<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Korea Market Regime on Korea Invest Insights</title><link>https://koreainvestinsights.com/series/korea-market-regime/</link><description>Recent content in Korea Market Regime on Korea Invest Insights</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>koreainvestinsights.com · @korea_invest_insights</copyright><lastBuildDate>Mon, 22 Jun 2026 21:16:06 +0900</lastBuildDate><atom:link href="https://koreainvestinsights.com/series/korea-market-regime/feed.xml" rel="self" type="application/rss+xml"/><item><title>What To Actually Watch In Korea's MSCI Review On June 24: Watch List Probability And The KB Financial Trade</title><link>https://koreainvestinsights.com/post/korea-msci-2026-developed-watch-list-probability-kb-financial-2026-06-22/</link><pubDate>Mon, 22 Jun 2026 08:00:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/korea-msci-2026-developed-watch-list-probability-kb-financial-2026-06-22/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;Connecting the dots
This piece follows &lt;a class="link" href="https://koreainvestinsights.com/post/kospi-benchmark-hard-to-beat-narrow-market-monte-carlo-2026-06-20/" &gt;How rare is an account that beats the pure KOSPI benchmark&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/kospi-disparity60-overheat-framework-partial-trim-2026-06-21/" &gt;The KOSPI 60-day disparity +28.6% overheat framework&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/semiconductor-2027-earnings-hyperscaler-payability-memory-nvidia-2026-06-21/" &gt;Who pays for the 2027 semiconductor consensus&lt;/a&gt;. Where the earlier pieces dealt with a &amp;ldquo;Samsung Electronics and SK Hynix hyper-concentration tape,&amp;rdquo; this one separates, by probability and by trade, &lt;strong&gt;whether the June 24 MSCI announcement is the event that changes that concentration&lt;/strong&gt;. It synthesizes two independent analyses (a probability frame and a Red Team check) without conflict.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Korea&amp;rsquo;s probability of an immediate Developed Markets promotion in 2026 is around 1%.&lt;/strong&gt; The substance of the June 24 event is not a &amp;ldquo;Developed promotion&amp;rdquo; but &lt;strong&gt;whether Korea is re-added to the Developed Markets Watch List (the pool under review for reclassification)&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Combining the two analyses, the probabilities are &lt;strong&gt;Watch List 30-35%, positive monitoring (base) about 53%, negative deferral 14%, immediate promotion 1%&lt;/strong&gt;. The key drag is that in the MSCI 2026 accessibility review Korea still received &amp;ldquo;needs improvement (-)&amp;rdquo; on five items: FX, registration, information flow, clearing and settlement, and transferability.&lt;/li&gt;
&lt;li&gt;The upside case rests on &lt;strong&gt;availability of investment instruments improving from &amp;ldquo;-&amp;rdquo; to &amp;ldquo;+&amp;quot;&lt;/strong&gt;, and on the government preparing a 24-hour FX market for July 2026 and an offshore won settlement network for 2027. That said, as of the assessment cut-off date (May 31), neither was yet in operation, so there is a shortage of &amp;ldquo;verified investor experience.&amp;rdquo;&lt;/li&gt;
&lt;li&gt;The most actionable conclusion is not a chase of large-cap semiconductors but a &lt;strong&gt;conditional re-rating trade in KB Financial&lt;/strong&gt;. At roughly 0.87x 2026F P/B on the current price, the market has over-loaded AI and KOSPI beta but under-priced the potential decline in the bank&amp;rsquo;s cost of capital.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Do not buy unconditionally ahead of the announcement.&lt;/strong&gt; The higher-expected-value approach is to enter on price terms after confirming the &amp;ldquo;under review / Watch List / market adoption&amp;rdquo; wording.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-what-to-actually-watch-on-june-24"&gt;1. What To Actually Watch On June 24
&lt;/h2&gt;&lt;p&gt;MSCI has announced that it will release the results of its 2026 annual market classification review &lt;strong&gt;shortly after 22:30 CEST on June 23, 2026&lt;/strong&gt;, which is &lt;strong&gt;shortly after 05:30 KST on June 24, 2026&lt;/strong&gt; in Korean time.&lt;/p&gt;
&lt;p&gt;The five key questions for Korea are these.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;Will Korea be promoted directly from Emerging to Developed? → Very unlikely, since it is procedurally hard to skip the Watch List and consultation stages.&lt;/li&gt;
&lt;li&gt;Will Korea be re-added to the &lt;strong&gt;Developed Markets Watch List (the pool under review for reclassification)&lt;/strong&gt;? → The crux of this event.&lt;/li&gt;
&lt;li&gt;Does MSCI view Korea&amp;rsquo;s FX reform as a &amp;ldquo;sufficiently implemented and verified change,&amp;rdquo; or as &amp;ldquo;planned but unverified&amp;rdquo;? → The decisive point.&lt;/li&gt;
&lt;li&gt;Will MSCI recognize the 24-hour FX market for July 2026 and the offshore won settlement network for 2027 as &lt;strong&gt;irreversible reforms&lt;/strong&gt;? → If it does, the Watch List probability rises sharply.&lt;/li&gt;
&lt;li&gt;If added, is the timeline a 2027 decision and 2028 inclusion, or a 2028 decision and 2029 inclusion? → The latter is more conservative and realistic.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;The MSCI classification framework uses three criteria: &lt;strong&gt;economic development, size and liquidity, and market accessibility&lt;/strong&gt;. Developed markets require &amp;ldquo;very high&amp;rdquo; on accessibility and &amp;ldquo;unrestricted&amp;rdquo; availability of investment instruments.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-official-evidence-snapshot"&gt;2. Official Evidence Snapshot
&lt;/h2&gt;&lt;h3 id="21-the-msci-2026-accessibility-reviews-verdict-on-korea"&gt;2.1 The MSCI 2026 Accessibility Review&amp;rsquo;s Verdict On Korea
&lt;/h3&gt;&lt;p&gt;The MSCI 2026 Global Market Accessibility Review was published on June 18, with an analysis cut-off date of &lt;strong&gt;May 31, 2026&lt;/strong&gt;. As a result, the 24-hour FX market scheduled to launch from July is not yet reflected in actual operating data.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Item&lt;/th&gt;
 &lt;th&gt;2026 Assessment&lt;/th&gt;
 &lt;th&gt;Read-through&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Capital flow restriction&lt;/td&gt;
 &lt;td&gt;++&lt;/td&gt;
 &lt;td&gt;Restrictions on capital inflows and outflows are not a major issue&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;FX market liberalization&lt;/td&gt;
 &lt;td&gt;-&lt;/td&gt;
 &lt;td&gt;The core unresolved item&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Investor registration &amp;amp; account setup&lt;/td&gt;
 &lt;td&gt;-&lt;/td&gt;
 &lt;td&gt;LEI/IRC coexist; omnibus-account friction&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Information flow&lt;/td&gt;
 &lt;td&gt;-&lt;/td&gt;
 &lt;td&gt;English disclosure and dividend information still insufficient&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Clearing &amp;amp; settlement&lt;/td&gt;
 &lt;td&gt;-&lt;/td&gt;
 &lt;td&gt;Settlement by investor ID; limited practical adoption&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Transferability&lt;/td&gt;
 &lt;td&gt;-&lt;/td&gt;
 &lt;td&gt;In-kind and off-exchange allowed but not yet established&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Short selling&lt;/td&gt;
 &lt;td&gt;+&lt;/td&gt;
 &lt;td&gt;Improved after the resumption, but regulatory complexity remains&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Availability of investment instruments&lt;/td&gt;
 &lt;td&gt;+&lt;/td&gt;
 &lt;td&gt;Improved from &amp;ldquo;-&amp;rdquo; to &amp;ldquo;+&amp;rdquo; in 2026&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Stability of institutional framework&lt;/td&gt;
 &lt;td&gt;+&lt;/td&gt;
 &lt;td&gt;Not yet at the developed-market &amp;ldquo;++&amp;rdquo; level&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The most important point is that while Korea improved from six &amp;ldquo;-&amp;rdquo; marks in 2025 to five &amp;ldquo;-&amp;rdquo; in 2026, &lt;strong&gt;the gap to the developed-market standard is still wide&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="22-the-bottlenecks-msci-flagged-directly"&gt;2.2 The Bottlenecks MSCI Flagged Directly
&lt;/h3&gt;&lt;p&gt;MSCI assessed that Korea &amp;ldquo;does not yet have a fully deliverable offshore currency market, and onshore FX constraints persist.&amp;rdquo; The 24-hour onshore trading for July 2026 and the offshore won settlement for 2027 were described not as completed mechanisms but as &lt;strong&gt;future initiatives&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;On registration and account setup, LEI has replaced the IRC, but the coexistence of both systems leaves friction; on information flow, English disclosure expanded in 2026 but further steps remain for 2027, and the overhaul of dividend procedures has only been partially adopted. On clearing and settlement, overdrafts and omnibus accounts have been allowed but actual adoption is limited, and short selling has been possible since the ban was lifted in early 2025, but MSCI stated it will continue to monitor operational friction and the compliance burden.&lt;/p&gt;
&lt;h3 id="23-the-procedurally-important-wording"&gt;2.3 The Procedurally Important Wording
&lt;/h3&gt;&lt;p&gt;When considering a promotion, MSCI specifies that the change in classification &amp;ldquo;should be viewed as irreversible.&amp;rdquo; In 2025, it set as conditions for a Korea consultation &lt;strong&gt;&amp;ldquo;resolution of all issues, full implementation of the reforms, and sufficient time for market participants to assess the effects.&amp;quot;&lt;/strong&gt; Applied strictly, this wording lowers the probability of an addition in 2026, because as of May 31 the core FX reform had not yet been implemented and the offshore won settlement network is targeted for 2027.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-probability-scenarios-synthesis-of-two-analyses"&gt;3. Probability Scenarios (Synthesis Of Two Analyses)
&lt;/h2&gt;&lt;p&gt;The first analysis put the Watch List probability at 35%; the Red Team check put it at 30%. The difference comes from an assumption about &amp;ldquo;how preemptively MSCI trusts the policy roadmap.&amp;rdquo; Because the MSCI accessibility assessment is based on &lt;strong&gt;actual investor experience (feedback from asset managers, custodians, and brokers)&lt;/strong&gt; rather than a roadmap alone, the more conservative 30% is the more defensible figure. The synthesis is as follows.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Scenario&lt;/th&gt;
 &lt;th&gt;First analysis&lt;/th&gt;
 &lt;th&gt;Red Team adjustment&lt;/th&gt;
 &lt;th&gt;Form of the 6/24 announcement wording&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;A. Added to the Watch List / pool under review for reclassification&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;35%&lt;/td&gt;
 &lt;td&gt;30%&lt;/td&gt;
 &lt;td&gt;&amp;ldquo;under review for potential reclassification to Developed Markets&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;B. Addition deferred + positive monitoring (base)&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;50%&lt;/td&gt;
 &lt;td&gt;53%&lt;/td&gt;
 &lt;td&gt;&amp;ldquo;continue to monitor implementation and market adoption&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;C. Addition deferred + negative wording&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;14%&lt;/td&gt;
 &lt;td&gt;14%&lt;/td&gt;
 &lt;td&gt;&amp;ldquo;fundamental accessibility issues remain unresolved&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;D. Immediate Developed promotion&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;1%&lt;/td&gt;
 &lt;td&gt;1%&lt;/td&gt;
 &lt;td&gt;An immediate reclassification announcement (procedurally near-excluded)&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;&lt;strong&gt;Headline synthesis: Watch List 30-35%, base case is positive monitoring (about 53%), immediate promotion 1%.&lt;/strong&gt;&lt;/p&gt;
&lt;h3 id="upside-case-factors-that-raise-the-watch-list-probability"&gt;Upside case (factors that raise the Watch List probability)
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;On size and liquidity, Korea is already a market that can be discussed as a developed candidate. The government, too, explains that the reason it remains in EM is market accessibility.&lt;/li&gt;
&lt;li&gt;In the 2026 review, &lt;strong&gt;availability of investment instruments improved from &amp;ldquo;-&amp;rdquo; to &amp;ldquo;+&amp;quot;&lt;/strong&gt;. The listing of Korea-index-linked derivatives on international exchanges has broadened the range of accessible products.&lt;/li&gt;
&lt;li&gt;The government roadmap is concrete. 24-hour dollar-won spot trading is scheduled to begin on July 6, 2026 (with pilot trading on June 29), and the offshore won settlement network is targeted for January 2027. This directly addresses the won accessibility issue MSCI has flagged the longest.&lt;/li&gt;
&lt;li&gt;The consensus is also more favorable than in the past. NH Investment &amp;amp; Securities suggested that, if added to the Watch List, inclusion would follow after about 24 months of observation in 2028, with potential medium-to-long-term passive inflows of about &lt;strong&gt;$29.2bn (about KRW 44 trillion)&lt;/strong&gt;. Note that this is a brokerage estimate, not an official MSCI figure.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="downside-case-why-the-probability-cannot-be-lifted-above-50"&gt;Downside case (why the probability cannot be lifted above 50%)
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;On the official accessibility table, Korea still has five &amp;ldquo;-&amp;rdquo; items. In particular, the &amp;ldquo;-&amp;rdquo; on FX market liberalization is critical. Developed-market FX hinges on onshore and offshore depth, real-time price transparency, hedging instruments, and best execution during large-scale rebalancing.&lt;/li&gt;
&lt;li&gt;The core reforms are &lt;strong&gt;pre-implementation or in their early stages&lt;/strong&gt; as of the announcement date. As of May 31, there is a shortage of actual investor-experience data.&lt;/li&gt;
&lt;li&gt;Applied strictly, MSCI&amp;rsquo;s 2025 consultation conditions (&amp;ldquo;issues resolved, full implementation, sufficient time to assess&amp;rdquo;) make an addition in 2026 a decision pulled forward.&lt;/li&gt;
&lt;li&gt;External surveys are not uniformly optimistic, either. In a report citing Bloomberg, a majority of 15 investors and strategists said more time is needed to prove the durability of the reforms, and expected Korea to remain in EM for the time being.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="4-a-read-through-guide-by-announcement-wording"&gt;4. A Read-Through Guide By Announcement Wording
&lt;/h2&gt;&lt;p&gt;When the announcement lands at 05:30 on June 24, check the following phrases first.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Strongly positive&lt;/strong&gt;: &amp;ldquo;under review for potential reclassification to Developed Markets,&amp;rdquo; &amp;ldquo;consultation on potential reclassification of Korea,&amp;rdquo; &amp;ldquo;Korea added to the Watch List.&amp;rdquo; In this case the posterior probability shifts quickly toward &amp;ldquo;actual inclusion in 2028-2029.&amp;rdquo;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Neutral / base&lt;/strong&gt;: &amp;ldquo;continue to monitor the implementation and market adoption,&amp;rdquo; &amp;ldquo;assess effectiveness over time.&amp;rdquo; An addition in 2026 fails, but the 2027 track stays alive.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Negative&lt;/strong&gt;: &amp;ldquo;fully operational offshore FX market is not available,&amp;rdquo; &amp;ldquo;operational friction persists,&amp;rdquo; &amp;ldquo;limited adoption.&amp;rdquo; Beyond non-addition, market expectations could break. Such phrasing already appears under the Korea item in the 2026 accessibility review.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="5-timeline-scenarios"&gt;5. Timeline Scenarios
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;If added to the Watch List in 2026&lt;/strong&gt;: The fastest path is 2026 addition → 2027 decision → 2028 inclusion. But given that the offshore won settlement network only launches in January 2027 and that MSCI emphasizes actual investor experience, &lt;strong&gt;2026 addition → 2028 decision → 2029 inclusion&lt;/strong&gt; is more realistic. NH Investment &amp;amp; Securities likewise views about 24 months of observation followed by a 2028 announcement and a 2029 actual transition as the realistic scenario.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;If not added in 2026&lt;/strong&gt;: The focus shifts to June 2027. If the 24-hour FX market operates stably, the offshore won settlement network launches, and English disclosure, omnibus accounts, transferability, and settlement practice are felt in practice, the probability of a Watch List addition in 2027 rises. This path naturally runs 2027 addition → 2028/2029 decision → 2029/2030 inclusion.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="6-market-impact-and-trades-illustrations-and-watch-points"&gt;6. Market Impact And Trades (Illustrations And Watch Points)
&lt;/h2&gt;&lt;p&gt;The names below are &lt;strong&gt;illustrations of the mechanism, not buy recommendations&lt;/strong&gt;. The key point is that although this looks like an &amp;ldquo;event to buy all of Korea,&amp;rdquo; the actual benefit concentrates in &lt;strong&gt;financials, brokers, and large liquid names that can monetize improved foreign access and a lower cost of capital&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="61-the-most-actionable-idea-kb-financials-conditional-re-rating"&gt;6.1 The Most Actionable Idea: KB Financial&amp;rsquo;s Conditional Re-rating
&lt;/h3&gt;&lt;p&gt;An improvement in MSCI accessibility translates into a &lt;strong&gt;lower cost of capital&lt;/strong&gt; for Korean financials. KB Financial has been estimated at 2026F BPS of KRW 180,958, ROE of 11.3%, and P/B of 0.81x; on the June 22 price of KRW 157,000, the simple-conversion P/B is about &lt;strong&gt;0.87x&lt;/strong&gt;. A bank is not directly MSCI infrastructure, but it monetizes the narrowing of the risk premium from improved won and foreign access in the simplest way.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Call&lt;/strong&gt;: Wait → conditional Buy.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Entry&lt;/strong&gt;: (strong case) If the announcement explicitly states &amp;ldquo;under review&amp;rdquo; or &amp;ldquo;Watch List&amp;rdquo; and the share price is at or below 0.90x 2026F BPS, i.e. about KRW 163,000 or lower, enter. (base case) If there is no Watch List but a &amp;ldquo;market adoption&amp;rdquo; phrase appears and a disappointment sell-off brings it to KRW 154,000 or lower (about 0.85x BPS), enter.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Catalyst&lt;/strong&gt;: The June 24 MSCI announcement, the June 29 24-hour FX pilot trading, the July 6 launch of 24-hour spot trading, and the 2Q26 results and shareholder-return update.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Invalidation&lt;/strong&gt;: CET1 falling below 13%, a retreat in the payout ratio, a spike in credit cost, or &amp;ldquo;fundamental accessibility issues remain&amp;rdquo; being emphasized in the announcement.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Asymmetry&lt;/strong&gt;: The downside is a Watch List miss and bank regulation; the upside is a re-rating from 0.87x to 1.0x P/B. On a simple calculation, that is about KRW 181,000.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="62-samsung-securities-strong-event-beta-but-large-regulatory-risk"&gt;6.2 Samsung Securities: Strong Event Beta, But Large Regulatory Risk
&lt;/h3&gt;&lt;p&gt;Samsung Securities at KRW 114,000 on June 22, on a P/E of 10.54x and P/B of 1.32x, is no longer a cheap price. 1Q26 net income was KRW 450.9 billion, up 81.5% year over year, beating consensus. If Korea is added to the Watch List, turnover and WM beta attach the fastest, but if leveraged-ETF and margin-loan regulation tightens, the Q turns down immediately.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Call&lt;/strong&gt;: Watchlist. If, after a Watch List addition, the same-day gain is within 5% and there is no regulatory headline, a short-term trade is permitted. On a miss, approach only at KRW 108,000 or lower (about 1.25x P/B).&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Key variable&lt;/strong&gt;: Korea&amp;rsquo;s supervisory authorities have signaled that approval of Samsung Electronics- and SK Hynix-linked leveraged ETFs was hasty, and are considering stabilization measures. This is event beta, not a fundamental compounder.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="63-samsung-electronics--sk-hynix-not-msci-beneficiaries-but-a-crowded-ai-trade"&gt;6.3 Samsung Electronics / SK Hynix: Not MSCI Beneficiaries But A Crowded AI Trade
&lt;/h3&gt;&lt;p&gt;For the large-cap semiconductors, the valuation is dominated by the &lt;strong&gt;AI memory cycle&lt;/strong&gt; rather than MSCI. SK Hynix is up more than 340% in 2026, on June 22 overtook Samsung Electronics to become Korea&amp;rsquo;s most valuable company, and the two names now make up more than 50% of the KOSPI. The reading that &amp;ldquo;MSCI expectations&amp;rdquo; are the cause of the semiconductor rally is weak. The actual driver is AI memory pricing and supply-demand.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Call&lt;/strong&gt;: Avoid as an MSCI trade; Wait only on the memory cycle. Buying on MSCI alone is a thesis mismatch. The alpha source for these names is the hyperscaler CAPEX and HBM bottleneck seen in &lt;a class="link" href="https://koreainvestinsights.com/post/semiconductor-2027-earnings-hyperscaler-payability-memory-nvidia-2026-06-21/" &gt;the 2027 semiconductor consensus piece&lt;/a&gt;, not MSCI.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="64-market-reaction-by-scenario"&gt;6.4 Market Reaction By Scenario
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Watch List addition&lt;/strong&gt;: Direct passive rebalancing does not occur immediately, but the expectation of a &amp;ldquo;narrowing Korea discount&amp;rdquo; gets priced in first. Financial holding companies, brokers, exchange infrastructure, and improving-shareholder-return names move first. That said, in the actual inclusion phase, the EM exit and a low DM weight could produce some passive outflows (a brokerage estimate of about $5.2bn, KRW 8 trillion).&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Positive monitoring (deferral)&lt;/strong&gt;: Some of the pre-priced expectation can unwind, but if the 2027 roadmap stays alive the drawdown is limited. The market again becomes sensitive to semiconductor earnings, the won-dollar rate, foreign flows, and the pace of dividend reform.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Negative deferral&lt;/strong&gt;: If MSCI explicitly states that the FX reform &amp;ldquo;does not replicate developed-market practice,&amp;rdquo; the probability of a 2027 retry also falls and short-term expectation money exits quickly.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="7-final-call"&gt;7. Final Call
&lt;/h2&gt;&lt;div class="thesis-callout"&gt;
 &lt;div class="thesis-callout__label"&gt;The core sentence&lt;/div&gt;
 &lt;div class="thesis-callout__body"&gt;
 The real question on June 24 is not "Has Korea become developed," but "Does MSCI trust Korea's institutional reform enough to put it in the pool under review for reclassification." Rather than buying ahead of the announcement, the higher-expected-value approach is to confirm the announcement wording and then enter a cost-of-capital beneficiary like KB Financial on price terms.
