The 2026 global equity market is, at its core, a story about the AI semiconductor supply chain. The countries that house the companies making the chips, memory, equipment and substrates that power AI data centres have dominated global performance rankings. The countries that don't have been left behind.
Nowhere is this dynamic more starkly visible than in the comparison between China and its neighbours. The SPDR S&P China ETF (GXC) is down -6.10% year-to-date. The iShares MSCI Taiwan ETF (EWT) is up +61.26%. The iShares MSCI South Korea ETF (EWY) has gained +104.63%. The gap between GXC and EWT — two markets in adjacent geographies competing across overlapping technology sectors — is 67.4 percentage points in a single calendar year. Between GXC and EWY: 110.7 percentage points.
This is not a macro cycle divergence. It is a structural supply chain divergence.
GXC Performance: A Market Stuck in Place
| Period | GXC Return |
|---|---|
| YTD (Jan 1 → June 11, 2026) | -6.10% |
| 1-Year | +2.99% |
| 3-Year | +20.17% |
| Since Iran War (Feb 27) | -7.52% |
GXC's 3-year return of +20.17% contrasts with Taiwan's +113.30% and Korea's +198.53% over the same window. China's equity market has essentially treaded water for three years while the AI semiconductor trade compressed multiples in Taiwan and Korea, then dramatically re-expanded them. The Iran war period (-7.52%) adds to the underperformance — China is a net energy importer and has faced additional headwinds from commodity inflation.
| Country | ETF | YTD | 3Y |
|---|---|---|---|
| South Korea | EWY | +104.6% | +198.5% |
| Taiwan | EWT | +61.3% | +113.3% |
| MSCI ACWI | ACWI | +10.1% | +65.2% |
| China | GXC | -6.1% | +20.2% |
| India | INDA | -11.6% | +13.0% |
Performance Comparison: GXC vs EWT vs EWY vs ACWI
Country ETF Tracker
- GXC
- EWT
- EWY
- ACWI
Trade on eToro – the easy-to-use investing app with 7000+ assets.
Your capital is at risk.
Other fees apply.
The Portfolio: Tencent, Alibaba, and No Chip Companies
The GXC portfolio structure explains the underperformance precisely. GXC tracks the S&P China BMI Index — the broadest China equity benchmark — covering 1,266 securities. Despite this breadth, the top holdings reflect China's listed equity universe: dominated by internet consumer platforms, not AI hardware manufacturers.
| Holding | Ticker | Weight | Sector |
|---|---|---|---|
| Tencent Holdings | 0700.HK | 10.29% | Internet / social media |
| Alibaba Group | 9988.HK | 7.87% | E-commerce / cloud |
| China Construction Bank | 0939.HK | 3.19% | Financials |
| ICBC | 1398.HK | 2.00% | Financials |
| PDD Holdings | PDD | 1.82% | E-commerce |
| Xiaomi | 1810.HK | 1.66% | Consumer electronics |
| Meituan | 3690.HK | 1.61% | Food delivery / tech |
| Bank of China | 3988.HK | 1.60% | Financials |
| Ping An Insurance | 2318.HK | 1.35% | Financials |
| NetEase | 9999.HK | 1.24% | Gaming / internet |
Tencent (10.29%) and Alibaba (7.87%) — the two largest holdings at 18.16% combined — are consumer internet platforms. They benefit from AI in marginal ways (AI-enhanced advertising, AI-assisted customer service, AI model deployment on cloud platforms) but are not fundamental components of the AI hardware supply chain that is driving TSMC, Samsung and SK Hynix to record earnings.
The absence that defines GXC: There is no Chinese equivalent of TSMC (advanced logic chip fabrication), no Chinese equivalent of SK Hynix (HBM memory), no Chinese equivalent of ASML (EUV lithography). The Chinese semiconductor ecosystem — SMIC for logic chips, CXMT for DRAM — exists but operates 1-2 process generations behind the most advanced nodes. The reasons are structural and policy-driven.
Key Facts
| Metric | GXC |
|---|---|
| Forward P/E | 10.87x |
| vs ACWI | -40% |
| Dividend Yield | 2.50% |
| AUM | $458.1M |
| Holdings | 1,266 |
| Expense Ratio | 0.59% |
| Current Price | $90.90 (June 11) |
Why China Is Not in the AI Supply Chain
US export controls are the ceiling. The October 2022 and subsequent US export control packages prohibit ASML from selling EUV (Extreme Ultraviolet) lithography machines to Chinese customers. EUV lithography is the technology required to manufacture chips at sub-7nm process nodes — the nodes used by Nvidia's AI GPUs, Apple's processors, and AMD's server chips. Without EUV access, Chinese fabs cannot produce chips at the performance levels demanded by AI training workloads. SMIC, China's most advanced foundry, is currently limited to 7nm (DUV-based) and above — a meaningful but not closing gap.
