NextFin News - Emerging-market funds are discovering that the AI boom they hoped would diversify their portfolios may instead be narrowing them. A growing share of the asset class’s performance is being pulled by a handful of semiconductor and memory-chip names in Taiwan and South Korea, while the same U.S. mega-cap technology complex that dominates global benchmarks continues to shape where capital flows, how benchmarks behave, and which managers can still claim true diversification.
The tension is simple: investors buy emerging markets to escape U.S. concentration, yet the strongest parts of the emerging-market complex now sit inside the global AI supply chain. That makes the asset class look less like a broad-country bet and more like a disguised technology trade. The question is not whether AI matters — it clearly does — but whether its current grip on emerging markets is a cyclical burst that will fade with the next inventory reset or a structural regime shift that has permanently altered how emerging-market capital is earned.
A Rally That Looks Broader Than It Is
The latest concern is not that emerging markets are underperforming. It is that they are outperforming for reasons many allocators did not buy them for in the first place. In 2026, funds tied to Asia and emerging markets have benefited from demand for the hardware behind the AI buildout, especially in South Korea and Taiwan. Morningstar’s coverage of the year’s best-performing funds said the gains have been driven by TSMC, SK Hynix and Samsung Electronics, and quoted a manager on the underlying theme: “The build out of AI infrastructure has had a huge impact on emerging markets, particularly Korea and Taiwan.”
That sentence matters because it captures the mechanism. Emerging markets are not being lifted evenly by growth in consumption, credit, reform or commodity prices. Instead, the marginal bid is concentrating in the same narrow industrial bottleneck that has powered the global AI capex cycle: advanced logic chips, memory, packaging and the manufacturing ecosystems around them. TSMC is the most obvious example, but it is not alone. SK Hynix and Samsung Electronics have also become direct beneficiaries of a buildout in data-center demand, high-bandwidth memory and other AI-linked components. When those names rise, they can lift the index and the fund at the same time. When they fall, the same concentration works in reverse.
That is why the complaint from fund managers is not about one stock or one country. It is about construction. A portfolio marketed as geographic diversification can end up inheriting the same thematic risk as a U.S. growth fund if the top weights are dominated by AI supply-chain winners. The diversification is real at the country level, but not necessarily at the factor level.
There is another layer to the problem. Emerging-market investors have long relied on the idea that a country mix spreads risk across different economic cycles. Yet the AI trade has compressed those cycles. Taiwan and South Korea have become more sensitive to the same variables that drive U.S. megacap technology: cloud capex, chip ordering trends, memory pricing and the market’s confidence in a sustained AI investment cycle. That makes the asset class more correlated with global tech than many managers expected, and it blunts one of the main reasons international allocators use EM in the first place.
Seen that way, the story is not really about a title number or a single company. It is about a structural change in the transmission of returns. Capital no longer enters emerging markets just through commodity demand, export growth or dollar weakness. It increasingly arrives through the AI hardware bill, and that bill has a short list of obvious winners.
Why The Same Trade Keeps Reappearing
The immediate explanation is cyclical. AI capex is strong. Orders are flowing. Semiconductor supply remains tight in the right places. Memory pricing is firm enough to reward producers. A cyclical read says the result is temporary: once data-center buildouts normalize, the earnings contribution to Korea and Taiwan should fade, and broader EM sectors could reclaim leadership.
That view is not wrong, but it is incomplete. A cyclical thesis explains why the trade can become stretched, but it does not explain why the concentration has been so persistent. For that, the structure matters. The global AI stack still depends on a small number of firms with the manufacturing depth, scale and ecosystem control to produce leading-edge chips and memory at acceptable yield. That is not a generic growth story. It is a capacity bottleneck story. When a bottleneck becomes the profit center, the market naturally keeps bidding the bottleneck owners.
The best way to see the mechanism is to trace the chain. U.S. AI capex rises first. That pushes orders into the semiconductor supply chain. The supply chain then channels profits into the firms with the most leverage over leading-edge production and memory supply. Those firms sit disproportionately in Taiwan and South Korea. Their market weights rise, which in turn raises the influence of those countries inside emerging-market benchmarks. The consequence is that an EM fund increasingly behaves like a chip-cycle fund with a regional wrapper.