 &lt;/div&gt;
&lt;/div&gt;
&lt;p&gt;In the June 24, 2026 announcement, an &amp;ldquo;actual Developed promotion&amp;rdquo; for Korea is not an event to expect. The probability is around 1%. The key event is a Watch List re-addition, and combining the two analyses, that is a &lt;strong&gt;30-35% positive surprise&lt;/strong&gt;. The bulk of the official evidence still points to &amp;ldquo;continue to monitor&amp;rdquo; rather than an &amp;ldquo;early addition,&amp;rdquo; because as of May 31 five &amp;ldquo;-&amp;rdquo; marks remain and the core FX reform is only implemented after the announcement.&lt;/p&gt;
&lt;p&gt;The base case is therefore &amp;ldquo;addition deferred + positive wording for a 2027 re-assessment.&amp;rdquo; Even so, factoring in the improvement in availability of investment instruments, the 24-hour FX and offshore won settlement roadmap, and the WGBI inclusion experience, the probability of entering the Watch List is clearly higher than in the past. Execution is led not by chasing semiconductors but by KB Financial&amp;rsquo;s conditional re-rating, and all you need to confirm is whether the announcement puts Korea among the &amp;ldquo;markets under review.&amp;rdquo;&lt;/p&gt;
&lt;p&gt;&lt;small&gt;The figures in this piece cite public materials noted in the text — the MSCI 2026 accessibility review and market classification framework, the Ministry of Economy and Finance, Reuters, Yonhap, Mirae Asset Securities, Samsung Securities, NH Investment &amp;amp; Securities, KCMI, and others — while the probabilities, passive-inflow estimates, and per-name re-rating ranges are estimates. The names are illustrations of the analytical flow, not investment recommendations; the actual investment decision and responsibility rest with the investor.&lt;/small&gt;&lt;/p&gt;
&lt;h2 id="8-evidence-classification-appendix"&gt;8. Evidence Classification Appendix
&lt;/h2&gt;&lt;h3 id="fact"&gt;[Fact]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;The MSCI 2026 Annual Market Classification Review is scheduled to be announced shortly after 22:30 CEST on June 23, 2026 (05:30 KST on June 24).&lt;/li&gt;
&lt;li&gt;The data cut-off date for the MSCI 2026 accessibility review is May 31, 2026.&lt;/li&gt;
&lt;li&gt;Korea received an assessment that it does not yet have a fully deliverable offshore currency market and that onshore FX constraints remain.&lt;/li&gt;
&lt;li&gt;The availability of investment instruments item improved from &amp;ldquo;-&amp;rdquo; to &amp;ldquo;+&amp;rdquo; in 2026.&lt;/li&gt;
&lt;li&gt;24-hour dollar-won spot trading is scheduled to begin on July 6, 2026, with pilot trading scheduled for June 29.&lt;/li&gt;
&lt;li&gt;Samsung Securities&amp;rsquo; 1Q26 net income was KRW 450.9 billion, up 81.5% year over year.&lt;/li&gt;
&lt;li&gt;SK Hynix overtook Samsung Electronics on June 22 to become Korea&amp;rsquo;s most valuable company.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="inference"&gt;[Inference]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;The 30-35% range is the defensible band for the Watch List probability, because the core FX reform has not been verified with actual operating data as of the announcement date.&lt;/li&gt;
&lt;li&gt;KB Financial expresses the MSCI accessibility improvement more purely than the semiconductors. At about 0.87x 2026F P/B, if the cost-of-capital decline thesis holds, there is room to approach 1.0x.&lt;/li&gt;
&lt;li&gt;Samsung Securities benefits quickly but carries large regulatory beta, making it a lower-quality trade than KB Financial.&lt;/li&gt;
&lt;li&gt;The alpha source for Samsung Electronics and SK Hynix is the AI memory bottleneck, not MSCI.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="speculation"&gt;[Speculation]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;If the announcement is at the &amp;ldquo;positive monitoring&amp;rdquo; level, disappointment selling could emerge, but if the 2027 path stays alive the drawdown in financial holdings may be limited.&lt;/li&gt;
&lt;li&gt;On a Watch List addition, the short-term gain may be larger for the brokers, but on a 3-6 month holding-quality basis KB Financial is better.&lt;/li&gt;
&lt;li&gt;In the actual DM inclusion phase, the EM exit and a low DM weight could produce some passive outflows.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="blocked"&gt;[Blocked]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;The final 2026 market classification result has not yet been announced.&lt;/li&gt;
&lt;li&gt;MSCI&amp;rsquo;s official passive inflow/outflow magnitudes are unconfirmed (the KRW 44 trillion inflow and KRW 8 trillion outflow are brokerage-industry estimates).&lt;/li&gt;
&lt;li&gt;The real-time FY26 consensus multiples for Samsung Electronics and SK Hynix cannot be fixed from official disclosures alone.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="sources"&gt;Sources
&lt;/h2&gt;</description></item><item><title>Who Pays For The 2027 Semiconductor Consensus? Hyperscaler Payability For Samsung, Hynix, Micron, Nvidia</title><link>https://koreainvestinsights.com/post/semiconductor-2027-earnings-hyperscaler-payability-memory-nvidia-2026-06-21/</link><pubDate>Sun, 21 Jun 2026 20:00:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/semiconductor-2027-earnings-hyperscaler-payability-memory-nvidia-2026-06-21/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;Connecting the dots
This piece is the synthesis of &lt;a class="link" href="https://koreainvestinsights.com/post/ai-supercycle-extension-agent-demand-ipo-funding-memory-storage-2026-06-12/" &gt;Why The AI Supercycle Runs Longer: IPO Funding And Memory/Storage&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/ai-datacenter-capex-5p3t-korea-power-substrate-storage-bottleneck-2026-06-05/" &gt;The $5.3tn Era Of AI Data Center CapEx&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/" &gt;Goldman&amp;rsquo;s Token Demand vs JPMorgan&amp;rsquo;s Memory ASP&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/samsung-hynix-micron-forward-per-parity-memory-catch-up-2026-06-03/" &gt;Sam-Hy-Mu Parity: When Korean Memory Got Cheap Again&lt;/a&gt;. Where those earlier pieces looked at demand, pricing, and valuation separately, this one ties them together under a single question: &lt;strong&gt;&amp;ldquo;Can demand actually afford to pay for the 2027 earnings consensus?&amp;quot;&lt;/strong&gt;&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;The 2027 consensus does not hinge on &lt;strong&gt;consumer electronics demand&lt;/strong&gt;, but on &lt;strong&gt;whether hyperscalers, AI cloud, and sovereign AI keep absorbing the prices of HBM, GPUs, and server DRAM&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;The verdict splits. &lt;strong&gt;Hyperscalers can pay conditionally&lt;/strong&gt;, &lt;strong&gt;governments and sovereign AI have the political will to pay but unclear ROI&lt;/strong&gt;, and &lt;strong&gt;PC and smartphone OEMs are already in the unpayable zone&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;In numbers, the combined 2027E CAPEX of the big four (Microsoft, Google, Meta, Amazon) alone is roughly &lt;strong&gt;~$782.2bn&lt;/strong&gt;, against combined FCF of roughly &lt;strong&gt;~$119.9bn&lt;/strong&gt;. They can pay on an accounting basis, but the residual cash cushion is thin.&lt;/li&gt;
&lt;li&gt;Nvidia&amp;rsquo;s FY2028 revenue consensus of roughly ~$551.7bn equals about &lt;strong&gt;70.5%&lt;/strong&gt; of the big four&amp;rsquo;s CAPEX. For that number to hold, you need not just the four but the &lt;strong&gt;entire global AI CAPEX pool&lt;/strong&gt; including Oracle, OpenAI, sovereign AI, China, and enterprise.&lt;/li&gt;
&lt;li&gt;The 2027E earnings of the three memory makers are even more sensitive to &lt;strong&gt;sustained memory ASPs and the durability of the HBM supply shortage&lt;/strong&gt; than to hyperscaler CAPEX. That is why the most interesting asymmetry is &lt;strong&gt;Samsung Electronics (a conditional candidate where the peak discount may be excessive)&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-the-core-question-and-the-fiscal-year-adjustment"&gt;1. The Core Question And The Fiscal-Year Adjustment
&lt;/h2&gt;&lt;p&gt;The purpose of this analysis is singular: to judge whether the 2027 earnings outlook for Samsung Electronics, SK hynix, Micron, and Nvidia is mere sell-side optimism, or a level verifiable against the payability of end demand.&lt;/p&gt;
&lt;p&gt;Five core questions:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;What are the 2027E revenue and earnings outlooks, and what does the picture look like once we adjust for fiscal-year differences?&lt;/li&gt;
&lt;li&gt;Can you simply add up the outlooks for the three memory makers and Nvidia, or does double-counting creep in?&lt;/li&gt;
&lt;li&gt;Can hyperscalers&amp;rsquo; 2027E CAPEX, FCF, and external financing capacity cover this revenue?&lt;/li&gt;
&lt;li&gt;Are governments and sovereign AI large enough to independently sustain 2027 earnings?&lt;/li&gt;
&lt;li&gt;Can PC and smartphone OEMs pass rising memory prices through to end consumers?&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;&lt;strong&gt;A note on fiscal years:&lt;/strong&gt; Samsung Electronics and SK hynix close in December, so they line up with CY2027. Micron&amp;rsquo;s FY2027 ends in August 2027. Nvidia&amp;rsquo;s FY2027 ends in January 2027, so it largely captures 2026 results; to see &amp;ldquo;2027,&amp;rdquo; you have to look at &lt;strong&gt;FY2028 (ending January 2028)&lt;/strong&gt; alongside it.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-the-2027e-earnings-consensus-snapshot"&gt;2. The 2027E Earnings Consensus Snapshot
&lt;/h2&gt;&lt;p&gt;The figures below are not company guidance but &lt;strong&gt;external consensus and market data&lt;/strong&gt; (per MarketScreener, closing prices of June 19, 2026). KRW figures are noted at roughly USD 1 = KRW 1,454 for ease of comparison.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Company&lt;/th&gt;
 &lt;th&gt;Comparison period&lt;/th&gt;
 &lt;th&gt;2027E revenue&lt;/th&gt;
 &lt;th&gt;2027E operating profit / EBIT&lt;/th&gt;
 &lt;th&gt;2027E net income&lt;/th&gt;
 &lt;th&gt;Valuation&lt;/th&gt;
 &lt;th&gt;First-pass verdict&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Samsung Electronics&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;CY2027&lt;/td&gt;
 &lt;td&gt;KRW 856.8tn (~$589.0bn)&lt;/td&gt;
 &lt;td&gt;KRW 460.1tn&lt;/td&gt;
 &lt;td&gt;KRW 373.4tn&lt;/td&gt;
 &lt;td&gt;KRW 354,000, 2027E P/E 6.12x&lt;/td&gt;
 &lt;td&gt;Consensus is aggressive; the stock is discounted as a &amp;ldquo;temporary peak&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;SK hynix&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;CY2027&lt;/td&gt;
 &lt;td&gt;KRW 483.8tn (~$333.0bn)&lt;/td&gt;
 &lt;td&gt;KRW 377.1tn&lt;/td&gt;
 &lt;td&gt;KRW 297.1tn&lt;/td&gt;
 &lt;td&gt;KRW 2,764,000, 2027E P/E 6.71x&lt;/td&gt;
 &lt;td&gt;HBM leadership is already largely priced in. A 6-7x peak P/E signals disbelief in durability&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Micron&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;FY2027&lt;/td&gt;
 &lt;td&gt;~$190.98bn&lt;/td&gt;
 &lt;td&gt;~$157.20bn&lt;/td&gt;
 &lt;td&gt;~$122.92bn (EPS ~$111)&lt;/td&gt;
 &lt;td&gt;$1,133.99, FY2027E P/E ~10.2x&lt;/td&gt;
 &lt;td&gt;Maximum memory beta, but with 2028 slowdown coming into view, the lowest margin of safety for new buyers&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Nvidia&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;FY2028&lt;/td&gt;
 &lt;td&gt;~$551.69bn&lt;/td&gt;
 &lt;td&gt;~$362.0bn&lt;/td&gt;
 &lt;td&gt;~$304.84bn (EPS ~$12.85)&lt;/td&gt;
 &lt;td&gt;$210.69, FY2028E P/E ~16.4x&lt;/td&gt;
 &lt;td&gt;Not expensive if the numbers are right. The issue is the scale of customer CAPEX those numbers require&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;&lt;strong&gt;The dispersion in forecasts is wide.&lt;/strong&gt; For SK hynix in particular, KB&amp;rsquo;s estimated 2027E operating profit was about &lt;strong&gt;71% higher&lt;/strong&gt; than the FnGuide consensus at the time (KRW 209.3tn). When the market&amp;rsquo;s 2027 picture for the same company is this divided, it means the consensus is not a &amp;ldquo;settled future&amp;rdquo; but a &amp;ldquo;high-conviction assumption about price durability.&amp;rdquo;&lt;/p&gt;
&lt;p&gt;And the 2027E operating margins for the three memory makers include figures in the 70% range. That is not a typical memory cycle, but a number that is &lt;strong&gt;only possible when a structural shortage plus long-term contracts hold&lt;/strong&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-dont-add-them-up-the-double-counting-trap"&gt;3. &amp;ldquo;Don&amp;rsquo;t Add Them Up&amp;rdquo;: The Double-Counting Trap
&lt;/h2&gt;&lt;p&gt;Start with the most important interpretation. The HBM, DRAM, and NAND revenue of the three memory makers and Nvidia&amp;rsquo;s GPU revenue &lt;strong&gt;partially overlap from the standpoint of end-demand spending&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;Nvidia&amp;rsquo;s revenue embeds the cost of systems, boards, and networking, including HBM. The memory makers&amp;rsquo; HBM revenue enters as the upstream of that system cost, then gets reflected in the price of the AI servers hyperscalers buy. So if you &lt;strong&gt;simply add &amp;ldquo;Nvidia revenue + memory makers&amp;rsquo; revenue&amp;rdquo; as a final spend figure, you count the same dollars twice&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;From an investment standpoint, the revenue at each node can all be real. But the end buyer has to pay for &lt;strong&gt;all&lt;/strong&gt; of it within the same AI data center budget: GPUs, HBM, commodity DRAM, SSDs, networking, servers, power, cooling, and construction. So rather than summing revenue, it is more rigorous to &lt;strong&gt;compare against the &amp;ldquo;total AI data center CAPEX pool.&amp;quot;&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="4-payability-test-by-demand-source"&gt;4. Payability Test By Demand Source
&lt;/h2&gt;&lt;h3 id="41-hyperscalers-conditionally-payable"&gt;4.1 Hyperscalers: Conditionally Payable
&lt;/h3&gt;&lt;p&gt;Summing the big four&amp;rsquo;s 2027E CAPEX and FCF gives the following (on a consensus basis).&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Demand source&lt;/th&gt;
 &lt;th style="text-align: right"&gt;2027E CAPEX&lt;/th&gt;
 &lt;th style="text-align: right"&gt;2027E FCF&lt;/th&gt;
 &lt;th&gt;Interpretation&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Microsoft&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$170.4bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$59.3bn&lt;/td&gt;
 &lt;td&gt;Ample investment capacity, but FCF dilution&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Alphabet&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$231.1bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$25.2bn&lt;/td&gt;
 &lt;td&gt;CAPEX absorbs most of FCF&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Meta&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$159.5bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$11.4bn&lt;/td&gt;
 &lt;td&gt;A structure where residual FCF gets very thin&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Amazon&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$221.2bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$24.0bn&lt;/td&gt;
 &lt;td&gt;Simultaneous investment in AWS, logistics, and AI&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Total&lt;/strong&gt;&lt;/td&gt;
 &lt;td style="text-align: right"&gt;&lt;strong&gt;~$782.2bn&lt;/strong&gt;&lt;/td&gt;
 &lt;td style="text-align: right"&gt;&lt;strong&gt;~$119.9bn&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Payable on an accounting basis, but limited cushion&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The arithmetic is simple. &lt;strong&gt;CFO proxy = CAPEX + FCF = $782.2bn + $119.9bn = roughly ~$902.1bn.&lt;/strong&gt; In other words, the claim that &amp;ldquo;hyperscalers have no cash flow to pay 2027 AI costs&amp;rdquo; is wrong. The problem is the opposite. &lt;strong&gt;Too much cash flow is being reallocated into AI CAPEX.&lt;/strong&gt; In this structure, buybacks, dividends, M&amp;amp;A, existing infrastructure replacement, power contracts, and data center leases all compete for the same cash.&lt;/p&gt;
&lt;p&gt;And the four alone are not enough. &lt;strong&gt;Nvidia&amp;rsquo;s FY2028 revenue of ~$551.69bn equals about 70.5% of the four&amp;rsquo;s $782.2bn CAPEX&lt;/strong&gt; (551.69 / 782.2 = 70.5%). For Nvidia&amp;rsquo;s revenue to materialize as the consensus expects, the four must pay not only for GPUs but also for non-GPU infrastructure (power, cooling, servers, networking, construction) at the same time, so you need the &lt;strong&gt;entire global AI CAPEX pool including Oracle, CoreWeave, OpenAI/Stargate, xAI, Chinese cloud, Middle Eastern sovereign AI, and enterprise&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;Goldman Sachs estimates 2026-2031 AI CAPEX at roughly $7.6tn, and Morgan Stanley estimates 2027E global data center spending at roughly ~$812.7bn. The scale is within range to explain the consensus, but both also present a financing gap: &lt;strong&gt;hyperscaler cash flow alone is insufficient, requiring external financing such as corporate bonds, ABS, and private credit&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="42-economic-payability-different-from-cash"&gt;4.2 Economic Payability: Different From Cash
&lt;/h3&gt;&lt;p&gt;What you can pay for in cash and what you can afford economically are different. Goldman Sachs typically puts the useful life of AI chips at &lt;strong&gt;4-6 years&lt;/strong&gt;. If 2027 AI CAPEX is $0.9tn-$1.1tn, then depreciation alone implies an annual economic cost of &lt;strong&gt;roughly ~$150bn-$275bn per year&lt;/strong&gt; ($0.9tn ÷ 6 years = $150bn / $1.1tn ÷ 4 years = $275bn). On top of that come power, cooling, leasing, networking, operations, model-training costs, and financing costs.&lt;/p&gt;
&lt;p&gt;So hyperscalers&amp;rsquo; payability ultimately depends on how fast &lt;strong&gt;AI inference revenue, enterprise AI seat expansion, ad/search/commerce improvements, and cloud GPU rental income&lt;/strong&gt; absorb this depreciation. Through 2027 they can hold on with cash flow and the financing markets. But to keep investing at the same pace in 2028-2029, AI monetization has to be far clearer.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Verdict: Cash payability is conditionally TRUE; payability on an economic ROIC basis is MIXED.&lt;/strong&gt;&lt;/p&gt;
&lt;h3 id="43-government-and-sovereign-ai-supplementary-demand-not-a-standalone-payer"&gt;4.3 Government And Sovereign AI: Supplementary Demand, Not A Standalone Payer
&lt;/h3&gt;&lt;p&gt;The demand is real. The EU announced an AI investment mobilization totaling EUR 200bn through InvestAI, and Saudi Arabia&amp;rsquo;s HUMAIN envisions an AI factory of up to 500MW and hundreds of thousands of GPUs with Nvidia (with a Phase 1 mention of 18,000 GB300 units). OpenAI&amp;rsquo;s Stargate is also a $500bn, 10GW program over four years (private-led, not a pure government budget).&lt;/p&gt;
&lt;p&gt;[Inference] Government and sovereign AI have lower price elasticity than private cloud (driven by defense, data sovereignty, and industrial-policy goals). They therefore act as a &lt;strong&gt;demand buffer&lt;/strong&gt; for Nvidia, HBM, and power infrastructure. But in terms of scale and pace of execution, they are closer to &lt;strong&gt;subsidies, long-term power contracts, credit enhancement, and anchor customers&lt;/strong&gt; than substitutes for hyperscaler CAPEX. Announced amounts should not be read as revenue outright. China in particular faces limited direct benefit to Nvidia and Micron due to export controls and localization.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Verdict: The claim that government demand alone sustains the 2027 supercycle is FALSE. It matters as downside protection when combined with hyperscalers.&lt;/strong&gt;&lt;/p&gt;
&lt;h3 id="44-consumer-electronics-oems-close-to-unable-to-pay"&gt;4.4 Consumer Electronics OEMs: Close To Unable To Pay
&lt;/h3&gt;&lt;p&gt;[Fact] Gartner forecast 2026 PC shipments at &lt;strong&gt;-10.4%&lt;/strong&gt;, smartphones at &lt;strong&gt;-8.4%&lt;/strong&gt;, and DRAM+SSD prices at &lt;strong&gt;+130%&lt;/strong&gt;, which it expects to push PC prices +17% and smartphone prices +13%, with the memory share of a PC&amp;rsquo;s BOM rising from 16% in 2025 to 23% in 2026. IDC likewise sees smartphones and PCs responding with price hikes, spec cuts, and launch delays as AI data centers absorb wafers.&lt;/p&gt;
&lt;p&gt;[Inference] Consumer electronics OEMs do not &amp;ldquo;pay&amp;rdquo; memory prices so much as respond with &lt;strong&gt;lower shipment volumes, reduced content per device, higher selling prices, and a downward model mix shift&lt;/strong&gt;. Premium devices can absorb ASP increases, but the mid-to-low-end market has high price elasticity, so demand destruction appears. In other words, PCs and smartphones are not the party justifying the 2027 memory consensus. That consensus effectively depends on demand for &lt;strong&gt;AI servers, HBM, enterprise SSDs, and DDR5 RDIMM&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Verdict: Not the core payer, but the victim of rising prices. FALSE.&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="5-supply-bottleneck-2027-is-a-p-problem-more-than-a-q-problem"&gt;5. Supply Bottleneck: 2027 Is A P Problem More Than A Q Problem
&lt;/h2&gt;&lt;p&gt;[Fact] Gartner forecasts 2026 semiconductor revenue of $1.320tn and 2027 of $1.555tn, with memory revenue of ~$633.3bn in 2026 and ~$748.1bn in 2027. It sees 2026 DRAM prices +125% and NAND +234%, with meaningful price relief potentially limited until the end of 2027.&lt;/p&gt;
&lt;p&gt;[Fact] TrendForce sees the HBM share of wafer input potentially rising to 18% at end-2025, 22% at end-2026, and 30% at end-2027, but the HBM bit-supply share may stay at only around 13% even in 2027. This means &lt;strong&gt;HBM eats up a lot of DRAM wafers, yet total bit supply grows only modestly&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;So 2027 earnings are a question less of demand volume Q than of &lt;strong&gt;price P and product mix&lt;/strong&gt;, and of who captures the incremental margin generated under constrained supply. That is why, even if commodity demand from PCs and smartphones slows, the HBM and server DDR5 shortage will not ease immediately, since wafer allocation toward AI servers has already moved.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="6-korea-read-through-and-stock-verdicts-illustrative-watch-points"&gt;6. Korea Read-Through And Stock Verdicts (Illustrative Watch Points)
&lt;/h2&gt;&lt;p&gt;The names below are &lt;strong&gt;illustrative basket examples and watch points&lt;/strong&gt;. They are an organization of investment hypotheses, not buy recommendations.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Stock&lt;/th&gt;
 &lt;th&gt;Call&lt;/th&gt;
 &lt;th&gt;One-line thesis&lt;/th&gt;
 &lt;th&gt;Key risk&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Samsung Electronics&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Watchlist → conditional Buy&lt;/td&gt;
 &lt;td&gt;If HBM4/HBM4E catch-up and DS operating leverage are confirmed, an early-6x 2027E P/E may be an excessive &amp;ldquo;peak discount&amp;rdquo;&lt;/td&gt;
 &lt;td&gt;HBM4 supply delays, failure to secure key-customer allocation, sharp drop in DS operating margin&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;SK hynix&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Wait&lt;/td&gt;
 &lt;td&gt;HBM leadership is the most certain, but the stock is already largely priced in. Earnings durability is attractive, but the margin of safety for new entry is low&lt;/td&gt;
 &lt;td&gt;Share/margin decline as rivals improve HBM4 yields, order adjustments&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Nvidia&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Watchlist&lt;/td&gt;
 &lt;td&gt;An FY2028E P/E in the 16x range is low if the numbers are right, but the burden of validating customer CAPEX and AI ROI has grown&lt;/td&gt;
 &lt;td&gt;In-house ASIC penetration, China regulation, falling GPU rental income, power bottlenecks&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Micron&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Avoid for new buys&lt;/td&gt;
 &lt;td&gt;Memory beta is large, but an FY2027E P/E in the 10x range plus 2028 slowdown risk make the asymmetry weaker than Korean memory&lt;/td&gt;
 &lt;td&gt;Early DRAM/NAND peak-out, inferior HBM share, FY2028E EPS downgrades&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The best setup is a &lt;strong&gt;&amp;ldquo;bottleneck node with pricing power that is still discounted as a peak.&amp;quot;&lt;/strong&gt; By that criterion, the most interesting one right now is Samsung Electronics. The market is still discounting HBM4, packaging, and memory price durability as a &amp;ldquo;peak cycle,&amp;rdquo; so if HBM4 catch-up is confirmed, there is asymmetry for an early-6x P/E to re-rate. SK hynix is high quality but the entry price is the issue; Nvidia is a quality asset but customer ROI validation has to come first; and Micron has the weakest asymmetry for new buying.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="7-red-team-when-this-conclusion-is-wrong"&gt;7. Red Team: When This Conclusion Is Wrong
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Rate and credit risk&lt;/strong&gt;: If rates rise or credit spreads widen, hyperscaler CAPEX wobbles first, given its heavy reliance on external financing.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;ASIC insourcing&lt;/strong&gt;: Google&amp;rsquo;s TPU, Amazon&amp;rsquo;s Trainium, and Meta&amp;rsquo;s in-house chips reshape Nvidia&amp;rsquo;s ASP ceiling and the HBM demand structure beyond 2027.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Memory capacity additions&lt;/strong&gt;: If Samsung and Micron expand HBM capacity aggressively and SK hynix&amp;rsquo;s spread narrows, 2027E margins plunge.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;&amp;ldquo;Revenue holds, FCF and monetization disappoint&amp;rdquo;&lt;/strong&gt;: The most dangerous signal is not CAPEX cuts but CAPEX held steady with weak AI monetization. Even if 2027 revenue holds, order intensity could break sharply in 2028.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="8-monitoring-checklist"&gt;8. Monitoring Checklist
&lt;/h2&gt;&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Indicator&lt;/th&gt;
 &lt;th&gt;Why it matters&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Big four CAPEX revisions&lt;/td&gt;
 &lt;td&gt;The top-level demand proxy for Samsung, Hynix, Micron, and Nvidia earnings&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;FCF after CAPEX&lt;/td&gt;
 &lt;td&gt;The key line between &amp;ldquo;able to pay&amp;rdquo; and &amp;ldquo;overspending&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Nvidia data center backlog / lead times&lt;/td&gt;
 &lt;td&gt;Visibility into FY2028 $550bn+ revenue&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;HBM LTAs, prepayments, capacity reservations&lt;/td&gt;
 &lt;td&gt;The most direct evidence of memory ASP durability&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;DRAM/NAND contract price vs spot&lt;/td&gt;
 &lt;td&gt;Gauging overheating and the top of the commodity memory supercycle&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Cloud GPU rental income&lt;/td&gt;
 &lt;td&gt;A proxy for actual monetization of AI infrastructure&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Inference cost and AI service margins&lt;/td&gt;
 &lt;td&gt;Hyperscalers&amp;rsquo; economic payability&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;PC/smartphone unit elasticity&lt;/td&gt;
 &lt;td&gt;Whether consumer electronics demand destruction is occurring&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Data center power permitting and PPAs&lt;/td&gt;
 &lt;td&gt;Risk of supply delays and CAPEX deferral&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;hr&gt;
&lt;h2 id="9-final-conclusion"&gt;9. Final Conclusion
&lt;/h2&gt;&lt;div class="thesis-callout"&gt;
 &lt;div class="thesis-callout__label"&gt;The core sentence&lt;/div&gt;
 &lt;div class="thesis-callout__body"&gt;
 What pays for the 2027 semiconductor consensus is not the consumer, but hyperscaler CAPEX. So the core question is not "Is there AI demand?" but "Do AI revenue and GPU utilization rise fast enough to justify CAPEX even beyond 2027?"
