The rally in emerging-market AI computing power stocks is rapidly broadening to a wider array of new targets, as investors appear to be rotating capital from Asia's three trillion-dollar chip manufacturing giants toward smaller-cap companies in the AI computing supply chain that are poised to provide critical hardware support for the booming construction of AI data centers.

Overall, the emerging-market AI computing trade is shifting from its first phase — dominated by Taiwan Semiconductor Manufacturing, SK Hynix, and Samsung Electronics, centered on advanced wafer foundry, 2.5D/3D advanced packaging, and HBM/DRAM/NAND memory — toward rack-level infrastructure segments including server racks, optical communications and interconnects, liquid cooling systems, and power distribution units. Taiwan's King Slide Works, a small-cap tech darling that holds a leading global position in the server/data center rack rail market, alongside smaller peers focusing on optoelectronic chips, optical modules, and optical interconnect technologies, have all emerged among the best-performing constituents of the MSCI Emerging Markets Index this month, with gains reaching as high as 90%. These companies are collectively reigniting bullish sentiment around Asia's AI computing investment theme, which had wavered in June and July following a supercycle that lasted over a year and saw a staggering $17 trillion in value across Asian AI computing leaders.

Where the rally heads next

In the view of Wall Street strategists, the most significant shift in the global semiconductor complex is that July's "AI crowded trade liquidation" increasingly looks like an unwinding of extreme leverage and speculative positions, rather than a fundamental reversal in the AI supply chain. Latest market data shows the Philadelphia Semiconductor Index (SOX) plunged nearly 29% from its June 22 record high to a July 29 low, but has since rebounded roughly 20% in just about three weeks — quickly re-entering technical bull market territory. South Korea, dubbed the "bellwether of AI computing," has amplified this semiconductor rebound story of "fundamentals intact, leverage first to break" to the extreme: the benchmark KOSPI surged 11.5% last week, snapping a seven-week losing streak, with Samsung Electronics and SK Hynix climbing 19% and 16% respectively. Measured from the July 30 close of 5,593.56, the KOSPI has rebounded approximately 24.75% to 6,977.94 by August 14. More critically, assets in single-stock leveraged ETFs in South Korea have plunged from roughly $50 billion to $17 billion, and JPMorgan estimates that hedge fund deleveraging is about 90% complete. Combined with net foreign buying of approximately 3 trillion won on August 14, this signals that global semiconductor positioning is shifting from "forced selling" back to "institutions like hedge funds re-assuming risk."

Who takes the baton from the chip giants?

For many investors, this marks the next major leg of Asia's AI computing bull market. Chipmakers like Samsung Electronics, SK Hynix, and Taiwan Semiconductor Manufacturing benefited from the earliest phase of AI infrastructure buildout. Now, the market's focus is turning to so-called small- and mid-cap "AI cluster pick-and-shovel" players — companies supplying core components for AI server racks, cooling equipment, and high-speed optical connectivity products.

"AI infrastructure is a full-stack capital cycle, and this investment logic is being validated layer by layer," said Gitania Kandhari, a senior portfolio manager at Morgan Stanley Investment Management, which oversees $2 trillion in assets in New York. Kandhari noted that U.S. hyperscale cloud providers have collectively committed and invested nearly $2.4 trillion in AI infrastructure buildout, positioning suppliers of core hardware components at the server rack level, as well as providers of system-level cooling and power distribution units, to capture a very substantial share. These companies, she said, essentially hold "strong order backlogs that provide multi-year revenue visibility." "The market is pricing in that visibility," she added.

As shown in the chart above, new AI investment targets are emerging across emerging markets — investors are no longer confined to the three chip titans, but are increasingly betting on AI infrastructure companies. (Note: Returns are in USD.) With China pushing forward its own AI computing infrastructure buildout aimed at laying the groundwork for long-term growth, nearly all AI infrastructure players in these emerging markets are also eyeing the growth opportunities presented by China's unprecedented AI technology push.

The investment case for AI infrastructure stocks rests on much the same dynamics that drove memory chipmakers' shares to stratospheric heights over the past year: persistent shortages of the equipment and components required to run massive AI training and inference systems, coupled with a very limited number of companies with the capacity to produce them.