HBM memory is a different problem. South Korea's dominance in HBM (High Bandwidth Memory) — the specialised DRAM stacked adjacent to GPU dies in AI accelerators — is not primarily about export controls. It is about accumulated engineering expertise and capacity investment spanning decades. CXMT (ChangXin Memory Technologies) is developing DRAM capability but is not producing AI-grade HBM at commercial scale. The 3-5 year pricing contracts that Samsung and SK Hynix are signing with hyperscalers lock out Chinese entrants for the duration of the current AI capex cycle.
China's AI champion strategy vs hardware reality. China's government has invested heavily in domestic AI model development (Baidu's ERNIE, Alibaba's Tongyi, Huawei's Pangu) and in Huawei's Ascend AI chip programme. Huawei's Ascend 910B and 910C chips are genuine AI accelerators that are finding domestic customers as Nvidia H100/H200 exports are restricted. But Huawei's chips are manufactured by SMIC on DUV-based processes — they are competitive enough for Chinese domestic AI deployment, but they are not components of the global AI infrastructure buildout that is driving hyperscaler capex of $720 billion in 2026.
The result: China is building a parallel AI ecosystem behind the export control wall. That ecosystem is real and growing. But it is not the ecosystem that is driving TSMC's 66.2% gross margins or SK Hynix's record quarterly revenue of $35 billion.
The Valuation Question: Value or Value Trap?
GXC at 10.87x forward P/E — a 40% discount to the MSCI ACWI — is cheap. China has been cheap for three years. The 2.50% dividend yield is meaningful. The breadth of 1,266 holdings provides diversification that single-stock AI plays (EWT, EWY) do not.
| Market | ETF | Fwd P/E | vs ACWI | YTD |
|---|---|---|---|---|
| Taiwan | EWT | 22.71x | +24.6% | +61.3% |
| MSCI ACWI | ACWI | 18.23x | — | +10.1% |
| China | GXC | 10.87x | -40% | -6.1% |
| South Korea | EWY | 8.35x | -54.2% | +104.6% |
| Brazil | EWZ | 8.22x | -54.9% | +9.6% |
The value case for GXC rests on two arguments. First, China's domestic consumption recovery — slowed by the property sector correction and deflationary pressures — could re-accelerate if fiscal stimulus strengthens, lifting Tencent, Alibaba and China's consumer-facing internet companies. Second, Chinese AI companies (Baidu, Alibaba Cloud, Huawei) could receive a domestic re-rating as China's parallel AI ecosystem matures.
The value trap argument is more straightforward: GXC is cheap because it deserves to be cheap. The export control ceiling on semiconductor technology is not going away. The geopolitical risk premium for holding Chinese equities — in an environment of ongoing US-China technology decoupling — is structural, not cyclical. Three years of cheapness has not produced a re-rating.
The 3-Year Verdict
Over three years, GXC has returned +20.17%. Taiwan has returned +113.30%. Korea has returned +198.53%. The S&P 500 has returned +70.07%. China has been the worst-performing major developed/emerging market in the AI era — by a substantial margin.
The analytical conclusion is not that China is uninvestable. It is that China is not in the trade that is driving 2026's global equity returns, and the structural reasons for that absence — export controls, decades of accumulated chip manufacturing expertise in Taiwan and Korea, HBM pricing contracts — are not resolving within the current market cycle.
GXC at 10.87x is pricing that reality accurately. Whether the next catalyst changes it depends on: (1) whether Huawei's domestic AI chip programme achieves performance parity on key benchmarks; (2) whether US-China diplomatic normalisation creates any easing of export control pressure; or (3) whether China's domestic AI ecosystem generates earnings growth sufficient to re-rate the platform companies that dominate GXC's portfolio.
None of these conditions are currently in evidence.
Track GXC and all country ETFs at countryetftracker.com. Global AI supply chain comparison at countryetftracker.com/compare.
Frequently Asked Questions
Why is China's GXC ETF down while Taiwan and Korea soar?
GXC is down -6.10% YTD while EWT is up +61.26% and EWY has doubled (+104.63%) because China's listed equity market — dominated by Tencent (10.29%) and Alibaba (7.87%) — is not part of the AI semiconductor supply chain driving 2026's global equity winners. US export controls prevent China from accessing EUV lithography equipment needed to manufacture advanced AI chips. China has no commercial-scale HBM memory producer. China's AI capabilities are real but are being built behind a technology access wall that keeps Chinese companies out of the global AI hardware supply chain generating TSMC's $35.9 billion Q1 2026 revenue and SK Hynix's record $35 billion quarterly result.
Is GXC cheap enough to buy at 10.87x forward P/E?
GXC at 10.87x offers a 40% discount to the MSCI ACWI and a 2.50% dividend yield. The value case rests on China's domestic AI ecosystem maturation and potential consumer spending recovery. The value trap argument is that GXC has been cheap for three years without catalysing a re-rating, the export control ceiling on semiconductor technology is structural and not easing, and the geopolitical risk premium for Chinese equities in a technology decoupling environment is unlikely to compress without a significant US-China diplomatic breakthrough. Whether GXC is a value opportunity or a value trap is the central question for China equity investors in 2026.