 &lt;/div&gt;
&lt;/div&gt;
&lt;p&gt;The high-earnings outlook for the four companies in 2027 is not mathematically impossible. The CAPEX of hyperscalers and the broader AI infrastructure ecosystem has already entered the $0.8tn-$1.0tn range, and including external financing, it can pay for 2027 supply-chain revenue. But &lt;strong&gt;three conditions must hold simultaneously&lt;/strong&gt;.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;Hyperscaler CAPEX sustains at least $0.8tn-$1.1tn in 2027, and players outside the four (Oracle, OpenAI, sovereign AI, China, enterprise) fill in Nvidia revenue and the non-GPU costs together.&lt;/li&gt;
&lt;li&gt;Memory prices are locked in not as a &amp;ldquo;temporary shortage premium&amp;rdquo; but through long-term contracts, prepayments, and capacity reservations.&lt;/li&gt;
&lt;li&gt;AI infrastructure converts into actual revenue. To keep investing at the same pace in 2028-2029, inference revenue, AI SaaS, ad/search improvement, and GPU utilization must justify depreciation and power costs.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;In short, the 2027 earnings outlook holds as a &lt;strong&gt;&amp;ldquo;hyperscaler-led conditional-payability scenario,&amp;quot;&lt;/strong&gt; but &lt;strong&gt;it does not hold on government demand and consumer electronics OEM demand alone.&lt;/strong&gt; In particular, the earnings of the three memory makers are even more sensitive to sustained memory ASPs and the durability of the HBM shortage than to hyperscaler CAPEX. So what to watch right now is not an &amp;ldquo;AI semiconductor basket,&amp;rdquo; but the &lt;strong&gt;bottleneck node that has pricing power and is still discounted as a peak&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;&lt;small&gt;The figures in this piece cite the MarketScreener consensus, company official results, and Gartner, IDC, Goldman Sachs, Morgan Stanley, TrendForce, Reuters, Mirae Asset, and KB materials as noted in the text, and all are market expectations as of mid-2026, not actual results. Stock names are examples illustrating the flow of analysis, not investment recommendations; actual investment decisions and responsibility rest with the investor.&lt;/small&gt;&lt;/p&gt;
&lt;h2 id="sources"&gt;Sources
&lt;/h2&gt;</description></item><item><title>The SpaceX IPO Is an 'AI Compute' Event, Not a 'Space' Event: What the Use of Proceeds and the Chey-Musk Meeting Mean for Korean Memory</title><link>https://koreainvestinsights.com/post/spacex-ipo-ai-compute-use-of-proceeds-sk-musk-korea-memory-2026-06-18/</link><pubDate>Thu, 18 Jun 2026 17:30:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/spacex-ipo-ai-compute-use-of-proceeds-sk-musk-korea-memory-2026-06-18/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;Connecting the thread
This piece follows &lt;a class="link" href="https://koreainvestinsights.com/post/ai-supercycle-extension-agent-demand-ipo-funding-memory-storage-2026-06-12/" &gt;Why the AI Supercycle Keeps Getting Longer — IPO Funding and Memory/Storage&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/spacex-ipo-korea-market-liquidity-ai-space-readthrough-2026-06-05/" &gt;The SpaceX IPO and Korean Market Liquidity&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/ai-datacenter-capex-5p3t-korea-power-substrate-storage-bottleneck-2026-06-05/" &gt;The $5.3tn AI Data Center CapEx Era&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/" &gt;Goldman&amp;rsquo;s Token Demand vs JPMorgan&amp;rsquo;s Memory ASP&lt;/a&gt;. Where the earlier pieces framed &amp;ldquo;the IPO as the new relay for AI funding,&amp;rdquo; this one tests that hypothesis concretely against &lt;strong&gt;the actual SpaceX use-of-proceeds plan and today&amp;rsquo;s Chey-Musk meeting&lt;/strong&gt;.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;The hypothesis: &lt;strong&gt;a successful SpaceX IPO accelerates AI infrastructure investment further.&lt;/strong&gt; The reasoning is simple — the use-of-proceeds plan in the S-1 places &lt;strong&gt;AI compute infrastructure expansion&lt;/strong&gt;, not launch vehicles, front and center.&lt;/li&gt;
&lt;li&gt;SpaceX&amp;rsquo;s biggest capex and loss center is not the rocket but &lt;strong&gt;xAI&lt;/strong&gt;. xAI lost roughly $6bn in 2025 and burned about $2.5bn in Q1 2026 alone. In short, the ~$75bn raise is &lt;strong&gt;fuel poured into a cash-burning sink called AI compute&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;That capital flows into demand for GPUs, HBM, memory, and power. And today&amp;rsquo;s reported &lt;strong&gt;meeting between SK Chairman Chey Tae-won and Elon Musk (scheduled for late June)&lt;/strong&gt; is the first link tying that flow to Korean memory — custom HBM for Tesla, HBM4 base die, memory for xAI data centers, storage for satellites, and even ESS.&lt;/li&gt;
&lt;li&gt;The conclusion is &lt;strong&gt;not a &amp;ldquo;SpaceX space stock&amp;rdquo; but &amp;ldquo;AI compute financing + a Korean HBM/memory/power option.&amp;quot;&lt;/strong&gt; That said, the meeting is still a sole-source report, and custom HBM is at an exploratory stage — both points must be kept separate.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-break-the-conventional-wisdom-spacex-is-not-a-space-stock"&gt;1. Break the Conventional Wisdom: SpaceX Is Not a &amp;ldquo;Space Stock&amp;rdquo;
&lt;/h2&gt;&lt;p&gt;Viewing the SpaceX IPO solely as a &amp;ldquo;space-themed listing&amp;rdquo; misses the point. Where &lt;a class="link" href="https://koreainvestinsights.com/post/spacex-ipo-korea-market-liquidity-ai-space-readthrough-2026-06-05/" &gt;the earlier piece&lt;/a&gt; examined the liquidity angle, this one looks at &lt;strong&gt;where the money is actually spent&lt;/strong&gt;. The answer is in the S-1 itself.&lt;/p&gt;
&lt;p&gt;SpaceX filed a Form S-1 with the SEC on May 20, 2026, and an amendment on June 1. Selling roughly 555.6m shares at $135 per share, it aims to &lt;strong&gt;raise about $75bn&lt;/strong&gt; at a target valuation of roughly $1.75tn — the largest IPO in U.S. history. (&lt;a class="link" href="https://www.cnbc.com/2026/05/20/spacex-ipo-live-updates.html" title="SpaceX&amp;#39;s historic IPO plans: Billions in losses and Musk&amp;#39;s massive ownership — CNBC"
 target="_blank" rel="noopener"
 &gt;CNBC&lt;/a&gt;)&lt;/p&gt;
&lt;p&gt;The crux is not the size but the &lt;strong&gt;use&lt;/strong&gt;. According to multiple analyses, the S-1&amp;rsquo;s use-of-proceeds plan places &lt;strong&gt;&amp;ldquo;AI compute infrastructure expansion&amp;rdquo; ahead of&lt;/strong&gt; the launch business, with most of the capital allocated to AI compute infrastructure, launch capacity, and Starlink build-out. (&lt;a class="link" href="https://kraneshares.com/spacex-ipo-5-key-takeaways-from-the-s-1-filing-and-how-to-get-exposure-today-with-agix/" title="SpaceX IPO: 5 Key Takeaways From the S-1 Filing — KraneShares"
 target="_blank" rel="noopener"
 &gt;KraneShares&lt;/a&gt;)&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-dissecting-the-use-of-proceeds"&gt;2. Dissecting the Use of Proceeds
&lt;/h2&gt;&lt;h3 id="21-the-biggest-capex-and-loss-center-is-xai"&gt;2.1 The Biggest Capex and Loss Center Is xAI
&lt;/h3&gt;&lt;p&gt;SpaceX&amp;rsquo;s ownership structure makes the picture clear. &lt;strong&gt;SpaceX owns xAI, and xAI owns X (formerly Twitter)&lt;/strong&gt;, with the AI data center and chip-related businesses sitting beneath. And the largest operating loss and capex sit precisely in this AI segment. xAI &lt;strong&gt;lost roughly $6bn in 2025 and burned about $2.5bn in Q1 2026 alone&lt;/strong&gt;. (&lt;a class="link" href="https://www.satellitetoday.com/finance/2026/06/03/assessing-spacex-finances-addressable-market-and-the-ai-pitch-ahead-of-ipo/" title="Assessing SpaceX Finances, Addressable Market, and the AI Pitch Ahead of IPO — Via Satellite"
 target="_blank" rel="noopener"
 &gt;Via Satellite&lt;/a&gt;)&lt;/p&gt;
&lt;p&gt;Capex itself has gone vertical.&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"&gt;&lt;code class="language-text" data-lang="text"&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;SpaceX total capex trajectory
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;2023: ~$4.4bn
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;2025: ~$20.7bn
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;Q1 2026: ~$10.1bn (single quarter)
&lt;/span&gt;&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;&lt;/div&gt;&lt;p&gt;This is not the curve of a &amp;ldquo;rocket company.&amp;rdquo; It is the curve of a company building AI compute and a satellite network at the same time. (&lt;a class="link" href="https://www.morningstar.com/stocks/6-charts-spacexs-s-1-financials" title="6 Charts on SpaceX&amp;#39;s Pre-IPO Financials — Morningstar"
 target="_blank" rel="noopener"
 &gt;Morningstar&lt;/a&gt;)&lt;/p&gt;
&lt;h3 id="22-ai-compute-is-not-an-abstraction--it-is-physical-data-centers"&gt;2.2 &amp;ldquo;AI Compute&amp;rdquo; Is Not an Abstraction — It Is Physical Data Centers
&lt;/h3&gt;&lt;p&gt;SpaceX&amp;rsquo;s AI compute runs on two tracks.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Ground data centers&lt;/strong&gt;: GPU clusters for xAI&amp;rsquo;s training and inference. This is where enormous GPU, HBM, and power demand originates.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Orbital data centers (AI1)&lt;/strong&gt;: SpaceX COO Gwynne Shotwell said the company plans to launch the first AI1 unit in late 2027, and before that will put compute onboard Starlink broadband and mobile satellites. (&lt;a class="link" href="https://www.datacenterdynamics.com/en/news/spacex-ipo-musks-firm-set-to-launch-first-orbital-data-center-ai1-satellites-in-2027-will-put-compute-on-starlink-craft/" title="SpaceX set to launch first &amp;#39;orbital data center&amp;#39; AI1 satellites in 2027 — DCD"
 target="_blank" rel="noopener"
 &gt;DCD&lt;/a&gt;)&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;So the proceeds are &amp;ldquo;money going to space&amp;rdquo; and at the same time &lt;strong&gt;&amp;ldquo;money going to AI compute.&amp;quot;&lt;/strong&gt; The two are not separate — Starlink&amp;rsquo;s launch capacity itself becomes the delivery vehicle for orbital compute infrastructure.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-why-this-is-ai-infrastructure-acceleration"&gt;3. Why This Is &amp;ldquo;AI Infrastructure Acceleration&amp;rdquo;
&lt;/h2&gt;&lt;p&gt;This is where the hypothesis gets tested. The funding relay laid out in &lt;a class="link" href="https://koreainvestinsights.com/post/ai-supercycle-extension-agent-demand-ipo-funding-memory-storage-2026-06-12/" &gt;the AI supercycle extension piece&lt;/a&gt; — Big Tech cash → corporate bonds → &lt;strong&gt;public markets&lt;/strong&gt; — sees its third stage actually open with the SpaceX IPO.&lt;/p&gt;
&lt;p&gt;The core logic goes like this.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;xAI is &lt;strong&gt;burning cash fast&lt;/strong&gt; right now ($6bn loss in 2025). The fuel for a loss-making AI compute business is, ultimately, financing.&lt;/li&gt;
&lt;li&gt;While at the private stage that fuel came from inside Musk&amp;rsquo;s ecosystem and private placements, &lt;strong&gt;the SpaceX IPO opens a new funding source: the public market.&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;If part of the $75bn flows into AI compute, that translates directly into &lt;strong&gt;orders for GPUs, HBM, server memory, eSSD, and power equipment&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;In other words, the more successful the SpaceX IPO, the more entrenched a structure in which AI compute capex is hard to stop. &lt;strong&gt;This is the mechanism of &amp;ldquo;successful IPO → AI infrastructure acceleration.&amp;quot;&lt;/strong&gt;&lt;/p&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Let&amp;rsquo;s be honest about the distinction, though. The proceeds may create new capex, or they may merely backstop already-planned capex financially. The direction of &amp;ldquo;acceleration&amp;rdquo; is right, but its magnitude must be confirmed by actual execution after listing.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;hr&gt;
&lt;h2 id="4-todays-link-the-chey-musk-meeting"&gt;4. Today&amp;rsquo;s Link: The Chey-Musk Meeting
&lt;/h2&gt;&lt;p&gt;The point where the hypothesis connects to Korea was reported today. Digital Daily reported exclusively that &lt;strong&gt;SK Chairman Chey Tae-won and Elon Musk are set to meet in the U.S. in late June&lt;/strong&gt;. An SK Group official&amp;rsquo;s position is that this is &amp;ldquo;hard to confirm.&amp;rdquo; (&lt;a class="link" href="https://www.ddaily.co.kr/page/view/2026061815040502328" title="최태원·머스크 회동 예정: 전사적 반도체·AI 인프라 협력 — 디지털데일리"
 target="_blank" rel="noopener"
 &gt;Digital Daily&lt;/a&gt;)&lt;/p&gt;
&lt;p&gt;The reported agenda is precisely the Korean end of the mechanism above.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Cooperation Area&lt;/th&gt;
 &lt;th&gt;Details&lt;/th&gt;
 &lt;th&gt;Korean Memory Link&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Semiconductors&lt;/td&gt;
 &lt;td&gt;Exploratory talks on custom HBM for Tesla&amp;rsquo;s in-house AI5/AI6 chips, HBM4 base die cooperation&lt;/td&gt;
 &lt;td&gt;SK hynix / Samsung Electronics HBM&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;AI Infrastructure&lt;/td&gt;
 &lt;td&gt;Memory chips needed for xAI data center expansion&lt;/td&gt;
 &lt;td&gt;Server DRAM / HBM / eSSD&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Space &amp;amp; Comms&lt;/td&gt;
 &lt;td&gt;Memory and storage for SpaceX spacecraft and low-orbit satellites&lt;/td&gt;
 &lt;td&gt;High-reliability memory / storage&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Energy&lt;/td&gt;
 &lt;td&gt;ESS (energy storage systems) for data centers and the power grid&lt;/td&gt;
 &lt;td&gt;Power / battery value chain&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;One official assessed that &amp;ldquo;SK Group is broadening the scope of cooperation to seize leadership in next-generation AI infrastructure.&amp;rdquo; (&lt;a class="link" href="https://www.ddaily.co.kr/page/view/2026061815040502328" title="최태원·머스크 회동 예정: 전사적 반도체·AI 인프라 협력 — 디지털데일리"
 target="_blank" rel="noopener"
 &gt;Digital Daily&lt;/a&gt;)&lt;/p&gt;
&lt;p&gt;The crux is this. The &lt;strong&gt;&amp;ldquo;public-market capital flowing to AI compute&amp;rdquo;&lt;/strong&gt; created by the SpaceX IPO meets the &lt;strong&gt;&amp;ldquo;HBM and memory that AI compute requires&amp;rdquo;&lt;/strong&gt; that SK holds — at the same table. The starting point of the capital (the IPO) and its destination (Korean memory) enter the same picture.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="5-korea-read-through-illustrative-watch-points"&gt;5. Korea Read-Through (Illustrative Watch Points)
&lt;/h2&gt;&lt;p&gt;What follows is not a stock recommendation but an &lt;strong&gt;illustration&lt;/strong&gt; of where to look first when the mechanism above actually starts working.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;SK hynix&lt;/strong&gt; — the most direct name in this link. The HBM leader and a core supply candidate for the meeting&amp;rsquo;s agenda (custom HBM for Tesla, memory for xAI data centers). That said, custom HBM is still at the &amp;ldquo;exploratory&amp;rdquo; stage, not an order.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Samsung Electronics&lt;/strong&gt; — a composite option of HBM4 base die, custom-chip foundry, and data center eSSD. If the chips in Musk&amp;rsquo;s ecosystem (AI5/AI6) use an external foundry, the option value rises.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Power / ESS / cooling&lt;/strong&gt; — AI compute is, in the end, a power problem. ESS being on the meeting agenda overlaps exactly with the &amp;ldquo;the bottleneck after GPUs is power&amp;rdquo; from &lt;a class="link" href="https://koreainvestinsights.com/post/ai-datacenter-capex-5p3t-korea-power-substrate-storage-bottleneck-2026-06-05/" &gt;the data center CapEx piece&lt;/a&gt;.&lt;/li&gt;
&lt;/ul&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Common condition: for these names to become candidates, &amp;ldquo;a hot SpaceX IPO&amp;rdquo; alone is not enough. &lt;strong&gt;Actual use-of-proceeds execution, official confirmation of the meeting, and conversion of custom HBM into orders&lt;/strong&gt; must all arrive together.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;hr&gt;
&lt;h2 id="6-red-team-when-this-hypothesis-weakens"&gt;6. Red Team: When This Hypothesis Weakens
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Capital reallocation vs new capex&lt;/strong&gt;: if the proceeds merely backfill already-planned investment financially, the incremental &amp;ldquo;acceleration&amp;rdquo; is small.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;xAI loss risk&lt;/strong&gt;: AI compute is a massive cash-burning sink ($6bn loss in 2025). Even if listing proceeds cover it, late monetization raises the burden of follow-on financing.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Unconfirmed meeting&lt;/strong&gt;: the Chey-Musk meeting is a sole-source report, and SK&amp;rsquo;s position is &amp;ldquo;hard to confirm.&amp;rdquo; The custom HBM on the agenda is also exploratory. It is too early to price it as fact.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Orbital data centers are long-dated&lt;/strong&gt;: AI1 is slated to launch in late 2027. The Korean memory upside from &amp;ldquo;satellite compute&amp;rdquo; is an option, not near-term earnings.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;IPO funding-drain backlash&lt;/strong&gt;: as seen in &lt;a class="link" href="https://koreainvestinsights.com/post/spacex-ipo-korea-market-liquidity-ai-space-readthrough-2026-06-05/" &gt;the earlier piece&lt;/a&gt;, if a large IPO becomes a funding source by selling existing AI positions, Samsung Electronics and SK hynix could become sell targets in the short term. The material for acceleration is also the material for short-term volatility.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="7-monitoring-checklist"&gt;7. Monitoring Checklist
&lt;/h2&gt;&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Checkpoint&lt;/th&gt;
 &lt;th&gt;Why It Matters&lt;/th&gt;
 &lt;th&gt;Read&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;SpaceX offer price / demand&lt;/td&gt;
 &lt;td&gt;Sets the scale of financing and the strength of risk-on&lt;/td&gt;
 &lt;td&gt;A hot deal strengthens the AI compute capex acceleration logic&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Actual use-of-proceeds execution&lt;/td&gt;
 &lt;td&gt;Whether &amp;ldquo;AI compute first&amp;rdquo; shows up as orders, not just words&lt;/td&gt;
 &lt;td&gt;GPU / HBM / power order data&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;xAI capex / data center build-out&lt;/td&gt;
 &lt;td&gt;A leading signal of AI compute demand&lt;/td&gt;
 &lt;td&gt;Accelerating build-out widens memory demand&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Chey-Musk meeting made official&lt;/td&gt;
 &lt;td&gt;Sole-source report → investable event&lt;/td&gt;
 &lt;td&gt;Official announcement / MOU&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Custom HBM / base die orders&lt;/td&gt;
 &lt;td&gt;Exploratory → earnings conversion&lt;/td&gt;
 &lt;td&gt;SK hynix / Samsung Electronics disclosures&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Starlink / AI1 satellite compute&lt;/td&gt;
 &lt;td&gt;Materialization of the orbital data center&lt;/td&gt;
 &lt;td&gt;2027 launch schedule&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;hr&gt;
&lt;h2 id="8-conclusion"&gt;8. Conclusion
&lt;/h2&gt;&lt;div class="thesis-callout"&gt;
 &lt;div class="thesis-callout__label"&gt;Bottom line&lt;/div&gt;
 &lt;div class="thesis-callout__body"&gt;
 The SpaceX IPO is not a space event but an AI compute financing event. That capital flows into GPUs, HBM, memory, and power, and today's Chey-Musk meeting is the first link tying that flow to Korean memory.