"After two years of highly concentrated gains, we are seeing signs of a broadening in AI supply chain leadership," said Nenad Dinic, an equity strategist at Bank Julius Baer in Zurich. He expects emerging market performance in the second half of the year to be more evenly distributed across the various core technology segments of the AI supply chain. Even as investors broaden the scope of AI infrastructure companies they pursue, these names remain heavily concentrated in China and South Korea. This means emerging market equity investing may continue to exhibit a distinctly imbalanced pattern.

"We would not view second-tier AI infrastructure names as a hedge against weakness in the AI mega-caps," Dinic said. "Ultimately, they depend on the same underlying AI capex and infrastructure cycle." By weight, Asian countries account for 82% of the MSCI Emerging Markets Index, leaving only 18% for Eastern Europe, the Middle East, Africa, and Latin America. The rise of emerging AI infrastructure players could further cement the dominance of Chinese and South Korean markets while increasing the technology sector's weight in the benchmark. This also introduces higher concentration risk. With so many large companies tied to AI, any macro shock or any reduction in North American hyperscaler AI capital spending could trigger a broad selloff across the entire emerging market equity index.

Capital always chases growth

The increasingly robust order visibility tied to AI computing is igniting the second-tier emerging market AI trade, which is rapidly broadening from Taiwan Semiconductor Manufacturing, SK Hynix, and Samsung Electronics toward rack-level infrastructure segments such as server racks, optical communications and interconnects, liquid cooling, and power distribution. This is not a case of capital abandoning the three chip giants; rather, as AI clusters are deployed at scale, the market is beginning to price in, layer by layer, the previously undervalued physical bottlenecks within the computing stack.

In the view of Wall Street financial heavyweights like Goldman Sachs and Morgan Stanley, which remain bullish on the AI theme, the AI supercycle is far from over — it is transitioning from the "AI chip buying spree" into the second phase of "building AI factories at scale." In this phase, the next wave of alpha will no longer belong exclusively to the strongest leaders in AI GPU/AI ASIC, but will systematically spread across the full-stack AI infrastructure layer: high-performance data center CPUs, DRAM/NAND/HBM memory, AI PCBs, liquid cooling systems, data center optical interconnect systems, ABF substrates/glass substrates, MLCCs, electronic fabrics, and broad-based wafer foundry.

The underlying logic of this broadening and rotation is AI's evolution from a "bottom-layer chip supercycle" into a full-stack AI capital cycle: an increase in GPU/TPU/AI ASIC counts simultaneously amplifies demand for servers, memory, enterprise NAND storage components, Ethernet switches, optical modules, data center optical communications/interconnects, high-speed connectors, server rack rails, cooling, and power distribution. The larger the training and inference clusters, the higher the port counts, per-rack value, and interconnect complexity. The nearly $2.4 trillion in hyperscaler investment commitments thus flows not only into compute chips but also, along the supply chain, into multi-year orders and revenue visibility for specialized equipment makers.

Morgan Stanley projects that by 2028, nearly $3 trillion in AI-related infrastructure investment will flow through the global economy, with more than 80% of that spending still ahead. Goldman Sachs' latest estimates show global AI capex growing from $765 billion annually in 2026 to $1.6 trillion annually by 2031, with cumulative capex of roughly $7.6 trillion between 2026 and 2031. U.S. data center power demand is expected to rise from 31GW in 2025 to 66GW by 2027, directly spilling AI infrastructure investment into server CPUs, DRAM/NAND/HBM, advanced packaging, liquid cooling, power equipment, transformers, gas turbines, grid connection equipment, data center REITs, and construction.

As the global AI computing theme accelerates from a "concentrated solo dance" of a few mega-cap stocks toward structural broadening driven by orders, supply bottlenecks, and per-rack value, it is important to note that this is not an indiscriminate small-cap AI bull market. Truly investable targets should simultaneously possess irreplaceable technological positions, customer qualification barriers, continuously rising per-rack value, reliable order visibility, and profitability that can convert into free cash flow. Furthermore, second-tier AI computing resource suppliers share the same AI capex cycle as the three chip giants and may not be able to fully hedge against the risk of hyperscaler investment cuts.