Context This is a bridge post linking the Samsung stock-bonus / memory-supercycle note, the AI infrastructure multiple map, and the NVIDIA Vera Rubin LPX / CMX inference-stack note. The related hub is the AI HBM Hub. The next linked note is SK Hynix vs Micron: HBM Technology Premium or U.S. Listing Premium?.
TL;DR
SK Hynix trading at a higher forward P/E than Samsung Electronics is unusual by historical standards. Memory pure-plays usually carried a discount to Samsung because of cyclicality, customer concentration and weaker business diversification.
But the current move is not simply a valuation error. It reflects a collision between SK Hynix’s HBM bottleneck premium and Samsung’s HBM laggard discount. SK Hynix is being reclassified as an AI-memory bottleneck supplier, while Samsung’s HBM4E catch-up and broad DRAM/NAND price leverage may still be underpriced.
The cleaner alpha is not chasing SK Hynix. It is the Samsung Electronics catch-up trade, conditional on HBM4E customer qualification, 2027 EPS revisions and DS margin durability.
1. Data Frame: Annual P/E Inverted, 12MF P/E Near Parity
A precise monthly three-year 12-month-forward P/E history requires Bloomberg, FactSet, LSEG or FnGuide terminal data. Public sources are enough to frame the current signal, but not to rebuild the full time series.
| Basis | Current Reading |
|---|---|
| FnGuide 2026E P/E | Samsung 6.77x, SK Hynix 6.79x, a slight Hynix inversion |
| 2026-05-29 close and 2026E EPS proxy | Samsung around 7.55x, SK Hynix around 8.02x, a roughly 6.2% Hynix premium |
| Mirae Asset 2026-05-22 12MF P/E | Samsung 6.2x, SK Hynix 5.9x, meaning 12MF parity is approaching but not cleanly inverted |
The conservative sentence is:
Annual 2026E P/E has inverted; 12-month-forward P/E is approaching parity.
That is different from saying Hynix will permanently deserve a higher multiple than Samsung. It says the market is beginning to treat SK Hynix as a scarce AI HBM bottleneck supplier.
2. Why It Is Unusual
Historically, Samsung deserved the higher multiple for four reasons.
- It had memory plus mobile, displays, consumer electronics, foundry and system LSI.
- Its balance sheet and cash-flow stability were stronger.
- SK Hynix was closer to a pure memory-cycle stock.
- Memory stocks were usually discounted for oversupply and ASP downside risk.
So “SK Hynix P/E > Samsung P/E” looks strange in the old framework. HBM changes the framework.
HBM is not commodity DRAM. It is a customer-qualified, package-integrated memory product that can gate GPU and AI ASIC shipments. Once qualified, it is hard for a customer to switch suppliers quickly. That gives SK Hynix a credible argument for a higher-quality earnings stream than a traditional DRAM cycle.
3. Why the SK Hynix Premium Makes Sense
SK Hynix’s 1Q26 numbers look nothing like a normal commodity memory company. It reported KRW 52.6T of revenue, KRW 37.6T of operating profit and a 72% operating margin. (SK hynix Newsroom)
That margin reflects more than DRAM pricing. It reflects a mix shift toward HBM and high-value server memory, plus the fact that qualified HBM supply can gate AI accelerator output.
The Hynix premium is justified if these conditions hold:
| Condition | Meaning |
|---|---|
| HBM share stays high | AI accelerator memory bottleneck remains intact |
| HBM4 / HBM4E leadership continues | Next-generation customer lock-in holds |
| LTAs expand | Peak-earnings doubt fades |
| High ROE persists | P/E and P/B premiums are easier to defend |
| Samsung catch-up is delayed | Relative premium remains intact |
The market is effectively asking whether HBM has moved from commodity memory into a contracted high-value memory franchise.
4. Why the Alpha Is Still Samsung Catch-Up
The problem is price. SK Hynix has already started to receive the HBM-leader premium. Samsung still carries an HBM laggard discount.
Samsung’s catch-up trade has three pillars.
4.1 HBM4E 12-high samples
Samsung reportedly shipped HBM4E 12-high samples to major global customers in May 2026. This is not yet revenue certainty. Customer qualification and production yield still matter. But the valuation path is clear. (Reuters)
HBM4E sample
→ customer qualification
→ mass production
→ EPS revision
→ HBM laggard discount compression
The market may not be fully pricing that path yet.
4.2 DRAM and NAND pricing leverage
This memory cycle is broader than HBM. TrendForce reported that Samsung’s memory ASP was up 146% versus its 2025 average in 1Q26, while SK Hynix saw DRAM ASP up in the mid-60% QoQ range and NAND ASP up in the mid-70% QoQ range. (TrendForce)
This matters for Samsung because its memory portfolio is broader: DRAM, NAND, HBM, eSSD and SOCAMM. A broad memory squeeze can drive wider EPS revisions at Samsung than a pure HBM thesis alone suggests.