 &lt;/div&gt;
&lt;/div&gt;
&lt;p&gt;The hypothesis &amp;ldquo;successful IPO → AI infrastructure acceleration&amp;rdquo; is &lt;strong&gt;clearly supported in direction&lt;/strong&gt; by the S-1&amp;rsquo;s use-of-proceeds plan (AI compute infrastructure front and center) and xAI&amp;rsquo;s capex/loss structure. But its magnitude and timing, and the transfer to Korean memory, must pass three gates: (1) actual use-of-proceeds execution, (2) official confirmation of the meeting, and (3) conversion of custom HBM into orders.&lt;/p&gt;
&lt;p&gt;The practical stance is one sentence. &lt;strong&gt;See SpaceX not as a space stock but as &amp;ldquo;AI compute financing + a Korean HBM/memory/power option,&amp;rdquo; while keeping the sole-source report and the exploratory stage separate from earnings and confirming step by step.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;&lt;small&gt;The facts in this piece cite the reporting, disclosures, and analyses noted in the body. The SpaceX IPO is in progress; the Chey-Musk meeting is at the sole-source report stage (SK says &amp;ldquo;hard to confirm&amp;rdquo;) and custom HBM is at the exploratory stage. The stock names are not investment recommendations but examples illustrating the analytical flow, and actual investment decisions and responsibility rest with the investor.&lt;/small&gt;&lt;/p&gt;
&lt;h2 id="sources"&gt;Sources
&lt;/h2&gt;</description></item><item><title>The U.S.-Iran Ceasefire MOU and Iran's Reconstruction: For Korean Equities This Is Not a 'Reconstruction Play' but a Power-Equipment, Plant and Shipbuilding Re-Rating</title><link>https://koreainvestinsights.com/post/us-iran-mou-iran-reconstruction-korea-power-plant-cyclical-2026-06-17/</link><pubDate>Wed, 17 Jun 2026 16:00:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/us-iran-mou-iran-reconstruction-korea-power-plant-cyclical-2026-06-17/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;Connecting the dots
This piece is the macro follow-up to &lt;a class="link" href="https://koreainvestinsights.com/post/ai-supercycle-midgame-rate-risk-yellow-not-red-2026-06-06/" &gt;The AI Supercycle Past Halftime: Rate and Funding Risk&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/foreign-return-after-24-day-kospi-selling-memory-rebalance-2026-06-12/" &gt;Have Foreigners Come Back? The Reversal After 24 Sessions of Selling&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/korea-outperformance-2026-structural-rerating-2026-04-24/" &gt;Korea&amp;rsquo;s 2026 Re-Rating&lt;/a&gt;. Where the earlier notes looked at rates, foreign flows and re-rating separately, this one translates &lt;strong&gt;a new variable, the shrinking of Middle East risk,&lt;/strong&gt; into Korea&amp;rsquo;s cyclical value chain.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;On the disclosed 14-article draft, this is &lt;strong&gt;a strategic win on points for Iran&lt;/strong&gt;. Washington secured the reopening of Hormuz and an end to the war, but Iran&amp;rsquo;s regime, its nuclear leverage and its path back to economic openness all survived.&lt;/li&gt;
&lt;li&gt;The real substance of &amp;ldquo;Iran reconstruction&amp;rdquo; is not reparations or aid but &lt;strong&gt;a private-investment (project-finance) pipeline that follows sanctions relief&lt;/strong&gt;. The most immediate cash flow is not the $300 billion fund but &lt;strong&gt;the resumption of Iranian crude sales&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;From a Korean investor&amp;rsquo;s standpoint this is a risk-on re-rating event that runs through &lt;strong&gt;power equipment, plant components, large EPC, shipbuilding, airlines and brokerages&lt;/strong&gt;. But chasing small-cap reconstruction theme stocks is dangerous.&lt;/li&gt;
&lt;li&gt;The conclusion is single. This event should be read not as an &amp;ldquo;Iran reconstruction play&amp;rdquo; but as &lt;strong&gt;a smaller Middle East risk premium + a sanctions-relief option + a re-pricing of Korea&amp;rsquo;s plant, power and shipbuilding value chain&lt;/strong&gt;. The next moves to watch are OFAC waivers, the final MOU text, and confirmation of Korean firms&amp;rsquo; LOIs and order disclosures.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-scope-and-core-questions"&gt;1. Scope and Core Questions
&lt;/h2&gt;&lt;p&gt;This analysis turns on five questions.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;The disclosed 14-article U.S.-Iran MOU draft — &lt;strong&gt;whom does the document favor?&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;Are the military winner of the war and the winner of the negotiated text the same?&lt;/li&gt;
&lt;li&gt;Is &amp;ldquo;$300 billion of Iran reconstruction&amp;rdquo; real money, or a political number?&lt;/li&gt;
&lt;li&gt;Through what transmission channel does it work on the Korean stock market?&lt;/li&gt;
&lt;li&gt;Should an investor today chase construction stocks, or position for a broader cyclical re-rating?&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;The premise must be stated clearly. What is public now is not a final treaty but &lt;strong&gt;a draft of MOU/framework character&lt;/strong&gt;. Domestic reports citing Bloomberg say at least three drafts were discussed between Washington and Tehran, with the common denominators being the reopening of the Strait of Hormuz, eased Iran sanctions, and a long-term negotiation over Iran&amp;rsquo;s nuclear program. (&lt;a class="link" href="https://www.asiae.co.kr/en/article/2026061504291984952" title="Bloomberg: At Least Three U.S.-Iran Draft Agreements — 아시아경제"
 target="_blank" rel="noopener"
 &gt;Asia Business Daily&lt;/a&gt;)&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-the-core-structure-of-the-14-article-mou"&gt;2. The Core Structure of the 14-Article MOU
&lt;/h2&gt;&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Article cluster&lt;/th&gt;
 &lt;th&gt;Core content&lt;/th&gt;
 &lt;th&gt;Investor reading&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Articles 1-3&lt;/td&gt;
 &lt;td&gt;End of war on all fronts; sovereignty, territorial integrity, non-interference; negotiate a final accord within 60 days&lt;/td&gt;
 &lt;td&gt;Premised on the survival of Iran&amp;rsquo;s regime. A signal that regime change is off the table&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Articles 4-5&lt;/td&gt;
 &lt;td&gt;U.S. lifts the naval blockade; Iran clears mines and normalizes merchant shipping&lt;/td&gt;
 &lt;td&gt;The global market secures Hormuz stability. Iran recovers its export route&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Article 6&lt;/td&gt;
 &lt;td&gt;A plan to raise at least $300 billion for reconstruction and economic development&lt;/td&gt;
 &lt;td&gt;The headline is large, but the substance is a private-investment / project-finance structure&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Article 7&lt;/td&gt;
 &lt;td&gt;Pledge to lift UN, IAEA and U.S. primary and secondary sanctions&lt;/td&gt;
 &lt;td&gt;The key article for reopening Iran&amp;rsquo;s economy&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Articles 8-9&lt;/td&gt;
 &lt;td&gt;Ban on building nuclear weapons reaffirmed; enriched uranium and nuclear needs left to follow-on talks&lt;/td&gt;
 &lt;td&gt;Not an immediate dismantlement. Iran retains negotiating leverage&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Articles 10-11&lt;/td&gt;
 &lt;td&gt;Crude and petrochemical exports; exemptions for finance, insurance and transport; release of frozen assets&lt;/td&gt;
 &lt;td&gt;The most immediate economic benefit&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Articles 12-14&lt;/td&gt;
 &lt;td&gt;Set up an implementing body; final negotiation after conditions are met; approval by a UNSC resolution&lt;/td&gt;
 &lt;td&gt;If it becomes a final accord, legal binding force can strengthen&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The core is &lt;strong&gt;asymmetry&lt;/strong&gt;. Iran offers &amp;ldquo;reaffirming the ban on building nuclear weapons, normalizing merchant shipping, joining the talks.&amp;rdquo; The U.S. offers &amp;ldquo;lifting the naval blockade, allowing crude exports, a sanctions-relief path, releasing frozen assets, a reconstruction fund.&amp;rdquo; This is not an instrument of surrender but &lt;strong&gt;a ceasefire text premised on the survival of Iran&amp;rsquo;s regime and the reopening of its economy&lt;/strong&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-who-won-the-war"&gt;3. Who Won the War
&lt;/h2&gt;&lt;h3 id="31-iran--a-strategic-win-on-points"&gt;3.1 Iran — A Strategic Win on Points
&lt;/h3&gt;&lt;p&gt;On the public draft, the biggest winner is Iran. What Iran obtained: regime survival, language on sovereignty and territorial integrity, the lifting of the naval blockade, exemptions for crude and petrochemical exports, the possibility of releasing frozen assets, a sanctions-relief roadmap, a $300 billion fund option, and follow-on talks rather than the immediate dismantlement of its nuclear program. Militarily it took damage, but politically it is &lt;strong&gt;the side that got hit and still survived&lt;/strong&gt;. In an authoritarian system and in the security politics of the Middle East, that is a very large victory.&lt;/p&gt;
&lt;h3 id="32-the-united-states--a-successful-loss-limiter-not-a-complete-win"&gt;3.2 The United States — A Successful Loss-Limiter, Not a Complete Win
&lt;/h3&gt;&lt;p&gt;The U.S. obtained an end to the war, the reopening of Hormuz, oil-price stability, and a restart of nuclear talks. Axios reported that the U.S.-Iran framework includes a 60-day ceasefire extension, the reopening of Hormuz, and a restart of nuclear talks. (&lt;a class="link" href="https://www.axios.com/2026/06/14/us-iran-ceasefire-extended-hormuz-reopen-trump" title="US, Iran reach deal to extend ceasefire, open strait — Axios"
 target="_blank" rel="noopener"
 &gt;Axios&lt;/a&gt;) That said, what would have been America&amp;rsquo;s top goals — regime change, immediate removal of enriched uranium, limits on ballistic missiles, dismantlement of the proxy structure — are not in the public text. Reuters likewise reported that the current MOU is not a final accord, and that Trump said in effect he could resume military action if he did not like the terms. (&lt;a class="link" href="https://www.reuters.com/world/middle-east/g7-leaders-demand-ceasefire-lebanon-welcome-iran-deal-2026-06-17/" title="Trump on Iran deal: &amp;#39;If I don&amp;#39;t like it, we&amp;#39;ll go back to shooting&amp;#39; — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;) So the U.S. is closer to &lt;strong&gt;a loss-limiter who built an exit off a position of military superiority&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="33-israel--a-relative-loser"&gt;3.3 Israel — A Relative Loser
&lt;/h3&gt;&lt;p&gt;The text is unfavorable to Israel. (1) The central parties to the ceasefire text are not Israel but the U.S. and Iran; (2) even the Lebanon front was folded into the ceasefire frame; and (3) ballistic missiles, Hezbollah and the proxy network were not nailed down as clauses for immediate, enforced dismantlement. The Guardian reported that in the G7 discussion it was raised that Iran&amp;rsquo;s ballistic missiles and regional threats should be handled in follow-on negotiations. (&lt;a class="link" href="https://www.theguardian.com/world/2026/jun/17/trump-g7-leaders-iran-missile-programme-ukraine" title="Trump backs G7 leaders&amp;#39; call for wider talks on Iranian missile programme — Guardian"
 target="_blank" rel="noopener"
 &gt;Guardian&lt;/a&gt;) In short, &lt;strong&gt;the side that fought the war was Israel, but the pricing power over the ceasefire text rests with the U.S. and Iran&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="34-global-markets-and-the-gulf--pragmatic-winners"&gt;3.4 Global Markets and the Gulf — Pragmatic Winners
&lt;/h3&gt;&lt;p&gt;From a market standpoint, the most important thing is the reduction of Hormuz risk. Reuters reported that after expectations of a deal, Gulf bourses mostly rose and Brent fell 5.1% to $82.86 a barrel. (&lt;a class="link" href="https://www.reuters.com/world/middle-east/most-gulf-markets-gain-iran-deal-2026-06-15/" title="Most Gulf markets gain on US, Iran deal — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;) For Korea, an energy importer, this is broadly positive — a lower oil-price risk premium, eased inflation worries, a return of risk appetite, a restart of foreign risk-on, and cyclical rotation.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="4-the-substance-of-irans-reconstruction"&gt;4. The Substance of Iran&amp;rsquo;s Reconstruction
&lt;/h2&gt;&lt;h3 id="41-reconstruction-is-not-one-thing-but-three"&gt;4.1 &amp;ldquo;Reconstruction&amp;rdquo; Is Not One Thing but Three
&lt;/h3&gt;&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Category&lt;/th&gt;
 &lt;th&gt;Substance&lt;/th&gt;
 &lt;th&gt;Certainty&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Physical repair&lt;/td&gt;
 &lt;td&gt;War-damage repair of refineries, ports, airports, steel, the power grid, etc.&lt;/td&gt;
 &lt;td&gt;Medium&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Economic reopening&lt;/td&gt;
 &lt;td&gt;After eased sanctions, restart of investment in energy, logistics, manufacturing, transport&lt;/td&gt;
 &lt;td&gt;High, but conditional&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Nuclear-deal incentive&lt;/td&gt;
 &lt;td&gt;The reward package when Iran meets its nuclear, inspection and Hormuz conditions&lt;/td&gt;
 &lt;td&gt;High, but dependent on a final accord&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The important point is that the $300 billion is &lt;strong&gt;not a U.S. cash reparation or ODA&lt;/strong&gt;. Reuters reported that the fund is wholly private-money-based and targets energy, logistics, manufacturing and transport, and that it is not U.S. taxpayer money or a government subsidy. (&lt;a class="link" href="https://www.reuters.com/business/finance/iran-deal-includes-300-billion-fund-more-than-half-which-already-committed-2026-06-16/" title="Iran deal includes $300 billion fund, more than half already committed — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;)&lt;/p&gt;
&lt;h3 id="42-the-nature-of-the-300-billion-fund"&gt;4.2 The Nature of the $300 Billion Fund
&lt;/h3&gt;&lt;p&gt;&lt;strong&gt;Verdict: there is substance, but right now it is closer to a conditional investment option than a hard commitment.&lt;/strong&gt; Per Reuters reporting, the fund is wholly private money, open to firms from the U.S., the Gulf, Asia, South America and Africa; an anonymous source says more than half is committed; the targets are energy, logistics, manufacturing and transport; disbursement follows a final accord; and the structure depends on Iran&amp;rsquo;s nuclear, inspection and regional-security compliance. In VC terms, what exists now is &lt;strong&gt;not a signed capital call but soft-circled capital + a political option&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="43-the-money-with-the-most-substance-is-crude"&gt;4.3 The Money With the Most Substance Is Crude
&lt;/h3&gt;&lt;p&gt;Right after the MOU is signed, the thing that can be turned into cash fastest is not the reconstruction fund but &lt;strong&gt;the resumption of Iranian crude sales&lt;/strong&gt;. Reuters reported that the U.S. agreed, upon conclusion of the MOU, to allow waivers for Iran&amp;rsquo;s crude and fuel sales and the related banking, shipping and insurance services, and that Iran could access more than 100 million barrels of stored crude. (&lt;a class="link" href="https://www.reuters.com/business/energy/us-iran-deal-allows-tehran-immediately-sell-oil-wsj-reports-2026-06-16/" title="Tehran can immediately sell oil upon signing US-Iran deal — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;)&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"&gt;&lt;code class="language-text" data-lang="text"&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;Short-term crude monetizable gross value
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;= 100 million barrels of stored crude x Brent ~$79/barrel
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;= ~$7.9 billion
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;Check: 100,000,000 barrels x $80/bbl = $8.0bn ~ $7.9bn
&lt;/span&gt;&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;&lt;/div&gt;&lt;p&gt;This sum is smaller than $300 billion, but its &lt;strong&gt;immediacy, liquidity and feasibility of execution are far higher.&lt;/strong&gt;&lt;/p&gt;
&lt;h3 id="44-crunching-the-numbers"&gt;4.4 Crunching the Numbers
&lt;/h3&gt;&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Item&lt;/th&gt;
 &lt;th&gt;Formula&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Result&lt;/th&gt;
 &lt;th&gt;Reading&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Reconstruction fund / GDP&lt;/td&gt;
 &lt;td&gt;300bn / 300.29bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;99.9%&lt;/td&gt;
 &lt;td&gt;Close to a full year of Iran&amp;rsquo;s 2026 nominal GDP&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Half-committed assumption&lt;/td&gt;
 &lt;td&gt;150bn / 300.29bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;50.0%&lt;/td&gt;
 &lt;td&gt;Anonymous-source basis; legal binding force unconfirmed&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Reconstruction fund / energy-repair cost&lt;/td&gt;
 &lt;td&gt;300bn / 58bn&lt;/td&gt;
 &lt;td style="text-align: right"&gt;5.17x&lt;/td&gt;
 &lt;td&gt;Not mere repair cost but an industrial-reopening package&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Stored-crude gross value&lt;/td&gt;
 &lt;td&gt;100m barrels x ~$79&lt;/td&gt;
 &lt;td style="text-align: right"&gt;~$7.9bn&lt;/td&gt;
 &lt;td&gt;The most immediate cash flow&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The IMF puts Iran&amp;rsquo;s 2026 nominal GDP at $300.29 billion, real growth at -6.1%, and inflation at 68.9%. (&lt;a class="link" href="https://www.imf.org/external/datamapper/profile/IRN" title="IMF DataMapper — Iran"
 target="_blank" rel="noopener"
 &gt;IMF&lt;/a&gt;) Rystad Energy estimated that repairing energy infrastructure from the Middle East war could cost up to $58 billion, of which oil &amp;amp; gas facilities could be up to $50 billion. (&lt;a class="link" href="https://www.reuters.com/business/energy/middle-east-war-damage-energy-assets-may-cost-up-58-billion-research-firm-rystad-2026-04-15/" title="Middle East war damage to energy assets may cost up to $58 billion — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;)&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="5-the-transmission-path-into-the-korean-stock-market"&gt;5. The Transmission Path Into the Korean Stock Market
&lt;/h2&gt;&lt;h3 id="51-first-channel--lower-oil-prices"&gt;5.1 First Channel — Lower Oil Prices
&lt;/h3&gt;&lt;p&gt;Korea is a net energy importer. A smaller Hormuz risk and a resumption of Iranian crude sales are broadly positive for the macro.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Sector&lt;/th&gt;
 &lt;th&gt;Impact&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Airlines&lt;/td&gt;
 &lt;td&gt;Lower fuel cost -&amp;gt; margin improvement&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Chemicals&lt;/td&gt;
 &lt;td&gt;Lower naphtha price -&amp;gt; possible relief in input costs&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Refining&lt;/td&gt;
 &lt;td&gt;Falling crude is an inventory-valuation-loss risk. Need to check refining margins&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Brokerages&lt;/td&gt;
 &lt;td&gt;Risk-on, higher trading value&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Semiconductors&lt;/td&gt;
 &lt;td&gt;Not a direct beneficiary, but a supporting variable via foreign risk-on&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;But refining is not a simple beneficiary. Falling crude can create inventory valuation losses, and the real profit is decided by the refining margin (crack spread).&lt;/p&gt;
&lt;h3 id="52-second-channel--risk-on-in-the-korean-market"&gt;5.2 Second Channel — Risk-On in the Korean Market
&lt;/h3&gt;&lt;p&gt;The Korean market has already reacted once. Yonhap reported that on June 15, 2026 the KOSPI rose 5.2% to close at 8,545.98 on the U.S.-Iran agreement to end the war and expectations of Hormuz&amp;rsquo;s reopening, and that the next day it added 2.11% to close at 8,726.6. (&lt;a class="link" href="https://en.yna.co.kr/view/AEN20260615007000320" title="Seoul stocks up for 3rd day above 8500 on U.S.-Iran deal — 연합뉴스"
 target="_blank" rel="noopener"
 &gt;Yonhap News&lt;/a&gt;) This reaction is not because of Iran reconstruction alone — a smaller Middle East tail risk, a lower oil-price premium, a possible easing of the burden on the U.S. rate path, a recovery of foreign risk-on, higher trading value, and cyclical rotation all worked together. So looking only at &amp;ldquo;Iran reconstruction plays&amp;rdquo; is too narrow. The bigger frame is &lt;strong&gt;a Korean cyclical re-rating&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="53-third-channel--the-reconstruction--plant--power-equipment-option"&gt;5.3 Third Channel — The Reconstruction / Plant / Power-Equipment Option
&lt;/h3&gt;&lt;p&gt;The reconstruction fund&amp;rsquo;s targets, as reported by Reuters, are energy, logistics, manufacturing and transport. (&lt;a class="link" href="https://www.reuters.com/business/finance/iran-deal-includes-300-billion-fund-more-than-half-which-already-committed-2026-06-16/" title="Iran deal includes $300 billion fund, more than half already committed — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;) These four areas overlap with fields where Korean firms are strong (refining and petchem plants, gas processing, power generation and transmission/distribution, ports, airports and logistics, steel and manufacturing equipment, shipbuilding and offshore plant, plant components). The core is &lt;strong&gt;not general construction but EPC, power and components&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="54-fourth-channel--the-gateway-of-sanctions-settlement-and-insurance"&gt;5.4 Fourth Channel — The Gateway of Sanctions, Settlement and Insurance
&lt;/h3&gt;&lt;p&gt;For actual investment to be possible, U.S. OFAC documents are needed. As of June 2026, OFAC still keeps the Iran sanctions as an active program. (&lt;a class="link" href="https://ofac.treasury.gov/sanctions-programs-and-country-information" title="Sanctions Programs and Country Information — OFAC"
 target="_blank" rel="noopener"
 &gt;OFAC&lt;/a&gt;) For Korean firms to move, at minimum the following must be clear: (1) crude and petchem trade waivers, (2) permission for bank settlement, (3) permission for shipping, insurance and reinsurance, (4) permission for EPC contracts, advance payments and guarantees, (5) the scope of the SDN trade ban, (6) whether dollar settlement is possible, and (7) whether secondary-sanctions immunity applies. Without these documents, an order may be &amp;ldquo;news&amp;rdquo; but it is not earnings.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="6-investment-priorities-by-sector"&gt;6. Investment Priorities by Sector
&lt;/h2&gt;&lt;p&gt;The names below are &lt;strong&gt;basket examples&lt;/strong&gt;. This is an organization of investment hypotheses, not a buy recommendation.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th style="text-align: right"&gt;Preference&lt;/th&gt;
 &lt;th&gt;Basket&lt;/th&gt;
 &lt;th&gt;Examples&lt;/th&gt;
 &lt;th&gt;Investment logic&lt;/th&gt;
 &lt;th&gt;Key risk&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;1&lt;/td&gt;
 &lt;td&gt;Power equipment, cable, transmission/distribution&lt;/td&gt;
 &lt;td&gt;HD Hyundai Electric, Hyosung Heavy Industries, LS ELECTRIC, LS, Iljin Electric, Taihan Cable &amp;amp; Solution&lt;/td&gt;
 &lt;td&gt;Even without Iran reconstruction, there is demand from AI data centers, the U.S. power grid and Middle East power investment&lt;/td&gt;
 &lt;td&gt;Valuations already high, peak-order worries&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;2&lt;/td&gt;
 &lt;td&gt;Plant components&lt;/td&gt;
 &lt;td&gt;Sungkwang Bend, TK Corporation, Hyrock Korea, BMT, DK-Lok&lt;/td&gt;
 &lt;td&gt;Profitability, delivery and order visibility can be better than EPC&lt;/td&gt;
 &lt;td&gt;If projects slip, downstream orders are delayed&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;3&lt;/td&gt;
 &lt;td&gt;Large EPC&lt;/td&gt;
 &lt;td&gt;Samsung E&amp;amp;A, Hyundai E&amp;amp;C, DL E&amp;amp;C, GS E&amp;amp;C, Daewoo E&amp;amp;C&lt;/td&gt;
 &lt;td&gt;Direct beneficiary of refining, gas, petchem and infrastructure repair&lt;/td&gt;
 &lt;td&gt;Sanctions, settlement, guarantees, cost ratios, schedule slippage&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;4&lt;/td&gt;
 &lt;td&gt;Shipbuilding, offshore&lt;/td&gt;
 &lt;td&gt;HD Korea Shipbuilding, HD Hyundai Heavy Industries, HD Hyundai Mipo, Samsung Heavy Industries, Hanwha Ocean&lt;/td&gt;
 &lt;td&gt;Normalized crude and gas exports; tanker, LNG and offshore-plant options&lt;/td&gt;
 &lt;td&gt;Time needed before orders; supercycle already priced in&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;5&lt;/td&gt;
 &lt;td&gt;Airlines, brokerages&lt;/td&gt;
 &lt;td&gt;Korean Air, Jeju Air, Kiwoom Securities, Mirae Asset Securities, etc.&lt;/td&gt;
 &lt;td&gt;Beneficiaries of lower oil, higher trading value, risk-on&lt;/td&gt;
 &lt;td&gt;Quick to price in short term; chasing is risky&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;6&lt;/td&gt;
 &lt;td&gt;Refining, chemicals&lt;/td&gt;
