Pearl Abyss is not priced like Crimson Desert failed. It is priced like the market accepts one successful game but has not yet accepted Pearl Abyss as a repeatable AAA platform company. The closest comps are Neowiz after Lies of P and CD Projekt around the Cyberpunk 2077 recovery and Phantom Liberty. Capcom is not the current comp; it is the target model.
Marvell and Broadcom's upcoming earnings calls are a test of whether Korea's semiconductor trade should move beyond a single HBM bet into AI ASICs, Ethernet networking, optical interconnects, packaging, silicon capacitors, high-speed substrates and test sockets. Broadcom's key benchmark is Q2 AI semiconductor guidance of $10.7B. Marvell's benchmark is Q1 FY27 revenue guidance of $2.4B ±5% and a data-center mix near 74%.
ARM's rally is not just a CPU renaissance story. It is a signal that AI infrastructure bottlenecks are moving from GPUs to CPU orchestration, memory movement, optical interconnects, power integrity, high-speed substrates and test sockets. ARM's thesis is real, but the stock already prices in a large part of the FY31 success case. The better risk-reward may sit in Marvell, Samsung Electro-Mechanics, Korean memory, high-speed PCB/substrates and test sockets.
KOSPI 7,848 in dual-Bull regime. VM (089970), SamC&S (252990) and TLB (356860) each clear all five quality-flow screeners as KOSDAQ surges ~4% on broad breadth expansion.
The past two months were defined by a decisive KOSPI large-cap win and an extreme relative slump in KOSDAQ. Based on a market-cap-weighted proxy, KOSPI gained +49.8% while KOSDAQ rose only +11.4%, leaving KOSDAQ -38.4 percentage points behind. But the recent 5- to 20-trading-day flow picture is different. KOSPI prices remain strong, yet foreigners plus institutions and program trading continue to sell. KOSDAQ prices remain weak, but foreigners plus institutions, program flows, and investment trusts are improving. This is not yet proof of KOSDAQ leadership. It is a bottom-test phase. Within that setup, Pearl Abyss ranks 7th among the top 100 KOSDAQ names by drawdown from its two-month high, trades at 8.0x 2026E PER, and has seen +KRW 22.08bn of foreign-plus-institutional net buying over the past five sessions.
An MLCC is an ultra-compact ceramic component used in almost every electronic device to stabilize power. A silicon capacitor is a high-performance component placed inside or immediately adjacent to AI GPU and HBM packages to suppress instantaneous power fluctuations. The critical insight is not that silicon capacitors replace MLCCs, but that the battlefield for passive components is expanding from the PCB surface into the interior of AI semiconductor packages. Samsung Electro-Mechanics' ₩1.5 trillion supply contract signals that MLCC and substrate companies may be re-classified as AI package power-integrity supply chain players. That said, the three core MLCC names surged an average of +35.6% this week and Samsung Electro-Mechanics now trades at 73x 2026E PER. A compelling industry shift and a compelling new entry price are two different things.
KOSPI jumped over 8% on May 21 as Samsung Electronics labor risk eased, Nvidia beat estimates, and Korea's semiconductor exports surged. Here's what moved markets.
The core of Google I/O 2026 was the reengineering of Google's entire product portfolio into an agentic OS through Gemini 3.5, Gemini Omni, Spark, Antigravity, AI Search, and Universal Cart. Unlike the simplistic narrative of AI killing search, Alphabet's Q1 2026 Search & other revenue grew 19%. The real change is AI usage volume: Google disclosed monthly token processing expanded from 9.7 trillion in May 2024 to over 32 quadrillion in May 2026. The Korean alpha from this trend lies not in chasing Alphabet, but in the bottlenecks of HBM, packaging, silicon capacitors, high-layer PCBs, and data center power.