4.3 DS profit scale and foundry optionality
Samsung’s DS division generated KRW 81.7T of revenue and KRW 53.7T of operating profit in 1Q26. The absolute profit pool is already enormous. (Samsung Global Newsroom)
If HBM4E uses logic base dies and advanced packaging, Samsung has more than a memory-stack story. It has a memory plus base-die plus packaging optionality that is not as cleanly priced as SK Hynix’s pure HBM leadership.
5. Relative-Value Strategy: Long Samsung / Neutral Hynix
This is not an anti-Hynix call. It is a statement that Samsung catch-up now offers the better relative-value setup.
| SK Hynix P/E Premium vs Samsung | Interpretation | Strategy |
|---|---|---|
| 0-5% | Acceptable for HBM leadership | Own both |
| 5-10% | Samsung catch-up becomes more attractive | Add Samsung weight |
| 10-15% | Hynix premium may be stretched | Samsung long / Hynix neutral |
| 15%+ with slowing EPS revisions | Mean reversion risk rises | Prefer Samsung |
Using the 2026-05-29 close and the 2026E EPS proxy from the user-provided data, the SK Hynix premium is around 6.2%. That already sits in the Samsung catch-up zone.
6. Samsung Upside Bridge
Using a 2026E Samsung EPS proxy of KRW 41,972 and a 2026-05-29 close of KRW 317,000, the simple P/E is about 7.55x.
| EPS Revision | Revised EPS | P/E at KRW 317,000 |
|---|---|---|
| 0% | KRW 41,972 | 7.55x |
| +10% | KRW 46,169 | 6.87x |
| +15% | KRW 48,268 | 6.57x |
| +20% | KRW 50,366 | 6.29x |
If Samsung EPS rises 15% and the market assigns an 8.0x P/E closer to SK Hynix:
KRW 48,268 × 8.0 = KRW 386,144
That is about 21.8% above KRW 317,000. This is not a formal target price. It is a sensitivity showing how powerful EPS revision plus P/E parity can be.
7. What Breaks the Thesis
| Risk | Meaning |
|---|---|
| Samsung HBM4E qualification delay | HBM laggard discount stays |
| Weak HBM4E yield | EPS revisions do not arrive |
| DRAM/NAND pricing cools | Samsung portfolio-breadth thesis weakens |
| Foundry / LSI losses widen | Non-memory drag offsets memory upside |
| SK Hynix 2027E EPS keeps rising | HBM leader premium remains justified |
| AI capex slows | Peak-earnings doubt returns |
Samsung HBM4E sample shipment is a fact. Customer qualification and mass-production economics remain blocked items.
Final View
SK Hynix trading above Samsung on forward P/E is unusual, but not irrational. HBM has moved closer to an AI infrastructure bottleneck than a commodity DRAM cycle, and SK Hynix deserves credit for leadership.
But the better alpha now is Samsung catch-up:
- SK Hynix is already receiving more HBM-leader premium.
- Samsung’s HBM4E, DRAM/NAND ASP leverage and DS profit scale may still be under-reflected.
- As the P/E spread widens above 5-10%, Samsung becomes increasingly attractive on relative value.
The resulting stance is Samsung Electronics: conditional Buy / SK Hynix: Hold-Wait. The next evidence points are HBM4E qualification, 2027 EPS revisions, DS margins and SK Hynix HBM4 / HBM4E share defense.
Evidence Map
[Fact]
- FnGuide 2026E P/E was reported at Samsung 6.77x and SK Hynix 6.79x. (Seoul Economic Daily)
- Samsung reportedly shipped 12-high HBM4E samples to major global customers. (Reuters)
- 1Q26 memory pricing surged across Samsung memory and SK Hynix DRAM/NAND. (TrendForce)
- Samsung DS reported KRW 81.7T revenue and KRW 53.7T operating profit in 1Q26. (Samsung Global Newsroom)
- SK Hynix reported KRW 52.6T revenue, KRW 37.6T operating profit and 72% operating margin in 1Q26. (SK hynix Newsroom)
[Inference]
- SK Hynix is being reclassified from commodity memory pure-play to AI HBM bottleneck supplier.
- Samsung can recover P/E parity if its HBM laggard discount compresses.
- The better relative-value alpha is Samsung catch-up rather than Hynix chase.
[Blocked]
- Exact current monthly 12MF P/E time series.
- Samsung HBM4E customer qualification, yield, volume and ASP.
- 2027 HBM LTA downside-price protection.
Sources
Disclaimer: For research and information purposes only. Not investment advice.