 &lt;td&gt;SK Innovation, S-Oil, Lotte Chemical, etc.&lt;/td&gt;
 &lt;td&gt;Possible lower input costs&lt;/td&gt;
 &lt;td&gt;Refining margins and spreads are the crux. Not a simple beneficiary of lower oil&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;7&lt;/td&gt;
 &lt;td&gt;Small-cap reconstruction theme&lt;/td&gt;
 &lt;td&gt;Low-liquidity construction / component theme stocks&lt;/td&gt;
 &lt;td&gt;Possible short-term surge&lt;/td&gt;
 &lt;td&gt;Thin substance; sharp drop risk once flows turn&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The best structure is &lt;strong&gt;&amp;ldquo;a firm where Iran reconstruction is extra upside if it happens, and core demand is alive even if it doesn&amp;rsquo;t.&amp;quot;&lt;/strong&gt; On that criterion, power equipment, cable and plant components offer better risk/reward than large EPC.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="7-market-reaction-and-current-position"&gt;7. Market Reaction and Current Position
&lt;/h2&gt;&lt;p&gt;The Korean market has already reacted once. The Seoul Economic Daily reported that on the MOU news construction, airline and brokerage stocks rose together, with Samsung E&amp;amp;A up 10.71%. (&lt;a class="link" href="https://en.sedaily.com/news/2026/06/15/wars-end-hopes-lift-korean-builders-airlines-brokerages" title="War&amp;#39;s End Hopes Lift Korean Builders, Airlines, Brokerages — Seoul Economic Daily"
 target="_blank" rel="noopener"
 &gt;Seoul Economic Daily&lt;/a&gt;) So the judgment now is this.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Bucket&lt;/th&gt;
 &lt;th&gt;Investment call&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Construction stocks that already surged&lt;/td&gt;
 &lt;td&gt;Need pullback / confirmation rather than chasing&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Airlines, brokerages&lt;/td&gt;
 &lt;td&gt;Short-term-trade in character. Check oil and trading value&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Power equipment, plant components&lt;/td&gt;
 &lt;td&gt;Core medium-term basket&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Large EPC&lt;/td&gt;
 &lt;td&gt;Add weight after OFAC waivers and actual LOIs&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Small-cap reconstruction theme&lt;/td&gt;
 &lt;td&gt;Avoid first&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The Korean government has also been reported as not yet having received a formal request from the U.S. and as taking a stance of monitoring the situation. (&lt;a class="link" href="https://en.sedaily.com/politics/2026/06/16/korea-weighs-role-in-300-billion-iran-reconstruction-fund" title="Korea Weighs Role in $300 Billion Iran Reconstruction Fund — Seoul Economic Daily"
 target="_blank" rel="noopener"
 &gt;Seoul Economic Daily&lt;/a&gt;) The Hankyoreh likewise reported that while Korean firms&amp;rsquo; participation was reported, specific company names, amounts and binding force have not been disclosed. (&lt;a class="link" href="https://english.hani.co.kr/arti/english_edition/e_international/1264019.html" title="Korean firms reportedly among backers of $300B Iran fund — Hani English"
 target="_blank" rel="noopener"
 &gt;Hani English&lt;/a&gt;) In other words, it is still too early to price in &amp;ldquo;confirmed Korean-firm participation.&amp;rdquo;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="8-investment-strategy"&gt;8. Investment Strategy
&lt;/h2&gt;&lt;h3 id="81-short-term--verify-rather-than-chase"&gt;8.1 Short Term — Verify Rather Than Chase
&lt;/h3&gt;&lt;p&gt;The construction, airline and brokerage stocks that have already had a first surge are an event-driven reaction. What matters is not the feeling of having missed out but &lt;strong&gt;the order in which substance gets confirmed&lt;/strong&gt;. The principles: (1) no chasing the surged construction stocks; (2) for oil-beneficiary names, check cost and margin sensitivity; (3) for brokerages, confirm the persistence of trading value; (4) for defense stocks, factor in a possible partial reduction of the Middle East premium; and (5) semiconductors can benefit from risk-on flows separate from Iran reconstruction.&lt;/p&gt;
&lt;h3 id="82-medium-term--centered-on-power-equipment-and-plant-components"&gt;8.2 Medium Term — Centered on Power Equipment and Plant Components
&lt;/h3&gt;&lt;p&gt;Over the medium term, the most reasonable are: (1) power equipment, cable, transmission/distribution; (2) plant components; (3) among large EPC, firms with good financials, order backlog and cost ratios; and (4) among shipbuilders/offshore, firms with tanker, LNG and offshore-plant options. The reason is simple. Even if Iran reconstruction is delayed, this basket has separate structural demand from &lt;strong&gt;AI data centers, the U.S. power grid, Middle East infrastructure and global energy capex&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="83-after-a-final-accord--numbers-not-themes"&gt;8.3 After a Final Accord — Numbers, Not Themes
&lt;/h3&gt;&lt;p&gt;Once a final accord is signed and Korean company names are disclosed, from then on it must be seen in numbers.&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"&gt;&lt;code class="language-text" data-lang="text"&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;Contract value = contract amount x expected operating margin x collectability x sanctions-risk discount rate
&lt;/span&gt;&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;&lt;/div&gt;&lt;p&gt;Even a $1 billion order, with a 3% operating margin, 30% collection risk and large sanctions risk, leaves limited shareholder value. Conversely, even a $300 million component order, at a 15% operating margin with clear advance payments, LCs and insurance, is worth more.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="9-core-monitoring-checklist"&gt;9. Core Monitoring Checklist
&lt;/h2&gt;&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Checkpoint&lt;/th&gt;
 &lt;th&gt;Why it matters&lt;/th&gt;
 &lt;th&gt;Investment call&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Final MOU text published&lt;/td&gt;
 &lt;td&gt;Whether the 14-article draft is actually reflected&lt;/td&gt;
 &lt;td&gt;If it differs from the draft, revise the thesis&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Official signing&lt;/td&gt;
 &lt;td&gt;First confirmation of the political event&lt;/td&gt;
 &lt;td&gt;If unsigned, the theme cools sharply&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;OFAC general license / waiver&lt;/td&gt;
 &lt;td&gt;The crux of actual deal feasibility&lt;/td&gt;
 &lt;td&gt;Without it, EPC orders are impossible&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Bank, insurance, shipping waivers&lt;/td&gt;
 &lt;td&gt;Whether settlement, shipping and insurance are possible&lt;/td&gt;
 &lt;td&gt;Premise for Korean-firm participation&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;IAEA inspection / enriched-uranium handling&lt;/td&gt;
 &lt;td&gt;Durability of the final accord&lt;/td&gt;
 &lt;td&gt;If it fails, the fund is near zero&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Disclosure of Korean company names&lt;/td&gt;
 &lt;td&gt;Rumor -&amp;gt; investable event&lt;/td&gt;
 &lt;td&gt;Check disclosures and press releases&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Contract terms&lt;/td&gt;
 &lt;td&gt;Profitability / cash-flow judgment&lt;/td&gt;
 &lt;td&gt;Margin matters more than contract amount&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Brent/WTI stabilization&lt;/td&gt;
 &lt;td&gt;Airline, chemical, refining sensitivity&lt;/td&gt;
 &lt;td&gt;A sharp oil rebound damages the thesis&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Israel / Lebanon variable&lt;/td&gt;
 &lt;td&gt;Spoiler risk&lt;/td&gt;
 &lt;td&gt;A re-escalation means risk-off&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;hr&gt;
&lt;h2 id="10-risk-factors"&gt;10. Risk Factors
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Political risk&lt;/strong&gt;: The MOU is not a final accord. Trump was reported to have said in effect that he could take military action again if he did not like the deal. (&lt;a class="link" href="https://www.reuters.com/world/middle-east/g7-leaders-demand-ceasefire-lebanon-welcome-iran-deal-2026-06-17/" title="Trump on Iran deal: &amp;#39;If I don&amp;#39;t like it, we&amp;#39;ll go back to shooting&amp;#39; — Reuters"
 target="_blank" rel="noopener"
 &gt;Reuters&lt;/a&gt;)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Nuclear-talks risk&lt;/strong&gt;: The ban on building nuclear weapons gets documented, but enriched uranium, inspection and nuclear needs are for follow-on talks. If it breaks down there, the reconstruction fund and sanctions relief could be delayed or fall through.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Sanctions risk&lt;/strong&gt;: If OFAC documents are unclear, Korean firms find it hard to move. If bank, insurance, reinsurance and dollar settlement are uncertain, EPC contracts stay headlines rather than earnings.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Investor-overheating risk&lt;/strong&gt;: A &amp;ldquo;leading Iran reconstruction stock&amp;rdquo;-style small-cap rally can see flows move ahead of substance. The upside is large but the loss is fast. From an institutional standpoint it is better to avoid.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="11-final-investment-conclusion"&gt;11. Final Investment Conclusion
&lt;/h2&gt;&lt;div class="thesis-callout"&gt;
 &lt;div class="thesis-callout__label"&gt;Bottom line&lt;/div&gt;
 &lt;div class="thesis-callout__body"&gt;
 Iran reconstruction did not hand the Korean market a construction-stock theme; through the removal of the Middle East risk premium and a sanctions-relief option, it handed it a rationale for a cyclical re-rating that runs through power equipment, plants, shipbuilding, airlines and brokerages.
 &lt;/div&gt;
&lt;/div&gt;
&lt;p&gt;The most reasonable actions now are these.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;No chasing of small-cap reconstruction theme stocks&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Keep power equipment, cable and plant components as the core basket&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Select large EPC only after OFAC waivers and actual LOIs&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Approach airlines and brokerages only as short-term trades&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;For refining and chemicals, check the crack spread and product spreads rather than oil prices&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Read semiconductors not as an alternative Iran-reconstruction theme but as a supporting risk-on beneficiary&lt;/strong&gt;&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;The winner on the text of the war is Iran, the short-term market winner is risk-on assets, and the Korean investor&amp;rsquo;s medium-term alpha lies in power equipment, plant components and high-quality EPC.&lt;/p&gt;
&lt;p&gt;&lt;small&gt;The facts in this piece cite the reports and official materials noted in the text; what is public now is not a final treaty but the MOU/draft stage. Company names are not an investment recommendation but basket examples illustrating the flow of analysis, and the actual investment judgment and responsibility rest with the investor.&lt;/small&gt;&lt;/p&gt;
&lt;h2 id="sources"&gt;Sources
&lt;/h2&gt;</description></item><item><title>Why the AI Supercycle Keeps Stretching: Agent Demand, IPO Funding, and the Cheapest Asset Still Standing — Memory and Storage</title><link>https://koreainvestinsights.com/post/ai-supercycle-extension-agent-demand-ipo-funding-memory-storage-2026-06-12/</link><pubDate>Fri, 12 Jun 2026 15:30:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/ai-supercycle-extension-agent-demand-ipo-funding-memory-storage-2026-06-12/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;This piece is the synthesis of a month of AI-infrastructure analysis. It connects &lt;a class="link" href="https://koreainvestinsights.com/post/broadcom-ai-semiconductor-100b-2027-korea-value-chain-2026-06-05/" &gt;Broadcom&amp;rsquo;s reaffirmed USD 100 billion target&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/marvell-trillion-broadcom-readthrough-korea-ai-connectivity-2026-06-02/" &gt;Jensen Huang&amp;rsquo;s trillion-dollar Marvell comment&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/" &gt;Goldman&amp;rsquo;s token demand vs. J.P. Morgan&amp;rsquo;s memory ASP peak-out&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/ai-datacenter-capex-5p3t-korea-power-substrate-storage-bottleneck-2026-06-05/" &gt;the USD 5.3 trillion AI data-center CapEx era&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/spacex-ipo-korea-market-liquidity-ai-space-readthrough-2026-06-05/" &gt;SpaceX&amp;rsquo;s listing and the Korean market&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/samsung-hynix-micron-forward-per-parity-memory-catch-up-2026-06-03/" &gt;Samsung-Hynix-Micron parity&lt;/a&gt; into a single map of the cycle.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;The structure of the AI supercycle is shifting. &lt;strong&gt;(1) An explosion in token demand created by AI agents → (2) expanded funding by Big Tech → (3) public-market funding via mega-IPOs like SpaceX and Anthropic&lt;/strong&gt; — the fuel hands off like a relay, and the lifespan of the cycle itself is lengthening.&lt;/li&gt;
&lt;li&gt;Nomura has priced this shift in most aggressively. Per reporting, it set &lt;strong&gt;a Samsung Electronics target of KRW 590,000, an SK Hynix target of KRW 5,000,000, and a KOSPI target of 10,000-11,000&lt;/strong&gt;, describing this year&amp;rsquo;s monthly memory revenue as &amp;ldquo;a vertical rise the likes of which we have never seen before.&amp;rdquo;&lt;/li&gt;
&lt;li&gt;The longer the cycle runs, the more the real question becomes: &amp;ldquo;Which AI stock is still left?&amp;rdquo; The answer is surprisingly simple. &lt;strong&gt;Memory and storage — the final destination of demand, yet carrying the lowest multiples.&lt;/strong&gt; Samsung Electronics and SK Hynix trade at a relative P/E of 0.80-0.82x versus Micron, and at roughly 6x on 2028 estimated earnings — among the lowest in the entire AI-chip basket.&lt;/li&gt;
&lt;li&gt;The counter-argument is real too: J.P. Morgan&amp;rsquo;s slowing 2027 ASP growth, rate pressure, and earnings-bubble warnings. So the conclusion is not &amp;ldquo;buy at any price&amp;rdquo; but &lt;strong&gt;&amp;ldquo;bet on the unwinding of memory and storage&amp;rsquo;s undervaluation, while confirming the evidence of cycle extension step by step.&amp;quot;&lt;/strong&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-the-cycles-fuel-hands-off-in-three-stages"&gt;1. The Cycle&amp;rsquo;s Fuel Hands Off in Three Stages
&lt;/h2&gt;&lt;p&gt;Until now, the core of AI-cycle skepticism was &amp;ldquo;the money dries up first.&amp;rdquo; The logic: once the cash going into GPUs and data centers exceeds Big Tech&amp;rsquo;s cash flow, investment eventually rolls over and the cycle ends. Yet stitching together the news flow of the past month reveals a different picture — funding sources handing off one stage at a time, extending the cycle.&lt;/p&gt;
&lt;h3 id="stage-1--demand-agents-detonate-token-consumption"&gt;Stage 1 — Demand: Agents Detonate Token Consumption
&lt;/h3&gt;&lt;p&gt;The starting point is not humans but agents. Token demand in the era when a person typed a question into a chat window is in a different dimension from the era when agents search, run code, and read documents on their own to get work done. Goldman Sachs projected that &lt;a class="link" href="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/" &gt;token usage will rise 24-fold by 2030&lt;/a&gt;, and Nomura&amp;rsquo;s head of research, Chang-won Jung, went a step further, even using the phrase &amp;ldquo;AI-driven memory demand will grow 10,000x-20,000x in 5 years&amp;rdquo; (citing Sisa Journal reporting). What matters more than the precision of the numbers is the direction. &lt;strong&gt;It is now machine usage, not human usage, that determines the slope of the demand curve.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;This demand is already confirmed in chip makers&amp;rsquo; actual order books. &lt;a class="link" href="https://koreainvestinsights.com/post/broadcom-ai-semiconductor-100b-2027-korea-value-chain-2026-06-05/" &gt;Broadcom reaffirmed AI-semiconductor revenue exceeding USD 100 billion in 2027&lt;/a&gt;, and Jensen Huang &lt;a class="link" href="https://koreainvestinsights.com/post/marvell-trillion-broadcom-readthrough-korea-ai-connectivity-2026-06-02/" &gt;publicly stated that Marvell could become a trillion-dollar company&lt;/a&gt;. As the customer base broadens to custom accelerators (XPUs), TPUs, and OpenAI&amp;rsquo;s own accelerators, demand has become multi-axis rather than resting on Nvidia alone.&lt;/p&gt;
&lt;h3 id="stage-2--funding-big-tech-builds-even-on-borrowed-money"&gt;Stage 2 — Funding: Big Tech Builds Even on Borrowed Money
&lt;/h3&gt;&lt;p&gt;The scale of money required to meet this demand has swelled to the unit of &lt;a class="link" href="https://koreainvestinsights.com/post/ai-datacenter-capex-5p3t-korea-power-substrate-storage-bottleneck-2026-06-05/" &gt;USD 5.3 trillion in data-center CapEx&lt;/a&gt;. Alphabet, Microsoft, Meta, and Amazon can no longer cover it with internal cash alone, so they are issuing more corporate bonds — and still not cutting investment. The reason is simple. &lt;strong&gt;Inference cost (cost per token) is falling 60-70% a year, creating a structure where margins improve the more you use&lt;/strong&gt;. Whoever stops investing loses share.&lt;/p&gt;
&lt;h3 id="stage-3--a-new-funding-source-the-public-market-opens"&gt;Stage 3 — A New Funding Source: The Public Market Opens
&lt;/h3&gt;&lt;p&gt;This is the new chapter of this cycle. Until now, AI-infrastructure capital came from inside Big Tech (cash + corporate bonds). But with &lt;a class="link" href="https://koreainvestinsights.com/post/spacex-ipo-korea-market-liquidity-ai-space-readthrough-2026-06-05/" &gt;SpaceX filing its S-1 and, per reporting, pursuing a USD 75 billion offering&lt;/a&gt;, &lt;strong&gt;a third funding channel — the public market&lt;/strong&gt; — is opening. Anthropic is likewise a mega-candidate whose listing is being talked about (we treat it as a scenario, as it is not officially confirmed). If these IPOs go through, retail and passive money will join the funding pipeline for AI-infrastructure investment.&lt;/p&gt;
&lt;p&gt;In summary:&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Stage&lt;/th&gt;
 &lt;th&gt;Funding source&lt;/th&gt;
 &lt;th&gt;Status&lt;/th&gt;
 &lt;th&gt;Meaning&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Stage 1&lt;/td&gt;
 &lt;td&gt;Big Tech internal cash&lt;/td&gt;
 &lt;td&gt;Ongoing&lt;/td&gt;
 &lt;td&gt;Improving inference margins create reinvestment capacity&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Stage 2&lt;/td&gt;
 &lt;td&gt;Corporate bonds / external borrowing&lt;/td&gt;
 &lt;td&gt;Expanding&lt;/td&gt;
 &lt;td&gt;Rate pressure exists, but no signal of investment cuts yet&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Stage 3&lt;/td&gt;
 &lt;td&gt;IPO public-market funding&lt;/td&gt;
 &lt;td&gt;Emerging&lt;/td&gt;
 &lt;td&gt;SpaceX has an official S-1; Anthropic is at the speculative stage&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The more diversified the funding sources, the lower the probability of the &amp;ldquo;money dries up and the cycle ends&amp;rdquo; scenario. &lt;strong&gt;This is the essence of the cycle&amp;rsquo;s extension.&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-what-nomura-saw-a-revenue-curve-rising-vertically"&gt;2. What Nomura Saw: A Revenue Curve Rising Vertically
&lt;/h2&gt;&lt;p&gt;The house that translated this structural change into price targets most aggressively is Nomura. Organized per reporting (Sisa Journal, 2026-06-12 media briefing):&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Item&lt;/th&gt;
 &lt;th&gt;Nomura&amp;rsquo;s call&lt;/th&gt;
 &lt;th&gt;Note&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Samsung Electronics target&lt;/td&gt;
 &lt;td&gt;&lt;strong&gt;KRW 590,000&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;A large upgrade versus the current price&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;SK Hynix target&lt;/td&gt;
 &lt;td&gt;&lt;strong&gt;KRW 5,000,000&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Top pick of the memory supercycle&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;KOSPI target&lt;/td&gt;
 &lt;td&gt;10,000-11,000&lt;/td&gt;
 &lt;td&gt;Cites a 60% probability of MSCI developed-index inclusion&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Core rationale&lt;/td&gt;
 &lt;td&gt;&amp;ldquo;This year&amp;rsquo;s monthly memory revenue is a vertical rise, a pattern never seen before&amp;rdquo;&lt;/td&gt;
 &lt;td&gt;Chang-won Jung, Co-Head of Asia Research&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The rationale matters more than the target itself. Nomura&amp;rsquo;s logic overlaps precisely with the picture we have tracked for a month. Past memory cycles rose and fell roughly once every 2-3 years, following PC and smartphone replacement cycles. This time, &lt;strong&gt;agents — a new demand source — have layered structural growth on top of the cycle&lt;/strong&gt;, and as a result the monthly revenue curve is climbing at a slope unlike any prior cycle.&lt;/p&gt;
&lt;p&gt;Of course, a foreign house&amp;rsquo;s aggressive target cannot in itself be a reason to buy. Still, the fact that this upgrade came right as &lt;a class="link" href="https://koreainvestinsights.com/post/foreign-return-after-24-day-kospi-selling-memory-rebalance-2026-06-12/" &gt;foreigners began returning after 24 sessions of selling&lt;/a&gt; can be read as a signal that the direction of foreign flows and the sell-side view have started pointing the same way.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-so-whats-left-the-most-powerful-and-the-cheapest"&gt;3. So What&amp;rsquo;s Left: The Most Powerful and the Cheapest
&lt;/h2&gt;&lt;p&gt;Even if you agree the cycle is lengthening, what to buy in an already much-higher market is a separate question. Here the valuation map gives the answer.&lt;/p&gt;
&lt;p&gt;Let&amp;rsquo;s lay out again the figures we confirmed in the &lt;a class="link" href="https://koreainvestinsights.com/post/samsung-hynix-micron-forward-per-parity-memory-catch-up-2026-06-03/" &gt;Samsung-Hynix-Micron parity analysis&lt;/a&gt; and the &lt;a class="link" href="https://koreainvestinsights.com/post/sam-hama-parity-follow-up-ai-chip-memory-pe-map-2026-06-05/" &gt;AI-chip P/E map follow-up&lt;/a&gt;.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Asset&lt;/th&gt;
 &lt;th&gt;Valuation&lt;/th&gt;
 &lt;th&gt;AI demand exposure&lt;/th&gt;
 &lt;th&gt;Verdict&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;US AI-chip large caps&lt;/td&gt;
 &lt;td&gt;High multiples maintained&lt;/td&gt;
 &lt;td&gt;Direct&lt;/td&gt;
 &lt;td&gt;Strong but already expensive&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Micron&lt;/td&gt;
 &lt;td&gt;US-listed AI-memory premium&lt;/td&gt;
 &lt;td&gt;Direct&lt;/td&gt;
 &lt;td&gt;Premium is justified, but expensive versus Korea&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;Samsung Electronics&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Relative P/E 0.82x vs. Micron, ~6x 2028E P/E&lt;/td&gt;
 &lt;td&gt;Direct (HBM4E catch-up + eSSD + foundry option)&lt;/td&gt;
 &lt;td&gt;&lt;strong&gt;Cheapest direct exposure&lt;/strong&gt;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;SK Hynix&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Relative P/E 0.80x vs. Micron&lt;/td&gt;
 &lt;td&gt;Direct (HBM leader)&lt;/td&gt;
 &lt;td&gt;&lt;strong&gt;Discounted versus its tech edge&lt;/strong&gt;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Storage (eSSD / NAND)&lt;/td&gt;
 &lt;td&gt;Early-stage multiple re-rating&lt;/td&gt;
 &lt;td&gt;Direct (inference / RAG / KV-cache storage)&lt;/td&gt;
 &lt;td&gt;A zone where reclassification is underway&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The core logic compresses into three lines.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;The final destination of demand is memory.&lt;/strong&gt; Whether agents burn tokens or SpaceX builds satellite networks and AI compute, what is ultimately needed is HBM, server DRAM, and eSSD. Memory always sits next to the GPU, and as inference grows, storage demand follows. An AI server uses roughly 3x the SSD of a general server.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;And yet it is priced as if it were late-cycle.&lt;/strong&gt; Among assets exposed to the same AI demand, only Korean memory trades at roughly 6x on 2028 estimated earnings. It means the market has not yet fully shed the old-cycle frame of &amp;ldquo;memory is about to roll over.&amp;rdquo;&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The longer the cycle runs, the harder this discount is to sustain.&lt;/strong&gt; As the funding source hands off through three stages and the lifespan of demand extends, the &amp;ldquo;about to roll over&amp;rdquo; assumption breaks down, and the room for multiples to normalize grows. Nomura&amp;rsquo;s targets are numbers that pre-empt that normalization.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;Storage sits one step earlier in the sequence. As we saw in the &lt;a class="link" href="https://koreainvestinsights.com/post/fadu-ai-infra-storage-bottleneck-p-q-new-segment-2026-06-02/" &gt;Fadu-SanDisk analysis&lt;/a&gt;, NAND and eSSD are in the early stage of being reclassified from &amp;ldquo;always the worse cycle asset than DRAM&amp;rdquo; to &amp;ldquo;the storage bottleneck of AI inference.&amp;rdquo; If the reclassification is correct, the upside is large — but so too are the items to verify (price, volume, new markets) remain many.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="4-red-team-when-this-logic-is-wrong"&gt;4. Red Team: When This Logic Is Wrong