“The build out of AI infrastructure has had a huge impact on emerging markets, particularly Korea and Taiwan.”
That quote is the clearest description of the first-order mechanism. The second-order effect is more interesting. If EM performance is being carried by a narrow AI hardware cohort, then a slowdown in global capex would not simply cool returns; it could also reverse the diversification narrative itself. Investors who bought EM as a hedge against U.S. megacap concentration would discover that they had simply swapped one concentration for another. That is a different risk profile, because it is not just about volatility. It is about where the asset class gets its alpha.
The strongest case for calling this a structural shift is that the old EM map no longer matches the new economic map. Ten years ago, the largest emerging-market return drivers were often banks, commodities, state-owned cyclicals or broad domestic growth stories. Today, the most influential names are the ones tied into AI infrastructure, and the center of gravity has shifted toward a smaller set of semiconductor ecosystems. If the index weight follows earnings, and earnings follow AI infrastructure, then the concentration is not a temporary anomaly. It is the result of a new profit engine.
Still, the cyclical counter-thesis deserves real weight. Semiconductor supply chains have a long history of boom-bust dynamics, and memory prices in particular can revert quickly once supply catches up. TSMC, SK Hynix and Samsung do not enjoy immunity from the usual capital-cycle laws. If AI demand cools, if hyperscaler spending slows, or if inventories rebuild too quickly, the current leadership can unwind faster than most managers expect. The fact that concentration has become visible is itself a warning that the trade is crowded. Crowding can produce sharp corrections even when the underlying theme remains intact.
The question is therefore not whether the AI trade can reverse. It can. The real question is whether reversal would restore old-style diversified EM leadership or merely rotate leadership among the same few large technology-linked names. A cyclical pullback would lower returns. It would not necessarily restore the diversified country story investors thought they were buying.
Who Wins If Concentration Persists, and What Would Prove This View Wrong?
If the current pattern persists, the beneficiaries are obvious. Taiwanese and Korean semiconductor leaders gain the most direct earnings leverage. Active managers with a tolerance for concentration can continue to own the winners and outperform the benchmark. And global investors who want AI exposure outside the U.S. may continue to find the cleanest expression in emerging Asia, not in broad commodity or consumer names.
The exposed side is equally clear. Broad emerging-market funds that market themselves as diversification vehicles may become more vulnerable to factor shocks than many investors realize. Country-level diversification will still exist, but it will be diluted by a common dependency on AI capex. That matters for allocators who use EM to reduce correlation with U.S. technology. If the same capex cycle and the same supply chain drive both portfolios, the hedge is weaker than it looks.
The medium-term base case is therefore a mixed one. In the next several quarters, AI-linked semiconductor leadership can keep dominating EM returns if hyperscaler spending remains strong and memory pricing stays supportive. Over that horizon, the market is still rewarding scarce capacity. But over a longer horizon, broadening in AI supply, more normalized capex and additional capacity could flatten the outperformance of the current winners and force managers to look for a new source of EM alpha.
The upside scenario is a further extension of the AI buildout. If demand for data-center infrastructure accelerates again, the same trio of forces — TSMC’s manufacturing leverage, SK Hynix’s memory exposure and Samsung’s scale — could keep EM benchmarks tilted toward the same countries and industries. The downside scenario is a classic capital-cycle reversal: if cloud spending slows, if chip inventories rise, or if pricing weakens, the current concentration premium could evaporate quickly and take benchmark returns with it.
The single falsifying signal to watch is a sustained break in AI capex momentum. If hyperscaler spending and semiconductor order books turn down for multiple quarters, the structural case for EM’s AI-driven concentration weakens materially. A broad-based rotation away from Taiwan and South Korea into the rest of the EM universe would be the first sign that the old diversification story is reasserting itself.
For now, though, the market is telling a less comforting story. Emerging markets are not escaping the AI trade. They are being reorganized by it. And that may be the clearest sign yet that the world’s biggest technology cycle is no longer a U.S.-only story.
The next shock to emerging markets may not come from outside the AI trade. It may come from the moment investors realize the trade already owns them.
Explore more exclusive insights at nextfin.ai.
