The Ex-China ETF Lie: You're Not Buying Diversification, You're Buying Chips

The consensus take on emerging markets this year is tidy, seductive, and in my opinion, misleading. The story goes like this: China is structurally broken — property in freefall, consumer confidence negative for over four years, a K-shaped economy that rewards AI factories while everything else stagnates. So the rational move is to buy an "emerging markets ex-China" ETF. You get the growth of the developing world without the Chinese drag. You've done your geopolitical homework. You're diversified.

That's the false narrative. It sounds responsible. It's not.

If you bought an ex-China EM ETF in the first half of 2026, you didn't buy diversification. You bought a concentrated levered bet on Taiwanese and South Korean semiconductors, disguised as an emerging market allocation. And the market has been paying you handsomely to play along.

What the numbers actually show

The iShares MSCI Emerging Markets ex China ETF (EMXC) — the category leader with $24.6 billion in assets — is 28% Taiwan and 22% South Korea. Combined, those two countries account for roughly half the fund. Its top holdings are Taiwan Semiconductor Manufacturing Company, Samsung Electronics, and SK Hynix. Three chip companies.

Meanwhile, EMXC delivered approximately 34% year-to-date through mid-2026. The broad iShares Core MSCI Emerging Markets ETF (IEMG), which retains an 18% China weight, returned about 18%. Vanguard's VWO, with 27% China exposure, managed only 9%.

The performance gap exists because Taiwan and South Korea are absorbing close to 30% of global AI capital expenditure. That's not an emerging market diversification story. That's a semiconductor supply-chain story that just happens to live inside an ETF with a politically correct name.

The Global X version is worse

The Global X Emerging Markets ex-China ETF (EMM) makes the category's structural problem look almost generous. Despite "ex-China" in its title, EMM holds 20.5% of its portfolio in Chinese companies — Alibaba, Tencent, PDD Holdings — all accessible through ADRs trading on U.S. exchanges. The fund charges 0.75% annually, nearly double the category average of 0.50%, and manages just $24.6 million in assets. It has a beta of 4.15, the highest in its category, and its top 15 holdings account for over half the fund.

This is a fund that cannot execute its own premise, charges a premium fee to do so, and has less total assets than most individual 401(k) accounts. It is, in my opinion, a structural embarrassment. If you wanted China exposure, you'd buy a broad EM fund at a lower cost. If you didn't, you'd buy a properly constructed ex-China fund. EMM delivers neither.

What the buyers think they're getting

Emerging market ETFs attracted $38 billion in net new assets during the first half of 2026, outpacing the full-year 2025 haul of $35 billion. Roughly 73% of EM ETF funds saw inflows. The capital migration was led by the ex-China narrative, and that narrative is not entirely wrong — just not what most investors think it is.

China's economy is genuinely unbalanced. GDP growth is forecast to moderate to roughly 4.3% in 2026. Household consumption is weak, the property market remains a drag, and the government has shifted from direct stimulus to optimizing existing trade-in subsidies. But China still holds 15% of global export market share and posted real export growth of roughly 8% in 2025. Its AI infrastructure capex is projected to exceed $70 billion in 2026 — 15–20% of U.S. hyperscaler spending.

So China isn't a one-note disaster. It's a complex story where AI-related sectors accelerate while the old economy stagnates. But most ex-China buyers aren't holding a nuanced view. They're holding a headline.

The structural reality

Here's what actually underpins the ex-China performance in 2026, decomposed:

1. AI cross-pollination, not diversification. The hyperscalers' infrastructure spending flows through Taiwan's advanced-node manufacturing (TSMC) and South Korea's high-bandwidth memory (SK Hynix, Samsung). These are not independent emerging market growth stories. They are nodes in a single supply chain. If AI capex slows, the ex-China ETF slows with it — faster than broad EM, because it has no China to cushion the hit.

2. Tariff geography, not decoupling. Trump's 2025 tariffs accelerated supply chain rerouting, but the mechanism wasn't a clean break from China. Multinational firms redirected smartphone, laptop, and monitor sourcing to India and Vietnam, where those products received tariff exemptions. Chinese-owned factories in Vietnam continued exporting to the U.S. The decoupling is real in headline terms but opaque in practice.

3. Valuation discipline is still on EM's side. The broad MSCI Emerging Markets index trades at a trailing P/E of roughly 19.8, compared to 30.5 for the S&P 500. That's a discount of more than 33%, above the historical norm of 20–30%. EM consensus earnings growth is projected at approximately 17% for 2026, outpacing major developed market regions. Aggregate EM economic growth is expected near 4%.

4. The Fed question is unresolved. Goldman Sachs and J.P. Morgan expect the Fed to remain on hold through mid-2026, with market pricing for a year-end rate hike sitting at roughly 50/50. Traditionally, EM assets need Fed easing to thrive. But the structural shift in what EM actually is — less commodity-exporting debtor, more AI supply-chain participant — makes it less sensitive to the dollar cycle than it was five years ago. That's a real change, but it hasn't been stress-tested against a stronger dollar environment.

The allocation question

The real decision for investors isn't whether emerging markets are a good idea. They're cheap, they're growing faster than developed markets, and their earnings trajectory is solid. The question is whether the ex-China wrapper is worth the concentrated bet it smuggles inside.

If you want AI semiconductor exposure, buy it explicitly. There are semiconductor ETFs that are more transparent about what you're holding, charge less, and don't pretend to be something they're not.

If you want genuine emerging market diversification — India's demographic dividend, Indonesia's commodity processing, Brazil's agricultural strength, Mexico's nearshoring potential — the broad EM index may actually serve you better. Yes, it drags some China exposure. But that China exposure also includes exporters, industrial companies, and infrastructure plays that the ex-China funds exclude.

The ex-China ETF has become a single-sector fund wearing a macro costume. That's not a sin if you understand what you're buying. It is a sin if you think you're doing geopolitical risk management.

For investors who want EM exposure with a clear view of what they're actually allocating to, I favor the broad EM index funds like IEMG for core diversification and a separate, explicit semiconductor or AI infrastructure ETF for the chip thesis. Combining them in one product with a misleading name doesn't make the portfolio cleaner — it makes it harder to manage.

In my opinion, the ex-China ETF is a legitimate way to bet on AI supply-chain growth at a discount to U.S. tech valuations. But it is not a diversified emerging market allocation. Investors who treat it as one are buying a narrative, not a portfolio.