&lt;/h2&gt;&lt;p&gt;The more comprehensive the piece, the more clearly the counter-argument must be written.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;When J.P. Morgan&amp;rsquo;s peak-out is right.&lt;/strong&gt; The year-over-year growth rate of DRAM and NAND ASPs may slow from 2027. As we saw in the &lt;a class="link" href="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/" &gt;P-Q-C separation analysis&lt;/a&gt;, this is a deceleration in the pace of increase, not a price crash — but share prices react first to momentum, not earnings. Momentum deceleration can arrive before the memory multiple normalizes.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;When rates squeeze the cycle.&lt;/strong&gt; As organized in the &lt;a class="link" href="https://koreainvestinsights.com/post/ai-supercycle-midgame-rate-risk-yellow-not-red-2026-06-06/" &gt;AI supercycle mid-game check&lt;/a&gt;, the more funding shifts toward corporate bonds and IPOs, the more sensitive the cycle becomes to rates. Infrastructure built on debt and public offerings wobbles first when rates rise. It is a yellow light right now, not a red one — but this signal must be watched continuously.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;When IPOs absorb funds rather than supply them.&lt;/strong&gt; If the SpaceX and Anthropic offerings are funded by selling existing AI positions, then in the short term liquid AI large caps like Samsung Electronics and SK Hynix could become the sell target (the funding source). The material for cycle extension is also the material for short-term volatility.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;When rhetoric like &amp;ldquo;10,000x, 20,000x&amp;rdquo; is itself a sign of overheating.&lt;/strong&gt; There is also the classic counter that the sell-side&amp;rsquo;s &amp;ldquo;vertical rise&amp;rdquo; rhetoric is itself a hallmark of late cycle. The warning from the &lt;a class="link" href="https://koreainvestinsights.com/post/ai-infra-earnings-bubble-deep-dive-research-thesis-os-2026-05-31/" &gt;earnings-bubble analysis&lt;/a&gt; — that when earnings are a bubble, share prices fall first, before estimates are cut — applies equally to the logic of this piece.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="5-checklist-how-to-confirm-cycle-extension-step-by-step"&gt;5. Checklist: How to Confirm Cycle Extension Step by Step
&lt;/h2&gt;&lt;p&gt;Things to confirm, instead of to predict.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Item to check&lt;/th&gt;
 &lt;th&gt;Strengthens the extension thesis&lt;/th&gt;
 &lt;th&gt;Weakens the extension thesis&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Token / agent usage&lt;/td&gt;
 &lt;td&gt;High growth sustained&lt;/td&gt;
 &lt;td&gt;Growth rate slows&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Big Tech CapEx guidance&lt;/td&gt;
 &lt;td&gt;Maintained / raised&lt;/td&gt;
 &lt;td&gt;Cut, or emphasis on &amp;ldquo;efficiency&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Corporate bond rates / spreads&lt;/td&gt;
 &lt;td&gt;Stable&lt;/td&gt;
 &lt;td&gt;Spreads widen&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;SpaceX IPO&lt;/td&gt;
 &lt;td&gt;Strong demand, stable trading&lt;/td&gt;
 &lt;td&gt;Offering shrinks, post-listing plunge&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Anthropic listing status&lt;/td&gt;
 &lt;td&gt;Enters official process&lt;/td&gt;
 &lt;td&gt;Indefinitely postponed&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;HBM long-term contracts&lt;/td&gt;
 &lt;td&gt;Prices held / raised&lt;/td&gt;
 &lt;td&gt;Renegotiation, volume deferral&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Korean memory relative P/E&lt;/td&gt;
 &lt;td&gt;Normalizes from 0.8x toward 1.0x&lt;/td&gt;
 &lt;td&gt;Re-declines below 0.8x&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Foreign flows&lt;/td&gt;
 &lt;td&gt;Net buying sustained / broadening&lt;/td&gt;
 &lt;td&gt;Reverts to net selling again&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;hr&gt;
&lt;h2 id="conclusion"&gt;Conclusion
&lt;/h2&gt;&lt;p&gt;The skepticism that what ends the AI supercycle is not demand but money was only half right. Demand is being detonated by agents, and money is relaying from Big Tech cash to corporate bonds and on to the public market, expanding its sources. A cycle with diversified funding runs longer than past ones.&lt;/p&gt;
&lt;p&gt;And in a lengthened cycle, the most advantageous seat is the asset that is the final destination of demand yet still sits at late-cycle pricing. &lt;strong&gt;That is memory, and one step earlier, storage.&lt;/strong&gt; Nomura&amp;rsquo;s KRW 590,000 for Samsung Electronics and KRW 5,000,000 for SK Hynix are not in themselves a buy signal but a milestone that this re-rating has begun spreading into the sell-side consensus.&lt;/p&gt;
&lt;p&gt;The practical stance is one sentence. &lt;strong&gt;Confirm the evidence of cycle extension (token usage, CapEx, IPO reception, HBM contracts) step by step, weight toward the unwinding of Korean memory&amp;rsquo;s relative discount and the reclassification of storage, but ease off the pace if signals of rate pressure and ASP-momentum deceleration light up.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;&lt;small&gt;The Nomura targets and comments in this piece cite Sisa Journal reporting (2026-06-12 media briefing), and Anthropic&amp;rsquo;s listing is at a speculative stage prior to official confirmation. Mentions of specific names are not investment recommendations but examples illustrating an analytical flow; the actual investment decision and responsibility rest with the investor.&lt;/small&gt;&lt;/p&gt;</description></item><item><title>Goldman's Token Demand vs. J.P. Morgan's Memory ASP Peak-Out: Do the Two Calls Really Clash?</title><link>https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/</link><pubDate>Sun, 31 May 2026 11:00:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;This piece follows on from &lt;a class="link" href="https://koreainvestinsights.com/post/ai-token-futures-cost-per-token-korea-semiconductor-thesis-2026-05-30/" &gt;AI Token Futures and Cost per Token&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/kr-deep-dive-samsung-electronics-2026-04-16/" &gt;Samsung Electronics 2026 Deep Dive&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/kr-deep-dive-sk-hynix-2026-04-16/" &gt;SK Hynix: The HBM Heavyweight&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/nvidia-q1-fy27-korea-ai-infra-supply-chain-2026-05-21/" &gt;Korea AI Infrastructure After Nvidia Q1 FY27&lt;/a&gt;. Where those earlier posts looked separately at cost per token, individual companies, and the spread of AI infrastructure, this one &lt;strong&gt;merges two different calls — Goldman on demand and J.P. Morgan on memory pricing — into a single framework&lt;/strong&gt; and lays out the investment timeline from 2026 to 2030.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;Goldman Sachs sees &lt;strong&gt;an explosion in token usage (24x by 2030) plus a steep drop in cost per token (60-70% per year)&lt;/strong&gt;. J.P. Morgan sees &lt;strong&gt;the year-over-year (YoY) growth rate of DRAM and NAND ASP slowing from 2027&lt;/strong&gt;. Because they track different variables, the two calls do not clash head-on.&lt;/li&gt;
&lt;li&gt;Combine them and a single path appears: &lt;strong&gt;2026 is memory ASP beta, 2027 is a peak-out in price momentum, and 2028-2030 is a rotation of alpha into the components and systems that lower cost per token&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;So the investment conclusion is not &amp;ldquo;buy memory without limit.&amp;rdquo; In 2026, memory beta such as HBM, server DRAM and eSSD is favorable, but after that you have to &lt;strong&gt;be selective about where the bottlenecks are&lt;/strong&gt; — ASIC, AI networking, optical, HBM leaders, eSSD, advanced packaging, and MLCC/FC-BGA.&lt;/li&gt;
&lt;li&gt;The two most common misreadings are: (1) seeing J.P. Morgan&amp;rsquo;s peak-out and concluding &amp;ldquo;the AI infrastructure cycle is over,&amp;rdquo; and (2) seeing Goldman&amp;rsquo;s demand and concluding &amp;ldquo;memory prices keep surging through 2030.&amp;rdquo; Both go too far.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-two-giants-two-seemingly-opposite-calls"&gt;1. Two Giants, Two Seemingly Opposite Calls
&lt;/h2&gt;&lt;p&gt;Looking at the same AI era, two global investment banks have painted pictures that seem to point in opposite directions.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Goldman Sachs&lt;/strong&gt; (official article, May 5, 2026) looks at the demand side. The core is two points.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Token usage grows 24x by 2030.&lt;/strong&gt; The call is that monthly usage reaches 120 quadrillion tokens in 2030; working backward, that puts the 2026 base at roughly 5 quadrillion. That is about 121% average annual growth over four years.&lt;/li&gt;
&lt;li&gt;At the same time, &lt;strong&gt;inference cost per token falls 60-70% per year.&lt;/strong&gt; The drivers are improving chip efficiency and better data-center architecture.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;img alt="Goldman Sachs AI token economics reconstructed index — usage rises 24x while cost per token falls 39-123x" class="gallery-image" data-flex-basis="390px" data-flex-grow="162" height="1162" loading="lazy" sizes="(max-width: 767px) calc(100vw - 30px), (max-width: 1023px) 700px, (max-width: 1279px) 950px, 1232px" src="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-ai-token-economics-reconstructed-index.png" srcset="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-ai-token-economics-reconstructed-index_hu_590079f5c9217e9c.png 800w, https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-ai-token-economics-reconstructed-index_hu_9581b0be0cd08913.png 1600w, https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-ai-token-economics-reconstructed-index.png 1891w" width="1891"&gt;
&lt;/p&gt;
&lt;p&gt;&lt;small&gt;Source: a reconstructed chart that simply indexes the figures in Goldman Sachs&amp;rsquo;s official article (2026-05-05). It is not the original chart but the original numbers — &amp;ldquo;24x tokens by 2030, cost per token down 60-70% a year&amp;rdquo; — plotted on a log scale.&lt;/small&gt;&lt;/p&gt;
&lt;p&gt;In the chart above, the blue line (usage) climbs steeply, and the orange and green lines (cost per token) fall even more steeply. At a 60% annual decline, cost four years out is about 2.6% of the 2026 level (roughly a 39x improvement); at 70%, about 0.8% (roughly a 123x improvement).&lt;/p&gt;
&lt;p&gt;By contrast, the &lt;strong&gt;J.P. Morgan&lt;/strong&gt; material looks at price. The picture is that DRAM and NAND ASP rise strongly in 2026 but that, &lt;strong&gt;from late 2026 into early 2027, the year-over-year growth rate decelerates quickly.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;&lt;img alt="J.P. Morgan DRAM and NAND ASP year-over-year change — peaking in 2026, slowing from 2027" class="gallery-image" data-flex-basis="601px" data-flex-grow="250" height="271" loading="lazy" sizes="(max-width: 767px) calc(100vw - 30px), (max-width: 1023px) 700px, (max-width: 1279px) 950px, 1232px" src="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/jpm-dram-nand-asp-yoy-peakout-chart.png" width="679"&gt;
&lt;/p&gt;
&lt;p&gt;&lt;small&gt;Source: a session-uploaded image (DRAM from WSTS / J.P. Morgan estimates, NAND from Gartner / J.P. Morgan estimates). Detailed figures such as FY26 DRAM +53% and NAND +30%, FY27 DRAM +1% and NAND -6% are based on a secondary summary; the original table is unverified.&lt;/small&gt;&lt;/p&gt;
&lt;p&gt;On the surface, &amp;ldquo;demand explodes&amp;rdquo; (Goldman) and &amp;ldquo;memory price gains roll over&amp;rdquo; (J.P. Morgan) seem to clash. But look closely and the two calls are &lt;strong&gt;talking about different things.&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-why-its-not-a-clash--separating-p-q-and-c"&gt;2. Why It&amp;rsquo;s Not a Clash — Separating P, Q and C
&lt;/h2&gt;&lt;p&gt;A memory company&amp;rsquo;s profit breaks down into a simple multiplication.&lt;/p&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Revenue = shipments (Q) × average selling price (P), and the token economy = usage (Q) × cost per token (C).&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;p&gt;Through this lens, it becomes clear that the two calls are watching different variables.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Dimension&lt;/th&gt;
 &lt;th&gt;Goldman Sachs&lt;/th&gt;
 &lt;th&gt;J.P. Morgan&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Variable watched&lt;/td&gt;
 &lt;td&gt;Token usage (Q), cost per token (C)&lt;/td&gt;
 &lt;td&gt;YoY change in DRAM/NAND ASP (P momentum)&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Core message&lt;/td&gt;
 &lt;td&gt;AI usage explodes long term, cost per token plunges&lt;/td&gt;
 &lt;td&gt;Memory price growth rate slows from 2027&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Time horizon&lt;/td&gt;
 &lt;td&gt;2026-2030&lt;/td&gt;
 &lt;td&gt;Late 2026 to early 2027&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;There is a trap that must be flagged here. &lt;strong&gt;The vertical axis of J.P. Morgan&amp;rsquo;s chart is not the absolute level of price but the &amp;ldquo;year-over-year growth rate (YoY %).&amp;quot;&lt;/strong&gt; The two are entirely different.&lt;/p&gt;
&lt;p&gt;Take an example.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Point in time&lt;/th&gt;
 &lt;th style="text-align: right"&gt;ASP index&lt;/th&gt;
 &lt;th style="text-align: right"&gt;YoY&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;4Q25&lt;/td&gt;
 &lt;td style="text-align: right"&gt;100&lt;/td&gt;
 &lt;td style="text-align: right"&gt;–&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;4Q26&lt;/td&gt;
 &lt;td style="text-align: right"&gt;300&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+200%&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;4Q27&lt;/td&gt;
 &lt;td style="text-align: right"&gt;315&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+5%&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;From 4Q26 to 4Q27 the price index still &lt;strong&gt;rises&lt;/strong&gt;, from 300 to 315. Yet the year-over-year growth rate plunges from +200% to +5%. In other words, J.P. Morgan&amp;rsquo;s 2027 &amp;ldquo;peak-out&amp;rdquo; may not mean prices collapse but that &lt;strong&gt;the pace of increase slows.&lt;/strong&gt; The stock market usually prices in this &lt;strong&gt;direction of the growth rate&lt;/strong&gt; before it prices in &amp;ldquo;record-high profit&amp;rdquo; itself. That is why share prices can cool first, right when profit is at its peak.&lt;/p&gt;
&lt;p&gt;To sum up, Goldman watches &lt;strong&gt;Q (usage) and C (cost)&lt;/strong&gt;, while J.P. Morgan watches &lt;strong&gt;the pace of P (price).&lt;/strong&gt; Because these are different variables, both can be right at the same time.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-total-inference-cost-may-actually-fall"&gt;3. Total Inference Cost May Actually Fall
&lt;/h2&gt;&lt;p&gt;Here a counterintuitive conclusion emerges. Even if usage grows 24x, if cost per token falls 60-70% per year, then &lt;strong&gt;the total inference-cost burden could actually be lower than in 2026.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;It is simple arithmetic. Assuming the same token mix and the same model complexity, total cost in 2030 is the usage index × the cost-per-token index.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;At a 60% annual decline: 24 × 2.6% ≈ &lt;strong&gt;61% of 2026&lt;/strong&gt;&lt;/li&gt;
&lt;li&gt;At a 70% annual decline: 24 × 0.8% ≈ &lt;strong&gt;19% of 2026&lt;/strong&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;img alt="Total inference-cost burden, simple sensitivity — even 24x usage is absorbed by the steep drop in cost per token" class="gallery-image" data-flex-basis="390px" data-flex-grow="162" height="1162" loading="lazy" sizes="(max-width: 767px) calc(100vw - 30px), (max-width: 1023px) 700px, (max-width: 1279px) 950px, 1232px" src="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-total-inference-cost-burden-index.png" srcset="https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-total-inference-cost-burden-index_hu_964b689d7d571adb.png 800w, https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-total-inference-cost-burden-index_hu_bfc3a249a4ddb966.png 1600w, https://koreainvestinsights.com/post/goldman-token-demand-vs-jpm-memory-asp-peakout-korea-semiconductor-2026-05-31/goldman-total-inference-cost-burden-index.png 1891w" width="1891"&gt;
&lt;/p&gt;
&lt;p&gt;&lt;small&gt;This is a heavily simplified sensitivity calculation. It does not reflect token mix, model size, context length, the rise in reasoning tokens, multimodality, redundant processing, latency constraints, or memory-bandwidth bottlenecks. In practice these factors could push cost back up.&lt;/small&gt;&lt;/p&gt;
&lt;p&gt;This chart is the heart of Goldman&amp;rsquo;s logic. &lt;strong&gt;Whoever lowers cost per token ends up making the money.&lt;/strong&gt; More than the rise in usage itself, it is the cost reduction that makes that usage affordable which keeps the industry&amp;rsquo;s cash flow alive. That said, as the caveat notes, if context length or reasoning tokens rise quickly, the real cost curve may fall less than this chart suggests.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="4-so-the-timeline-shifts"&gt;4. So the Timeline Shifts
&lt;/h2&gt;&lt;p&gt;Combine the two calls and the following path is logically consistent.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Period&lt;/th&gt;
 &lt;th&gt;What happens&lt;/th&gt;
 &lt;th&gt;Where it&amp;rsquo;s favorable&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;2026&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Agent usage surges → HBM, server DRAM, eSSD and NAND allocation tightens → DRAM and NAND ASP spike&lt;/td&gt;
 &lt;td&gt;Memory ASP beta works strongly&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;2027&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;A higher base plus some supply additions and long-term agreements (LTAs) slow the YoY ASP growth rate. B2B AI memory stays firm; B2C consumer memory hits price resistance&lt;/td&gt;
 &lt;td&gt;Share prices price in &amp;ldquo;slowing growth rate&amp;rdquo; before &amp;ldquo;peak profit.&amp;rdquo; Segment divergence begins&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;&lt;strong&gt;2028-2030&lt;/strong&gt;&lt;/td&gt;
 &lt;td&gt;Usage keeps rising, but the drop in cost per token absorbs it&lt;/td&gt;
 &lt;td&gt;Alpha rotates from broad memory beta to the &lt;strong&gt;token-cost-reduction stack&lt;/strong&gt;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The core message is one. From 2027, &amp;ldquo;how much further DRAM and NAND prices rise&amp;rdquo; matters less than &lt;strong&gt;&amp;ldquo;which components and systems lower cost per token.&amp;quot;&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="5-example-investment-ideas-observation-points-not-recommendations"&gt;5. Example Investment Ideas (Observation Points, Not Recommendations)
&lt;/h2&gt;&lt;p&gt;Below are not stock recommendations but &lt;strong&gt;examples&lt;/strong&gt; of &amp;ldquo;where to look first&amp;rdquo; along the timeline above. AI and memory stocks have already re-rated quickly, so this is not a call to buy right now but a map for when the conditions are met.&lt;/p&gt;
&lt;h3 id="example-1--2026-memory-asp-beta"&gt;Example 1 — 2026 Memory ASP Beta
&lt;/h3&gt;&lt;p&gt;The phase where HBM, server DRAM and eSSD allocation tightens and DRAM and NAND prices spike. Memory large-caps such as &lt;strong&gt;SK Hynix, Samsung Electronics and Micron&lt;/strong&gt; benefit directly. The counterargument is that once the 2027 slowdown in the growth rate begins, the share price can take multiple compression first, even if profit is high.&lt;/p&gt;
&lt;h3 id="example-2--token-cost-reduction-stack"&gt;Example 2 — Token-Cost-Reduction Stack
&lt;/h3&gt;&lt;p&gt;The real heart of Goldman&amp;rsquo;s logic is not the rise in usage but the &lt;strong&gt;fall in cost per token.&lt;/strong&gt; Customers will pay more for chips and systems that lower their cost per token. &lt;strong&gt;Custom ASICs, AI networking and optical, and HBM leaders&lt;/strong&gt; fit this best. (The interconnect, substrate and power bottlenecks seen in the &lt;a class="link" href="https://koreainvestinsights.com/post/marvell-q1-fy2027-korea-semiconductor-readthrough-2026-05-28/" &gt;Marvell read-through&lt;/a&gt; are this same thread.)&lt;/p&gt;
&lt;h3 id="example-3--essd--enterprise-nand"&gt;Example 3 — eSSD / Enterprise NAND
&lt;/h3&gt;&lt;p&gt;Agent inference demands far more retrieval (RAG), logs, context, KV cache and checkpoint storage than plain training. If it is true that AI servers use roughly 3x the SSD capacity of ordinary servers, then NAND can be reclassified not as a plain commodity-cycle asset but as an &lt;strong&gt;AI storage bottleneck.&lt;/strong&gt; The counterargument is the possibility that price resistance in consumer SSDs dilutes enterprise strength.&lt;/p&gt;
&lt;h3 id="example-4--advanced-packaging--mlcc--fc-bga"&gt;Example 4 — Advanced Packaging / MLCC / FC-BGA
&lt;/h3&gt;&lt;p&gt;If Goldman&amp;rsquo;s 2030 token call is right, the complexity of server and rack architecture keeps rising. It is not only GPUs, ASICs and HBM that grow; demand for substrate area, power stabilization, decoupling capacitors and high-speed signal quality grows alongside. High-spec MLCC and FC-BGA suppliers such as &lt;strong&gt;Samsung Electro-Mechanics&lt;/strong&gt; fall here.&lt;/p&gt;
&lt;h3 id="different-peak-outs-by-segment"&gt;Different Peak-Outs by Segment
&lt;/h3&gt;&lt;p&gt;What matters is not &amp;ldquo;memory as a whole&amp;rdquo; but the differences by segment. J.P. Morgan&amp;rsquo;s peak-out logic does not apply equally to all memory.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Segment&lt;/th&gt;
 &lt;th&gt;Likelihood peak-out applies&lt;/th&gt;
 &lt;th&gt;Tension with long-term demand&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;HBM&lt;/td&gt;
 &lt;td&gt;Low-Medium&lt;/td&gt;
 &lt;td&gt;High (demand keeps reinforcing the bandwidth wall)&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Server DRAM&lt;/td&gt;
 &lt;td&gt;Medium&lt;/td&gt;
 &lt;td&gt;Medium&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;eSSD / Enterprise NAND&lt;/td&gt;
 &lt;td&gt;Medium&lt;/td&gt;
 &lt;td&gt;High (structural demand possible)&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Mobile DRAM&lt;/td&gt;
 &lt;td&gt;High&lt;/td&gt;
 &lt;td&gt;Low (fast consumer price resistance)&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Commodity DRAM/NAND&lt;/td&gt;
 &lt;td&gt;High&lt;/td&gt;
 &lt;td&gt;Low (peak-out logic applies best)&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Common condition: the examples above need more than &amp;ldquo;oversold&amp;rdquo; or &amp;ldquo;because it&amp;rsquo;s AI.&amp;rdquo; A candidate must be somewhere where &lt;strong&gt;actual orders, contract prices and earnings estimates are newly rising, and where bottlenecks and entry barriers are confirmed.&lt;/strong&gt;&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;hr&gt;
&lt;h2 id="6-the-two-most-common-misreadings"&gt;6. The Two Most Common Misreadings
&lt;/h2&gt;&lt;p&gt;In this phase the market is prone to two mistakes pointing in opposite directions.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Misreading 1 — &amp;ldquo;J.P. Morgan says peak-out, so the AI infrastructure cycle is over.&amp;quot;&lt;/strong&gt; No. A peak-out is a slowdown in the &lt;strong&gt;growth rate&lt;/strong&gt; of price, not a collapse in the &lt;strong&gt;level&lt;/strong&gt; of price. And it applies best to commodity memory and less to bottlenecks such as HBM, eSSD, ASIC and packaging.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Misreading 2 — &amp;ldquo;Goldman says 24x tokens, so DRAM and NAND prices keep surging through 2030 too.&amp;quot;&lt;/strong&gt; This too goes too far. Goldman says explosive usage and a &lt;strong&gt;steep drop in cost per token&lt;/strong&gt; at the same time. If cost falls quickly, the YoY growth rate of memory prices can slow even as usage rises.&lt;/p&gt;
&lt;p&gt;The right reading is in between. &lt;strong&gt;A memory super-cycle in 2026, a peak-out in price momentum in 2027, and a long-term expansion of the token-cost-reduction stack in 2028-2030.&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="7-fund-managers-comment"&gt;7. Fund Manager&amp;rsquo;s Comment
