The U.S. Market Is On Fire, But One Surprising Country Has Outperformed U.S. Stocks 10-to-1 Since the Start of 2025

iShares MSCI South Korea ETF (EWY) has surged 303.3% in U.S. dollar terms since the start of 2025, versus 29.9% for SPY, highlighting a sharp rotation in global equity leadership. The rally is being driven by AI-memory demand at Samsung Electronics and SK Hynix, with Korea benefiting from the HBM supercycle and a weaker dollar; JPMorgan says international equities are up 31% in dollar terms, outperforming U.S. equities by 1,520 bps. Risks remain tied to memory pricing, FX, and geopolitics, but the broad international breadth argues for continued diversification away from U.S. exceptionalism.

Analysis

The biggest second-order effect is not simply “Korea up,” but that the AI memory complex is becoming a quasi-macro asset class with its own cycle, distinct from U.S. semis. If HBM pricing stays tight through the second half, the earnings revision cycle should continue to favor the memory suppliers and the equipment vendors that can expand advanced packaging and test capacity; if not, the market is likely to de-rate the entire basket faster than fundamentals would imply because positioning is now crowded and reflexive.

The FX component is doing a lot of work here. A weaker dollar mechanically amplifies foreign equity returns for U.S. investors, so part of the outperformance is not stock selection but currency beta; that means a USD bounce would hit the ETF wrapper even if local earnings remain fine. That creates an asymmetry: the trade is better expressed in select operating names tied to the HBM bottleneck than in the broad Korea basket, which is heavily concentrated and vulnerable to sudden geopolitical gaps.

The contrarian miss is that the market may be extrapolating AI demand too linearly while underestimating supply response. Micron, Samsung, and SK Hynix all have clear incentives to add capacity, and once utilization normalizes, memory usually mean-reverts violently. The right risk window is 3-9 months, not 3-5 years: the near-term catalyst is another round of guidance raises, but the main reversal risks are a USD rally, a pause in hyperscaler capex, or a policy shock on the peninsula.

For U.S. equities, this is a warning that leadership breadth is rotating away from the old mega-cap growth monopoly. That does not automatically mean SPY should be shorted, but it does argue for relative-value positioning into earnings season: the market is rewarding scarcity of supply in a few AI-critical components more than downstream AI beneficiaries. The opportunity is to own the bottleneck and fade the beneficiaries whose margins depend on those inputs staying cheap.

Key Decisions for Investors

• Long MU vs short a basket of AI downstream beneficiaries (or NVDA delta-neutral) for 3-6 months: express the view that memory pricing, not AI narrative alone, is the cleaner earnings driver; target 1.5-2.0x upside if HBM margins remain firm, with risk capped if capex peaking starts to show.
• Buy call spreads on EWY only as a tactical 1-3 month trade, not a strategic allocation: use limited premium because the ETF is heavily concentrated and vulnerable to FX/geopolitical reversals; attractive only if you want convexity around continued memory strength.
• Favor SK Hynix/Samsung exposure via local listings or ADR proxies over broad Korea exposure: the risk/reward is better because you avoid non-memory sectors diluting the AI thesis; watch for any USD rebound as the primary de-rating trigger.
• Hedge existing U.S. large-cap growth exposure with a small short in SPY or a relative short in U.S. mega-cap tech against long Asian memory exposure: the trade benefits if global leadership continues to broaden and U.S. exceptionalism remains in question over the next quarter.
• Set a 4-8 week trigger to reduce Korea beta if the USD index turns up or if hyperscaler capex commentary softens: those are the fastest ways to unwind the trade, and the ETF will likely gap lower before fundamentals fully roll over.