TL;DR
- The number: KOSPI closed at 7,384.56 on May 6 — the first-ever close above 7,000, +6.45% on the day. Foreigners bought a record ₩3.1 trillion ($2.1B) in a single session.
- The engine: Samsung’s Q1 operating profit hit ₩57.2 trillion ($39.3B), +750% YoY. Semiconductors generated 94% of that. HBM demand is the sole driver.
- The binary: Trump meets Xi in Beijing on May 14–15. Semiconductor export controls, rare earths, and tariffs are all on the table. The outcome defines the next leg.
The Numbers That Actually Matter
Markets run on narratives, but I trade on numbers. Here is what May 6 actually looked like:
- KOSPI intraday high: 7,426.60 (seventh sidecar suspension of the year triggered)
- Samsung 1-day move: +14.4% — the stock has quadrupled in one year and doubled in 2026 alone
- Foreign net purchase: ₩3.1 trillion — the largest single-day foreign buy in Korean market history
- KOSPI YTD: +78%, following a +76% year in 2025
For context, the Nasdaq — the best-performing major US index in 2026 — is up 13% year-to-date. The S&P 500 is up 8%. KOSPI is outperforming the Nasdaq by a factor of six.
This is not a correlation trade. This is a structural re-rating.
Samsung’s $1 Trillion Is Backed by Real Earnings
Let me state this plainly: Samsung’s valuation is not a bubble.
The company reported Q1 2026 operating profit of ₩57.2 trillion ($39.3B), a 750% year-over-year increase. Semiconductor operations accounted for 94% of that figure. The driver is HBM — High Bandwidth Memory — the stacked DRAM architecture that every major AI GPU requires.
Nvidia’s Blackwell B200 GPU requires 2–3x more HBM per unit than its predecessor H100. As hyperscalers (Microsoft, Google, Meta, Amazon) accelerate AI data center buildout well ahead of initial 2026 estimates, HBM demand has outstripped supply. Samsung and SK Hynix together hold an effective duopoly on HBM production capacity.
To put the earnings surge in historical context: Samsung’s ₩7.5 trillion Q1 operating profit in 2024 was already considered a recovery. ₩57.2 trillion in Q1 2026 is in a different category entirely. This is what a commodity supercycle looks like when the commodity is AI memory.
Samsung is now the second Asian company — after TSMC — to surpass a $1 trillion market cap. The gap to Apple, Nvidia, and Microsoft remains wide, but the trajectory validates the thesis that Korean memory is now a core infrastructure asset, not a cyclical commodity bet.
Why Foreign Capital Is Flooding Korea
The ₩3.1 trillion single-day foreign purchase requires explanation beyond “investors like Samsung.”
The structural driver is passive fund rebalancing. MSCI periodically adjusts country weights in its global indices based on market cap and liquidity. As Samsung’s market cap crossed $1 trillion and the KOSPI surged past 7,000, the mechanical rebalancing required global passive funds to increase Korean exposure. This is not discretionary — it is forced buying by index-tracking funds managing trillions.
Korea’s government “Value-Up” program — launched in 2025 to push low-P/B companies toward higher dividend payouts and buybacks — provided a second structural tailwind. This reform, modeled loosely on Japan’s TSE governance push, has been driving a P/B re-rating across Korean industrials and financials.
The result: KOSPI is up 78% YTD in 2026, after +76% in 2025 — the best two-year run since 1999. The S&P 500’s six consecutive winning weeks and the Nasdaq’s 13% YTD gain are impressive. But they are running at one-sixth the pace of KOSPI.
The Trump-Xi Summit Changes Everything (Or Nothing)
The rally’s next direction is a binary event: the Beijing summit on May 14–15.
The agenda is crowded:
- AI semiconductor export controls — Can the US loosen restrictions on high-performance AI chip exports to China? This directly impacts Samsung and SK Hynix’s ability to expand HBM sales into the Chinese market.
- Rare earth supply — China restricted rare earth exports as a retaliatory measure. A resumption of supply would remove a key input cost risk for the semiconductor supply chain.
- Tariff framework — Existing US tariffs on Chinese goods reach 54% on some categories. A bilateral “Board of Trade” framework is reportedly under discussion for non-sensitive sectors.
- Iran — The ongoing Iran conflict is expected to dominate significant summit time. Treasury Secretary Bessent has confirmed Iran will be a primary topic. This creates a risk: Iran diplomacy crowding out trade resolution.
The market is pricing in a partial deal. Polymarket shows 50.5% odds of “no US-China tariff agreement by May 31.” Investor positioning in US-listed Chinese ETFs has turned bullish, with notable call-spread activity in iShares China Large-Cap ETF ahead of the summit.
What a Deal Does (And Doesn’t) Fix for Semiconductors
A common misconception is that trade progress on tariffs directly benefits Korean semiconductor exports. The mechanism is more specific.
Korean memory — particularly HBM — is not subject to the same export restrictions as leading-edge logic chips (e.g., Nvidia’s H100/B200). Samsung and SK Hynix already sell substantial volumes into China. What the summit matters for is:
- Rare earth inputs: China controls ~90% of global rare earth processing, including materials critical for semiconductor manufacturing. A supply resumption reduces input cost pressure.
- Downstream demand: If Chinese AI buildout accelerates due to reduced export restrictions on Nvidia chips, HBM demand follows proportionally.
- Macro risk-off: A failed summit triggers tariff escalation, risk-off sentiment, and pressure on all Asian tech equities — including Korea.
The AI chip supply chain is deeply interconnected. Even if Korean memory is not the direct subject of negotiation, the summit outcome shapes the demand environment that makes HBM profitable.
My Verdict / Positioning
On KOSPI: I remain constructive. The structural case — HBM duopoly, Value-Up reform, passive fund rebalancing — is intact. The earnings cycle is delivering. The risk is concentration: KOSPI’s performance is now highly correlated to Samsung, and Samsung’s performance is correlated to Nvidia B200 demand. This is a high-beta position on global AI capex.
On the summit: I do not recommend speculative positioning ahead of a known binary event. The market has already partially priced in optimism. A full deal outcome likely produces a 3–5% relief rally in KOSPI and Korean semiconductor names. A failure or Iran-dominated meeting that produces no trade progress could retrace 5–8%.
Tactical positioning I am running:
- Long KOSPI via Korean semiconductor ETF (hedged at current levels against summit tail risk)
- Long Nasdaq / TQQQ (AI cycle remains intact regardless of bilateral politics)
- Monitoring: Samsung HBM4 volume ramp timeline — any forward guidance revision is the single most important data point for the next quarter
What I am watching specifically: The joint statement language. If “semiconductor” appears in the context of reduced restrictions, cover the hedge and extend. If it does not, or if the statement focuses solely on agriculture and Boeing purchases, reduce KOSPI exposure and wait.
The Bottom Line
KOSPI 7,000 is not a bubble. It is the market’s recognition that South Korea holds a structural advantage in the most capital-intensive phase of the AI build cycle. Samsung’s $1 trillion market cap reflects ₩57.2 trillion in Q1 earnings — real profit from a real supercycle.
The summit in Beijing on May 14–15 is this week’s event risk. The base case is a partial deal announcement — not enough to remove all uncertainty, but enough to extend the rally through Q2. I stay long, but with a tighter stop than I would normally run into an unresolved binary.
The data says Korea is the trade. The question is how much headline risk you can stomach while holding it.
This content is for informational purposes only and does not constitute investment advice or a solicitation to buy or sell any securities. All investments involve risk, including the possible loss of principal. Past performance is not indicative of future results.