&lt;/h2&gt;&lt;p&gt;The two calls are not enemies but two axes of one picture. Goldman speaks to &amp;ldquo;how much, and how cheaply, we come to use AI&amp;rdquo;; J.P. Morgan speaks to &amp;ldquo;how long memory prices keep rising fast in that process.&amp;rdquo; The investor&amp;rsquo;s job is not to pick one of the two but to &lt;strong&gt;not miss the inflection points on the timeline.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;The two most dangerous choices are these.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Throwing out all of AI infrastructure on seeing the single word peak-out&lt;/strong&gt; — the mistake of giving away even the winners with bottlenecks at a bargain price.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Chasing commodity memory to the bitter end on token demand alone&lt;/strong&gt; — the mistake of taking multiple compression head-on in the phase where the price growth rate rolls over.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;The signals to check now, put simply, are these.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;What you watch&lt;/th&gt;
 &lt;th&gt;Goldman (demand) strengthens&lt;/th&gt;
 &lt;th&gt;J.P. Morgan (peak-out) strengthens&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Token usage&lt;/td&gt;
 &lt;td&gt;High growth sustained&lt;/td&gt;
 &lt;td&gt;Growth rate slows&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Cost per token&lt;/td&gt;
 &lt;td&gt;Continues to fall 60%+ a year&lt;/td&gt;
 &lt;td&gt;Decline rate slows, latency cost rises&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;HBM long-term agreements&lt;/td&gt;
 &lt;td&gt;Prices held or raised&lt;/td&gt;
 &lt;td&gt;Renegotiation, volume deferral&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Server DRAM contract price&lt;/td&gt;
 &lt;td&gt;Further increase&lt;/td&gt;
 &lt;td&gt;Slowing growth rate, spot discounts&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Enterprise SSD&lt;/td&gt;
 &lt;td&gt;CSP long-term agreements expand&lt;/td&gt;
 &lt;td&gt;Consumer SSD price resistance spreads in&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;In short, &lt;strong&gt;recognizing memory beta in 2026 while preparing, from 2027, to shift weight toward the bottlenecks that lower cost per token&lt;/strong&gt; is the most reasonable stance available now. It is neither chasing memory to the end nor throwing it all out in fear of the peak-out.&lt;/p&gt;
&lt;p&gt;&lt;small&gt;This piece is an analysis reconstructed from Goldman Sachs&amp;rsquo;s official article (2026-05-05), J.P. Morgan-related charts and secondary summaries, and TrendForce&amp;rsquo;s official outlook. The full text of J.P. Morgan&amp;rsquo;s original report, its quarterly ASP tables, and its segment-level detail could not be verified, and company names are examples illustrating the analytical flow, not investment recommendations. Actual investment decisions and their responsibility rest with the investor.&lt;/small&gt;&lt;/p&gt;</description></item><item><title>ADR Hits a Six-Year Low: Will the Megacap Chip Squeeze Tighten, or Is It Time for the Laggards?</title><link>https://koreainvestinsights.com/post/korea-adr-record-low-megacap-concentration-vs-laggard-rebound-2026-05-30/</link><pubDate>Sat, 30 May 2026 10:30:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/korea-adr-record-low-megacap-concentration-vs-laggard-rebound-2026-05-30/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;This note follows up on &lt;a class="link" href="https://koreainvestinsights.com/post/korea-adr-breadth-narrow-leadership-kospi-kosdaq-2026-05-27/" &gt;Korea ADR at 67: The Index Holds, So Why Are Stocks Weak?&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/korea-foreign-investor-flow-memory-megacap-rotation-2026-05-24/" &gt;Korea Foreign Investor Flow Analysis&lt;/a&gt;, and &lt;a class="link" href="https://koreainvestinsights.com/post/kosdaq-smart-money-return-pearl-abyss-rebound-2026-05-22/" &gt;Smart Money Returns to the KOSDAQ&lt;/a&gt;. Where the earlier pieces looked separately at market breadth (ADR), foreign flows into megacaps, and KOSDAQ liquidity, this one ties all three together to answer the practical question: &lt;strong&gt;&amp;ldquo;Will the concentration tighten, or will the laggards rebound?&amp;quot;&lt;/strong&gt;&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;On May 29, 2026 the KOSPI set a fresh record at &lt;strong&gt;8,476pt (+3.55%)&lt;/strong&gt;, but the same day&amp;rsquo;s ADR fell to &lt;strong&gt;roughly 52%&lt;/strong&gt; — the lowest since March 2020, a six-year trough. A record-high index paired with nine of ten stocks falling is an extraordinarily narrow market.&lt;/li&gt;
&lt;li&gt;The stars of this squeeze are Samsung Electronics and SK Hynix (the &amp;ldquo;megacap chip duo,&amp;rdquo; or what Korean traders call &amp;ldquo;Jeon-Nik&amp;rdquo;). The starting point isn&amp;rsquo;t a simple theme but a &lt;strong&gt;genuine memory earnings cycle&lt;/strong&gt; — yet single-stock leverage ETFs piled a wave of speculative flow on top of it.&lt;/li&gt;
&lt;li&gt;The core conclusion is singular. &lt;strong&gt;Don&amp;rsquo;t try to predict the unwind — confirm it with signals.&lt;/strong&gt; The default right now is &amp;ldquo;concentration persists,&amp;rdquo; and laggards should be added only when ADR, turnover, foreign flows, and earnings estimates all turn together.&lt;/li&gt;
&lt;li&gt;And the most important distinction: &lt;strong&gt;&amp;ldquo;riding the concentration&amp;rdquo; and &amp;ldquo;riding the concentration product (leverage ETFs)&amp;rdquo; are two completely different things.&lt;/strong&gt;&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;h2 id="1-what-is-actually-happening-in-the-korean-market"&gt;1. What Is Actually Happening in the Korean Market
&lt;/h2&gt;&lt;p&gt;Start with the facts. ADR doesn&amp;rsquo;t measure how high the index is — it measures &lt;strong&gt;the breadth inside the market&lt;/strong&gt;.&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"&gt;&lt;code class="language-text" data-lang="text"&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;ADR = Advancing issues / Declining issues × 100
&lt;/span&gt;&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;&lt;/div&gt;&lt;p&gt;When ADR is 100, advancers and decliners are balanced. Well below 100 means &amp;ldquo;regardless of the index, most stocks are falling.&amp;rdquo; The roughly 52% reading on May 29 is a level not seen since the 2020 pandemic crash.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Item&lt;/th&gt;
 &lt;th&gt;As of May 29&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;KOSPI&lt;/td&gt;
 &lt;td&gt;8,476.15pt, +3.55%, all-time high&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;ADR&lt;/td&gt;
 &lt;td&gt;About 52%, a six-year low&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Market breadth&lt;/td&gt;
 &lt;td&gt;A breadth collapse close to &amp;ldquo;nine of ten stocks falling&amp;rdquo;&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The previous note showed ADR at 67; in the interval, the market&amp;rsquo;s internals narrowed further. In other words, the record-high index is &lt;strong&gt;not the result of a broadly strong market, but of a handful of megacaps doing the pulling&lt;/strong&gt;.&lt;/p&gt;
&lt;h3 id="two-flows-that-amplified-the-squeeze"&gt;Two Flows That Amplified the Squeeze
&lt;/h3&gt;&lt;p&gt;Two distinct flows magnified the concentration.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;First, single-stock leverage ETFs.&lt;/strong&gt; On May 27 alone, total turnover in Samsung Electronics and SK Hynix single-stock leverage ETFs was about 10.4 trillion won, against a combined market cap of roughly 4.9 trillion won. Dividing turnover by market cap gives a daily turnover ratio of about 207%. That signals short-term, event-driven flow rather than normal long-term capital. The combined turnover of the four KODEX and TIGER Samsung Electronics/SK Hynix leverage products alone came to about 9.42 trillion won.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Second, the MSCI rebalancing.&lt;/strong&gt; MSCI applied its May review at the May 29 close, and domestic brokers estimated 1.2 to 1.4 trillion won of passive inflows from higher weightings for SK Hynix and Samsung Electronics. Money that simply tracks the index was structurally forced to buy more of the duo.&lt;/p&gt;
&lt;p&gt;In short, today&amp;rsquo;s concentration is the product of &lt;strong&gt;real earnings + passive flows + leverage flows&lt;/strong&gt; all aligning in one direction.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="2-two-interpretations"&gt;2. Two Interpretations
&lt;/h2&gt;&lt;p&gt;There are two broad ways to read this.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Interpretation&lt;/th&gt;
 &lt;th&gt;One-line summary&lt;/th&gt;
 &lt;th&gt;Rationale&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;A. The squeeze tightens further&lt;/td&gt;
 &lt;td&gt;Strong leaders run for a long time&lt;/td&gt;
 &lt;td&gt;Memory earnings cycle, passive and leverage flows, MSCI weight increases&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;B. It&amp;rsquo;s the laggards&amp;rsquo; chance to rebound&lt;/td&gt;
 &lt;td&gt;A market this narrow snaps back&lt;/td&gt;
 &lt;td&gt;ADR at a six-year low; extreme breadth collapses don&amp;rsquo;t last&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;Both are partly right. A is correct in that &amp;ldquo;the strongest earnings momentum still sits with the duo,&amp;rdquo; and B is correct in that &amp;ldquo;extreme ADR lows have statistically been followed by rebounds.&amp;rdquo;&lt;/p&gt;
&lt;p&gt;The real issue is &lt;strong&gt;what to do when both are true&lt;/strong&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-the-answer-dont-predict--confirm"&gt;3. The Answer: Don&amp;rsquo;t Predict — Confirm
&lt;/h2&gt;&lt;p&gt;Bottom line first: &lt;strong&gt;no one can know in advance&lt;/strong&gt; when the concentration unwinds. Strong leaders always run longer than expected. So abandoning the leaders too early on the reasoning that &amp;ldquo;rotation into laggards is coming soon&amp;rdquo; usually costs you.&lt;/p&gt;
&lt;p&gt;Instead, judge the unwind not by &amp;ldquo;has the duo rolled over,&amp;rdquo; but by &lt;strong&gt;&amp;ldquo;has the rest of the market, ex-duo, gained the power to rise on its own?&amp;quot;&lt;/strong&gt; Put plainly, these are the five things to watch:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Ex-duo relative strength&lt;/strong&gt;: Does the KOSPI excluding the duo start to outperform the duo itself?&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Broadening advancers&lt;/strong&gt;: With the index holding, does the number of advancing stocks grow and does ADR recover above 70?&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Broadening foreign flows&lt;/strong&gt;: Do foreign net buys stop clustering in the duo and spread to other names?&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Leverage ETF cooldown&lt;/strong&gt;: Does leverage-ETF turnover cool while the underlying Samsung and SK Hynix shares hold firm?&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Laggard estimate upgrades&lt;/strong&gt;: Are the earnings estimates (EPS) of the neglected names being revised higher?&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;When three or four of these light up together, that&amp;rsquo;s when you can acknowledge a rotation. There&amp;rsquo;s no need to bet ahead of time with only one or two flashing.&lt;/p&gt;
&lt;h3 id="the-most-important-distinction-riding-the-squeeze-vs-riding-the-leverage-etf"&gt;The Most Important Distinction: &amp;ldquo;Riding the Squeeze&amp;rdquo; vs. &amp;ldquo;Riding the Leverage ETF&amp;rdquo;
&lt;/h3&gt;&lt;p&gt;Here&amp;rsquo;s a trap worth flagging. &amp;ldquo;I&amp;rsquo;ll ride the concentration in the duo&amp;rdquo; and &amp;ldquo;I&amp;rsquo;ll buy the duo&amp;rsquo;s leverage ETF&amp;rdquo; are entirely different statements.&lt;/p&gt;
&lt;p&gt;A single-stock leverage ETF is not a stock — it&amp;rsquo;s a &lt;strong&gt;leverage product that resets daily&lt;/strong&gt;. You&amp;rsquo;re buying not SK Hynix itself but &amp;ldquo;twice SK Hynix&amp;rsquo;s daily return,&amp;rdquo; and the longer you hold, the more the two become completely different assets.&lt;/p&gt;
&lt;p&gt;These products carry &lt;strong&gt;negative compounding (volatility decay)&lt;/strong&gt;. Even before listing, Korea&amp;rsquo;s Financial Supervisory Service warned that &amp;ldquo;if a price repeatedly rises and falls, cumulative returns can diverge sharply from the underlying.&amp;rdquo; On the managers&amp;rsquo; own simulations, holding a high-volatility name at 2x leverage for about three months often produces a &lt;strong&gt;loss of typically 8% to 12% on the leverage ETF even when the underlying is flat (0%)&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;The key is this: &lt;strong&gt;even if the squeeze doesn&amp;rsquo;t unwind and the duo merely trades sideways, the leverage ETF loses over time.&lt;/strong&gt; On top of that, with no diversification the daily limit is a wide ±60%, and on June 28 the FSS opened an investigation into possible wash trades and overheating. In other words, it&amp;rsquo;s a product whose exit you can&amp;rsquo;t easily control.&lt;/p&gt;
&lt;p&gt;So if you want to ride the concentration, the right approach is &lt;strong&gt;the underlying (cash) shares, not chasing the leverage ETF&lt;/strong&gt;. Riding negative compounding — a &amp;ldquo;publicly disclosed inefficiency&amp;rdquo; — isn&amp;rsquo;t alpha; it&amp;rsquo;s just leverage.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="4-investment-idea-examples-observation-points-not-recommendations"&gt;4. Investment Idea Examples (Observation Points, Not Recommendations)
&lt;/h2&gt;&lt;p&gt;The below are not stock recommendations but &lt;strong&gt;examples&lt;/strong&gt; showing &amp;ldquo;what conditions make a name a candidate.&amp;rdquo; This isn&amp;rsquo;t a call to buy now, but a map of where to look first when the five confirmation signals above light up.&lt;/p&gt;
&lt;h3 id="example-1--the-next-bottleneck-after-gpuhbm-power-substrates-passives"&gt;Example 1 — The Next Bottleneck After GPU/HBM: Power, Substrates, Passives
&lt;/h3&gt;&lt;p&gt;As AI servers grow more complex, demand rises not only for GPUs and HBM but also for &lt;strong&gt;parts that stabilize power, substrates, and passive components&lt;/strong&gt;. MLCCs, FC-BGA, and high-layer-count boards (MLB) are the prime examples. That said, some names here have already run a long way. One large component stock rose about 2.56x between late April and late May. So the point is &lt;strong&gt;not &amp;ldquo;what looks cheap&amp;rdquo; but &amp;ldquo;what is seeing fresh upward earnings revisions.&amp;quot;&lt;/strong&gt;&lt;/p&gt;
&lt;h3 id="example-2--traces-of-foreigners-pre-positioning"&gt;Example 2 — Traces of Foreigners&amp;rsquo; Pre-Positioning
&lt;/h3&gt;&lt;p&gt;There&amp;rsquo;s one intriguing signal. On May 29, as retail dumped en masse and memory melted up, &lt;strong&gt;foreigners quietly net-bought some deeply beaten-down materials, parts, and equipment names&lt;/strong&gt;. Wonik IPS, Leeno Industrial, and EO Technics were among them. Because domestic institutions mostly sold alongside retail, a &amp;ldquo;foreigners buying, domestic institutions selling&amp;rdquo; divergence emerged. Whether that divergence resolves with domestic institutions returning is the second confirmation point.&lt;/p&gt;
&lt;h3 id="example-3--where-does-the-money-the-leverage-etfs-bleed-go"&gt;Example 3 — Where Does the Money the Leverage ETFs Bleed Go?
&lt;/h3&gt;&lt;p&gt;The most contrarian view is this: when everyone is buying the duo&amp;rsquo;s leverage ETFs, watch &lt;strong&gt;where the money those products bleed via negative compounding will flow&lt;/strong&gt;. When leverage holders bleed in sideways and choppy ranges, that money ultimately has to flow back into more efficient exposure — index products, cash equities, and neglected quality names. This is a bet on direction rather than timing, which makes it relatively safe.&lt;/p&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Common condition: all of the examples above need more than just &amp;ldquo;oversold.&amp;rdquo; A name only becomes a candidate when &lt;strong&gt;earnings aren&amp;rsquo;t impaired, estimates are being revised higher, and turnover and foreign flows are attaching for the first time&lt;/strong&gt; — all at once.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;hr&gt;
&lt;h2 id="5-the-june-calendar-is-the-real-variable-for-rotation"&gt;5. The June Calendar Is the Real Variable for Rotation
&lt;/h2&gt;&lt;p&gt;The biggest enemy of a laggard rebound is, surprisingly, &lt;strong&gt;rates and liquidity&lt;/strong&gt;. June is packed with heavyweight events.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Event&lt;/th&gt;
 &lt;th&gt;Date&lt;/th&gt;
 &lt;th&gt;Why it matters&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;9th nationwide local elections&lt;/td&gt;
 &lt;td&gt;June 3&lt;/td&gt;
 &lt;td&gt;Short-term political uncertainty&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;US May CPI&lt;/td&gt;
 &lt;td&gt;June 10&lt;/td&gt;
 &lt;td&gt;Whether inflation reignites&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;FOMC&lt;/td&gt;
 &lt;td&gt;June 16–17&lt;/td&gt;
 &lt;td&gt;Chair Kevin Warsh&amp;rsquo;s first meeting&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;SpaceX IPO (target)&lt;/td&gt;
 &lt;td&gt;June 12&lt;/td&gt;
 &lt;td&gt;Assumed to be a major liquidity-absorbing event&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;US April PCE inflation came in at +3.8% year over year, so the inflation burden is still not low. If June CPI accelerates further or the new Fed chair is read as hawkish, &lt;strong&gt;rates could rise and rotation into high-valuation laggards (especially high-PER component names) could be delayed&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;And keep one statistical trap in mind. The claim that &amp;ldquo;the win rate of rebounds after extreme ADR lows is nearly 100%&amp;rdquo; omits that the sample is small and that most past cases occurred during &lt;strong&gt;rate-cutting regimes&lt;/strong&gt;. We may instead be in the opposite regime, with inflation stirring again. You can&amp;rsquo;t mechanically transplant past win rates onto the present.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="6-wrap-up--fund-managers-comment"&gt;6. Wrap-Up — Fund Manager&amp;rsquo;s Comment
&lt;/h2&gt;&lt;p&gt;Back to the question: &amp;ldquo;Will the megacap chip squeeze tighten, or is it the laggards&amp;rsquo; chance to rebound?&amp;rdquo; The honest answer is &lt;strong&gt;&amp;ldquo;both are possible, and right now we&amp;rsquo;re in the stage of waiting for the signal that separates them.&amp;quot;&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;The two riskiest choices are:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Abandoning the leaders too early on the back of a low ADR alone&lt;/strong&gt; — handing over strong earnings momentum at a discount.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Chasing an already-overheated single-stock leverage ETF too late&lt;/strong&gt; — being dragged along with no exit by negative compounding and ±60% volatility.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;So the most rational stance right now boils down to this:&lt;/p&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Hold the duo &lt;strong&gt;via the underlying (cash) shares&lt;/strong&gt;, do not chase the leverage ETF, and add laggards &lt;strong&gt;only after confirming that ADR, turnover, foreign flows, and earnings estimates all turn together&lt;/strong&gt;.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;p&gt;Unwinding a concentration isn&amp;rsquo;t something you predict — it&amp;rsquo;s something you confirm. Not missing the moment that first confirmation signal lights up, using the five-point checklist above — that&amp;rsquo;s the most practical preparation you can make right now.&lt;/p&gt;
&lt;p&gt;&lt;small&gt;The market figures in this article are based on news and disclosures between May 27 and 29, 2026, and the named stocks are examples to illustrate the analytical flow, not investment recommendations. All actual investment decisions and their consequences rest with the investor.&lt;/small&gt;&lt;/p&gt;</description></item><item><title>Korea ADR at 67: Why the Index Can Hold While Most Stocks Are Weak</title><link>https://koreainvestinsights.com/post/korea-adr-breadth-narrow-leadership-kospi-kosdaq-2026-05-27/</link><pubDate>Wed, 27 May 2026 16:25:00 +0900</pubDate><guid>https://koreainvestinsights.com/post/korea-adr-breadth-narrow-leadership-kospi-kosdaq-2026-05-27/</guid><description>
 &lt;blockquote&gt;
 &lt;p&gt;This note follows &lt;a class="link" href="https://koreainvestinsights.com/post/macro-snapshot-complex-risk-off-recovery-triggers-2026-05-17/" &gt;the complex risk-off and recovery-trigger framework&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/korea-foreign-ownership-kospi-samsung-hynix-divergence-2026-05-26/" &gt;KOSPI foreign ownership versus Samsung and SK Hynix&lt;/a&gt;, &lt;a class="link" href="https://koreainvestinsights.com/post/korea-foreign-investor-flow-memory-megacap-rotation-2026-05-24/" &gt;Korea foreign-investor flow analysis&lt;/a&gt; and &lt;a class="link" href="https://koreainvestinsights.com/post/national-growth-fund-kosdaq-smart-money-policy-bottlenecks-2026-05-24/" &gt;the National Growth Fund / KOSDAQ smart-money map&lt;/a&gt;. Those pieces looked at macro gates, KOSPI mega-cap flows and KOSDAQ policy capital. This one asks a simpler question: &lt;strong&gt;how healthy is the tape beneath the index?&lt;/strong&gt;&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;h2 id="tldr"&gt;TL;DR
&lt;/h2&gt;&lt;ul&gt;
&lt;li&gt;Korea&amp;rsquo;s 20-day ADR fell from &lt;strong&gt;113.1 to 67.3&lt;/strong&gt; over the last month. KOSPI fell from &lt;strong&gt;116.3 to 68.5&lt;/strong&gt; and KOSDAQ from &lt;strong&gt;111.6 to 66.7&lt;/strong&gt;. The average stock is already in a correction even if the index still holds.&lt;/li&gt;
&lt;li&gt;This is not broad risk-on. It is &lt;strong&gt;narrow leadership&lt;/strong&gt; concentrated in AI infrastructure bottlenecks, MLCC / FC-BGA / SOCAMM / back-end semis, and selected shipbuilding / defense / power names.&lt;/li&gt;
&lt;li&gt;The next trade is not chasing first-line leaders. It is watching for &lt;strong&gt;ADR recovery + rising turnover + first foreign / institutional flow into second-line candidates&lt;/strong&gt;. The local screen points to HPSP, SFA Semicon, Hana Micron, Dongjin Semichem and KMW.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;img alt="Korea equity ADR trend: 20-day ADR and daily advance ratio over the last month" class="gallery-image" data-flex-basis="403px" data-flex-grow="167" height="762" loading="lazy" sizes="(max-width: 767px) calc(100vw - 30px), (max-width: 1023px) 700px, (max-width: 1279px) 950px, 1232px" src="https://koreainvestinsights.com/post/korea-adr-breadth-narrow-leadership-kospi-kosdaq-2026-05-27/korea-adr-breadth-chart-2026-05-26.jpg" srcset="https://koreainvestinsights.com/post/korea-adr-breadth-narrow-leadership-kospi-kosdaq-2026-05-27/korea-adr-breadth-chart-2026-05-26_hu_dc8816563b244f9.jpg 800w, https://koreainvestinsights.com/post/korea-adr-breadth-narrow-leadership-kospi-kosdaq-2026-05-27/korea-adr-breadth-chart-2026-05-26.jpg 1280w" width="1280"&gt;
&lt;/p&gt;
&lt;p&gt;&lt;small&gt;Source: Research OS local DB &lt;code&gt;prices_daily&lt;/code&gt; + &lt;code&gt;universe.csv&lt;/code&gt;. Data cut is the May 26, 2026 close. May 27 intraday data is not included.&lt;/small&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="1-what-adr-tells-us"&gt;1. What ADR Tells Us
&lt;/h2&gt;&lt;p&gt;ADR measures market breadth, not index level.&lt;/p&gt;
&lt;p&gt;The formula is simple.&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"&gt;&lt;code class="language-text" data-lang="text"&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;Daily ADR = advancing stocks / declining stocks × 100
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;20D ADR = sum of advancing stocks over 20 sessions / sum of declining stocks over 20 sessions × 100
&lt;/span&gt;&lt;/span&gt;&lt;span style="display:flex;"&gt;&lt;span&gt;Advance ratio = advancing stocks / total stocks × 100
&lt;/span&gt;&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;&lt;/div&gt;&lt;p&gt;An ADR of 100 means advancing and declining stocks are balanced. Below 80, declining stocks are clearly dominating. If ADR sits in the 60s or 70s while the index remains firm, the market is usually being carried by a few large caps or a narrow set of themes.&lt;/p&gt;
&lt;p&gt;That is Korea now.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Market&lt;/th&gt;
 &lt;th style="text-align: right"&gt;2026-04-16 20D ADR&lt;/th&gt;
 &lt;th style="text-align: right"&gt;2026-05-26 20D ADR&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Change&lt;/th&gt;
 &lt;th&gt;Read&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;KOSPI&lt;/td&gt;
 &lt;td style="text-align: right"&gt;116.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;68.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-47.8p&lt;/td&gt;
 &lt;td&gt;Large-cap breadth has weakened&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;KOSDAQ&lt;/td&gt;
 &lt;td style="text-align: right"&gt;111.6&lt;/td&gt;
 &lt;td style="text-align: right"&gt;66.7&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-44.9p&lt;/td&gt;
 &lt;td&gt;Small/mid-cap breadth has cooled sharply&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;All Korea&lt;/td&gt;
 &lt;td style="text-align: right"&gt;113.1&lt;/td&gt;
 &lt;td style="text-align: right"&gt;67.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-45.8p&lt;/td&gt;
 &lt;td&gt;The average stock is already weak&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The important point is that this is not only a KOSDAQ problem. KOSPI&amp;rsquo;s 20-day ADR is also down to 68.5. So the right read is not &amp;ldquo;KOSDAQ is weak but KOSPI is healthy.&amp;rdquo; It is this:&lt;/p&gt;

 &lt;blockquote&gt;
 &lt;p&gt;Korean market breadth has narrowed, and surviving capital is compressed into AI infrastructure, shipbuilding, defense and a few other leadership pockets.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;hr&gt;
&lt;h2 id="2-why-the-late-may-bounce-is-not-enough-yet"&gt;2. Why the Late-May Bounce Is Not Enough Yet
&lt;/h2&gt;&lt;p&gt;On May 20, only 333 stocks rose while 2,082 fell. Daily ADR was &lt;strong&gt;16.0&lt;/strong&gt; and the advance ratio was &lt;strong&gt;13.5%&lt;/strong&gt;. That was close to a short-term capitulation session.&lt;/p&gt;
&lt;p&gt;The May 21-22 rebound was strong.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Date&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Advancers&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Decliners&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Daily ADR&lt;/th&gt;
 &lt;th style="text-align: right"&gt;20D ADR&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Advance Ratio&lt;/th&gt;
 &lt;th&gt;Read&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;2026-05-20&lt;/td&gt;
 &lt;td style="text-align: right"&gt;333&lt;/td&gt;
 &lt;td style="text-align: right"&gt;2,082&lt;/td&gt;
 &lt;td style="text-align: right"&gt;16.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;58.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;13.5%&lt;/td&gt;
 &lt;td&gt;Short-term capitulation&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;2026-05-21&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1,695&lt;/td&gt;
 &lt;td style="text-align: right"&gt;720&lt;/td&gt;
 &lt;td style="text-align: right"&gt;235.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;61.9&lt;/td&gt;
 &lt;td style="text-align: right"&gt;68.6%&lt;/td&gt;
 &lt;td&gt;Technical rebound&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;2026-05-22&lt;/td&gt;
 &lt;td style="text-align: right"&gt;2,097&lt;/td&gt;
 &lt;td style="text-align: right"&gt;283&lt;/td&gt;
 &lt;td style="text-align: right"&gt;742.2&lt;/td&gt;
 &lt;td style="text-align: right"&gt;67.9&lt;/td&gt;
 &lt;td style="text-align: right"&gt;86.1%&lt;/td&gt;
 &lt;td&gt;Strong rebound&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;2026-05-26&lt;/td&gt;
 &lt;td style="text-align: right"&gt;749&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1,660&lt;/td&gt;
 &lt;td style="text-align: right"&gt;45.1&lt;/td&gt;
 &lt;td style="text-align: right"&gt;67.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;30.2%&lt;/td&gt;
 &lt;td&gt;Breadth weakened again&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The problem is May 26. After the strong rebound, decliners rose again to 1,660 and the advance ratio fell back to 30.2%.&lt;/p&gt;
&lt;p&gt;So the current state is not confirmed breadth expansion. It is a bottoming attempt in breadth that is still being tested.&lt;/p&gt;
&lt;p&gt;That links directly to the earlier &lt;a class="link" href="https://koreainvestinsights.com/post/macro-snapshot-complex-risk-off-recovery-triggers-2026-05-17/" &gt;macro risk-off framework&lt;/a&gt;. The condition there was that oil, long rates, the dollar, KRW, Chinese credit and foreign flows needed to stabilize together. The ADR data says that recovery has not yet spread through the market.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="3-current-regime-narrow-leadership"&gt;3. Current Regime: Narrow Leadership
&lt;/h2&gt;&lt;p&gt;The current Korean equity regime is &lt;strong&gt;Narrow Leadership / Selective Risk-On&lt;/strong&gt;.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Item&lt;/th&gt;
 &lt;th&gt;Read&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Market breadth&lt;/td&gt;
 &lt;td&gt;Weak. All-market 20D ADR is 67.3&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Leadership&lt;/td&gt;
 &lt;td&gt;Very strong in AI infrastructure and selected shipbuilding / defense&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Trading difficulty&lt;/td&gt;
 &lt;td&gt;High. Stock selection matters more than index direction&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;New exposure&lt;/td&gt;
 &lt;td&gt;Prefer pullbacks and second-line names over chasing first-line leaders&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Portfolio stance&lt;/td&gt;
 &lt;td&gt;Hold relative-strength leaders, replace weak positions, avoid using all cash before breadth recovers&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The mistake is to say &amp;ldquo;the index is holding, so the whole market is fine.&amp;rdquo; With ADR in the 60s, most stocks are not fine. The opposite mistake is to say &amp;ldquo;breadth is bad, so avoid everything.&amp;rdquo; That is also wrong, because capital is still very active in a narrow set of sectors.&lt;/p&gt;
&lt;p&gt;The surviving groups are:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;AI infrastructure bottlenecks:&lt;/strong&gt; Samsung Electro-Mechanics, Jeju Semiconductor, Daeduck Electronics, Simmtech, Haesung DS, Hana Micron, HPSP&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Memory mega-caps:&lt;/strong&gt; SK Hynix, Samsung Electronics&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Shipbuilding / defense / nuclear-SMR:&lt;/strong&gt; HD Hyundai Heavy Industries, Hanwha Ocean, Doosan Fuel Cell&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Power / optical / network:&lt;/strong&gt; selected power cable, RF and optical names&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;The market is weak, but leadership is not. That is not a contradiction. When breadth collapses, capital often crowds even harder into the few themes that still work.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="4-where-leadership-actually-was"&gt;4. Where Leadership Actually Was
&lt;/h2&gt;&lt;p&gt;The last-month leadership list is compressed into AI infrastructure and shipbuilding / defense.&lt;/p&gt;
&lt;p&gt;Returns are %, turnover and flows are KRW 100 million units.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Stock&lt;/th&gt;
 &lt;th style="text-align: right"&gt;1M&lt;/th&gt;
 &lt;th style="text-align: right"&gt;5D&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Avg Turnover&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Foreign 1M&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Institution 1M&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Retail 1M&lt;/th&gt;
 &lt;th&gt;Read&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Jeju Semiconductor&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+173.6&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+28.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;3,692.8&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+1,440.8&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+595.8&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-1,962.1&lt;/td&gt;
 &lt;td&gt;LPDDR second-order discovery. Hot&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Samsung Electro-Mechanics&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+146.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+52.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;8,986.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-9,036.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+4,953.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+3,922.2&lt;/td&gt;
 &lt;td&gt;MLCC + FC-BGA leader. Institution-led&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Daeduck Electronics&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+81.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+20.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1,198.1&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+524.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+827.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-954.7&lt;/td&gt;
 &lt;td&gt;FC-BGA / MLB core name&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Simmtech&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+74.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+32.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;822.6&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+678.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+1,376.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-2,070.4&lt;/td&gt;
 &lt;td&gt;SOCAMM / substrate core&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Haesung DS&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+72.7&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+19.2&lt;/td&gt;
 &lt;td style="text-align: right"&gt;297.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+132.7&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+656.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-53.9&lt;/td&gt;
 &lt;td&gt;Heat spreader / substrate option&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Hana Micron&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+48.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+5.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1,200.6&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+2,172.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+158.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+135.8&lt;/td&gt;
 &lt;td&gt;Foreign-led back-end recovery&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;HD Hyundai Heavy Industries&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+51.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+21.1&lt;/td&gt;
 &lt;td style="text-align: right"&gt;3,642.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-5,344.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+7,038.2&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-1,987.9&lt;/td&gt;
 &lt;td&gt;Institution-led shipbuilding + nuclear option&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Hanwha Ocean&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+2.1&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+16.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;2,369.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+1,472.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+1,393.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+6,409.7&lt;/td&gt;
 &lt;td&gt;5D flow improvement, still a laggard&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;The key is that not all leadership has the same flow quality.&lt;/p&gt;
&lt;p&gt;Samsung Electro-Mechanics rose 146.0% over one month, but foreigners sold KRW 903.6 billion. Institutions and retail absorbed the supply. That fits the &lt;a class="link" href="https://koreainvestinsights.com/post/samsung-electro-mechanics-100tn-murata-hyundai-market-cap-2026-05-26/" &gt;SEMCO KRW 100T market-cap note&lt;/a&gt;: the AI passive-component re-rating is real, but chasing efficiency has fallen.&lt;/p&gt;
&lt;p&gt;Hana Micron, by contrast, rose 48.3% with KRW 217.25 billion of one-month foreign net buying. That is why it looks more like a second-line back-end expansion candidate than an already over-owned first-line leader.&lt;/p&gt;
&lt;p&gt;HD Hyundai Heavy Industries was sold by foreigners but bought heavily by institutions. That connects to the &lt;a class="link" href="https://koreainvestinsights.com/post/hd-hyundai-heavy-industries-smr-terrapower-natrium-option-2026-05-27/" &gt;HD Hyundai Heavy Industries SMR option analysis&lt;/a&gt;: the shipbuilding / engine / SMR story is alive, but the rally is institution-led and price location matters.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="5-is-kosdaq-weak-or-selectively-coming-back"&gt;5. Is KOSDAQ Weak, Or Selectively Coming Back?
&lt;/h2&gt;&lt;p&gt;KOSDAQ&amp;rsquo;s 20D ADR is 66.7. On the surface, that is weak. But it does not mean the whole KOSDAQ should be avoided.&lt;/p&gt;
&lt;p&gt;The earlier &lt;a class="link" href="https://koreainvestinsights.com/post/kosdaq-smart-money-return-pearl-abyss-rebound-2026-05-22/" &gt;KOSDAQ smart-money and Pearl Abyss rebound note&lt;/a&gt; argued that flows can turn before prices. The ADR data narrows that frame.&lt;/p&gt;
&lt;p&gt;What matters is not buying KOSDAQ broadly. It is finding &lt;strong&gt;second-line names where turnover is just starting to accelerate, foreign/institutional flows are positive, and the 20-day moving-average extension is still manageable&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;The local screen highlights:&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th style="text-align: right"&gt;Rank&lt;/th&gt;
 &lt;th&gt;Stock&lt;/th&gt;
 &lt;th&gt;Theme&lt;/th&gt;
 &lt;th style="text-align: right"&gt;5D&lt;/th&gt;
 &lt;th style="text-align: right"&gt;20D&lt;/th&gt;
 &lt;th style="text-align: right"&gt;5D Avg Turnover&lt;/th&gt;
 &lt;th style="text-align: right"&gt;Turnover Accel.&lt;/th&gt;
 &lt;th style="text-align: right"&gt;20D MA Gap&lt;/th&gt;
 &lt;th style="text-align: right"&gt;fi5&lt;/th&gt;
 &lt;th&gt;Read&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;1&lt;/td&gt;
 &lt;td&gt;HPSP&lt;/td&gt;
 &lt;td&gt;Semi equipment / AI infra&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+12.7&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+3.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1,989.2&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1.28x&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+2.9&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+509.3&lt;/td&gt;
 &lt;td&gt;Cleanest second-line candidate&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;2&lt;/td&gt;
 &lt;td&gt;SFA Semicon&lt;/td&gt;
 &lt;td&gt;Back-end&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+19.9&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-2.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;3,020.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;3.41x&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+7.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+259.1&lt;/td&gt;
 &lt;td&gt;Back-end expansion candidate&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;3&lt;/td&gt;
 &lt;td&gt;Hana Micron&lt;/td&gt;
 &lt;td&gt;Back-end&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-1.2&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+16.6&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1,386.3&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1.18x&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+3.8&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+129.4&lt;/td&gt;
 &lt;td&gt;Pullback candidate&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;4&lt;/td&gt;
 &lt;td&gt;Dongjin Semichem&lt;/td&gt;
 &lt;td&gt;Materials&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+5.4&lt;/td&gt;
 &lt;td style="text-align: right"&gt;-4.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;642.5&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1.15x&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+3.1&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+447.4&lt;/td&gt;
 &lt;td&gt;Materials-flow recovery candidate&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td style="text-align: right"&gt;5&lt;/td&gt;
 &lt;td&gt;KMW&lt;/td&gt;
 &lt;td&gt;RF / AI-RAN&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+11.7&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+20.8&lt;/td&gt;
 &lt;td style="text-align: right"&gt;259.0&lt;/td&gt;
 &lt;td style="text-align: right"&gt;1.30x&lt;/td&gt;
 &lt;td style="text-align: right"&gt;n/a&lt;/td&gt;
 &lt;td style="text-align: right"&gt;+226.8&lt;/td&gt;
 &lt;td&gt;AI-RAN event-confirmation candidate&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;&lt;code&gt;fi5&lt;/code&gt; is five-day foreign plus institutional net buying. Some local flow fields can be missing or incomplete, so Kiwoom / KRX flow validation is required before stock-level execution.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="6-breadth-expansion-triggers"&gt;6. Breadth Expansion Triggers
&lt;/h2&gt;&lt;p&gt;This is not yet a broad-market buy zone. At an all-market ADR of 67.3, decliners still dominate.&lt;/p&gt;
&lt;p&gt;The confirmation checklist is:&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Trigger&lt;/th&gt;
 &lt;th&gt;Threshold&lt;/th&gt;
 &lt;th&gt;Meaning&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;20D ADR recovers to 80&lt;/td&gt;
 &lt;td&gt;All-market ADR above 80&lt;/td&gt;
 &lt;td&gt;Decliner dominance is easing&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;20D ADR recovers to 100&lt;/td&gt;
 &lt;td&gt;All-market ADR above 100&lt;/td&gt;
 &lt;td&gt;Advancers and decliners are balanced&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Daily advance ratio above 55%&lt;/td&gt;
 &lt;td&gt;2-3 consecutive sessions&lt;/td&gt;
 &lt;td&gt;The rebound is not a one-day bounce&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;KOSDAQ turnover rises&lt;/td&gt;
 &lt;td&gt;Turnover up with more advancers&lt;/td&gt;
 &lt;td&gt;Small/mid-cap breadth can broaden&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Foreign selling is absorbed&lt;/td&gt;
 &lt;td&gt;FX stable and index holds/rises&lt;/td&gt;
 &lt;td&gt;More absorption than true risk-off&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;By sector, the sequence to watch is:&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Sector&lt;/th&gt;
 &lt;th&gt;Signal&lt;/th&gt;
 &lt;th&gt;Meaning&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;AI infra second-line&lt;/td&gt;
 &lt;td&gt;HPSP, SFA Semicon, Hana Micron, Dongjin Semichem turnover rising&lt;/td&gt;
 &lt;td&gt;Internal expansion within semis&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Optical / RF / AI-RAN&lt;/td&gt;
 &lt;td&gt;KMW, RFHIC, Oi Solution flow turning&lt;/td&gt;
 &lt;td&gt;Marvell / NVIDIA AI-RAN linkage&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;FC-BGA / MLB&lt;/td&gt;
 &lt;td&gt;Daeduck, ISU Petasys, Korea Circuit re-accelerating&lt;/td&gt;
 &lt;td&gt;Custom ASIC / AI networking confirmation&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Test sockets / back-end&lt;/td&gt;
 &lt;td&gt;ISC, Leeno, TSE, Doosan Tesna turnover rising&lt;/td&gt;
 &lt;td&gt;SOCAMM / ASIC test-infra expansion&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Shipbuilding / defense second-line&lt;/td&gt;
 &lt;td&gt;Flows rotate into laggards during leader pullbacks&lt;/td&gt;
 &lt;td&gt;Rotation within an existing leadership theme&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;This links back to &lt;a class="link" href="https://koreainvestinsights.com/post/marvell-broadcom-earnings-korea-ai-bottleneck-preview-2026-05-23/" &gt;the Marvell / Broadcom Korea AI bottleneck preview&lt;/a&gt;. If the market is rotating from a single HBM trade into custom ASICs, AI networking, optical links and power integrity, the broad ADR can stay weak while those lower-stack bottlenecks keep attracting turnover.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="7-practical-read"&gt;7. Practical Read
&lt;/h2&gt;&lt;p&gt;The current Korean tape can be summarized in two lines:&lt;/p&gt;

 &lt;blockquote&gt;
 &lt;p&gt;The broad market is weak.&lt;br&gt;
Leadership is not dead.&lt;/p&gt;

 &lt;/blockquote&gt;
&lt;p&gt;The action plan must reflect both.&lt;/p&gt;
&lt;table&gt;
 &lt;thead&gt;
 &lt;tr&gt;
 &lt;th&gt;Action&lt;/th&gt;
 &lt;th&gt;Condition&lt;/th&gt;
 &lt;th&gt;Targets&lt;/th&gt;
 &lt;/tr&gt;
 &lt;/thead&gt;
 &lt;tbody&gt;
 &lt;tr&gt;
 &lt;td&gt;Hold existing leaders&lt;/td&gt;
 &lt;td&gt;Relative strength remains intact despite weak ADR&lt;/td&gt;
 &lt;td&gt;Samsung Electro-Mechanics, Daeduck and other AI infra names&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Avoid chasing first-line leaders&lt;/td&gt;
 &lt;td&gt;20D ADR below 80&lt;/td&gt;
 &lt;td&gt;Hot first-line names&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Watch second-line candidates&lt;/td&gt;
 &lt;td&gt;Turnover acceleration + positive fi5 + manageable MA gap&lt;/td&gt;
 &lt;td&gt;HPSP, SFA Semicon, Hana Micron&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Replace weak positions&lt;/td&gt;
 &lt;td&gt;Holdings underperform the market and flows are weak&lt;/td&gt;
 &lt;td&gt;Non-leadership / non-core positions&lt;/td&gt;
 &lt;/tr&gt;
 &lt;tr&gt;
 &lt;td&gt;Manage cash&lt;/td&gt;
 &lt;td&gt;Portfolio concentration is high before breadth recovers&lt;/td&gt;
 &lt;td&gt;Do not fully deploy cash while breadth is still weak&lt;/td&gt;
 &lt;/tr&gt;
 &lt;/tbody&gt;
&lt;/table&gt;
&lt;p&gt;Chasing first-line leaders is inefficient here. Samsung Electro-Mechanics, Jeju Semiconductor and Simmtech have already moved sharply. But avoiding all equities because breadth is weak would miss the narrow leadership regime.&lt;/p&gt;
&lt;p&gt;The cleaner stance is: &lt;strong&gt;hold leaders, watch second-line names, and require ADR above 80 before broadening exposure.&lt;/strong&gt;&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="8-conclusion"&gt;8. Conclusion
&lt;/h2&gt;&lt;p&gt;Korea is not a dead market. But it is not a broad market either.&lt;/p&gt;
&lt;p&gt;As of the May 26, 2026 close, the all-market 20-day ADR is &lt;strong&gt;67.3&lt;/strong&gt;. KOSPI is &lt;strong&gt;68.5&lt;/strong&gt; and KOSDAQ is &lt;strong&gt;66.7&lt;/strong&gt;. That is a poor backdrop for buying the average stock. At the same time, Samsung Electro-Mechanics, Jeju Semiconductor, Daeduck Electronics, Simmtech, Hana Micron and HD Hyundai Heavy Industries still show strong leadership.&lt;/p&gt;
&lt;p&gt;So the regime is &lt;strong&gt;narrow leadership&lt;/strong&gt;, not broad risk-on.&lt;/p&gt;
&lt;p&gt;The three checks are:&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;Does all-market ADR recover above 80?&lt;/li&gt;
&lt;li&gt;Do KOSDAQ turnover and the number of advancers rise together?&lt;/li&gt;
&lt;li&gt;Do foreign and institutional flows begin to appear in AI-infra second-line names?&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;Until those conditions improve, chasing first-line leaders is less attractive than observing second-line candidates. If breadth recovers, the market can broaden. If it does not, only a few leaders will survive. The right stance is neither optimism nor pessimism. It is to accept that market breadth is narrow and follow where real turnover and quality flow are appearing.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="appendix-evidence-classification"&gt;Appendix. Evidence Classification
&lt;/h2&gt;&lt;h3 id="fact"&gt;[Fact]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;As of May 26, 2026, the all-market 20D ADR was &lt;strong&gt;67.3&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;On April 16, 2026, the all-market 20D ADR was &lt;strong&gt;113.1&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;On May 26, 2026, 749 Korean stocks advanced, 1,660 declined, and daily ADR was &lt;strong&gt;45.1&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;The last-month leadership list includes Samsung Electro-Mechanics, Jeju Semiconductor, Daeduck Electronics, Simmtech, Haesung DS, Hana Micron and HD Hyundai Heavy Industries.&lt;/li&gt;
&lt;li&gt;HPSP, SFA Semicon, Hana Micron and Dongjin Semichem screened relatively well on turnover and flow conditions.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="inference"&gt;[Inference]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;The current market is narrow leadership, not broad risk-on.&lt;/li&gt;
&lt;li&gt;More important than simple foreign buying is whether a sector can hold price and attract turnover despite foreign selling.&lt;/li&gt;
&lt;li&gt;Within AI infrastructure, rotation can move from SOCAMM / LPDDR into FC-BGA / MLB, back-end, optical and RF.&lt;/li&gt;
&lt;li&gt;Second-line turnover and flow turns have better expected value than chasing first-line leaders after large moves.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="speculation"&gt;[Speculation]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;Marvell earnings may strengthen custom ASIC, optical and AI-RAN themes more than SOCAMM alone.&lt;/li&gt;
&lt;li&gt;Broadcom earnings may re-ignite AI networking, FC-BGA / ABF and high-speed MLB names.&lt;/li&gt;
&lt;li&gt;If ADR recovers above 80, the probability of second-line expansion rises.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id="blocked"&gt;[Blocked]
&lt;/h3&gt;&lt;ul&gt;
&lt;li&gt;May 27, 2026 closing ADR is not included in this note.&lt;/li&gt;
&lt;li&gt;Some five-day flow fields may be missing or incomplete in the local DB.&lt;/li&gt;
&lt;li&gt;AI-RAN candidates such as RFHIC and Oi Solution require separate Kiwoom / KRX flow validation before firm stock-level judgment.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;small&gt;Data source: Research OS local DB &lt;code&gt;prices_daily&lt;/code&gt;, &lt;code&gt;universe.csv&lt;/code&gt;, &lt;code&gt;korea_adr_recent_20260526.csv&lt;/code&gt;, &lt;code&gt;korea_leaders_20260415_20260526.csv&lt;/code&gt;, &lt;code&gt;second_line_theme_flow_candidates_20260527.csv&lt;/code&gt;. Data cut: May 26, 2026 close. This is research commentary, not investment advice.&lt;/small&gt;&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Disclaimer: For research and information purposes only. Not investment advice. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.&lt;/em&gt;&lt;/p&gt;</description></item></channel></rss>