EM AI Computing Rally Spreads to Second-Tier Names, Optical Communication and Server Rack Stocks Surge Up to 90%
Nashnova编辑部
The emerging-market AI trade is migrating from chipmakers like TSMC and SK Hynix into server racks, optical networking, liquid cooling and power distribution — three mid-cap names surged as much as 90% this month, but the rally's lifeline still runs through North American hyperscaler capex plans.
What exactly is spreading?
AI infrastructure spending is no longer just about chips. Capital is flowing into everything that keeps chips running — server racks, optical modules, liquid cooling, power distribution units.
This month, Kawako Tech, Siaophotonics and Changxin Bochuang — three mid-caps — ranked among the top gainers in the MSCI Emerging Markets Index, with the strongest up 90%.
This means → the market sees AI hardware bottlenecks shifting from "making chips" to "housing, powering and cooling them." Whoever solves the second problem inherits the momentum.
What just happened to the semiconductor sector?
The Philadelphia Semiconductor Index plunged nearly 29% from its June 22 all-time high to its July 29 low, then rebounded roughly 20% within about three weeks.
Korea's KOSPI rallied from 5,593.56 on July 30 to 6,977.94 on August 14 — a gain of about 24.75%. Samsung Electronics rose 19%; SK Hynix gained 16%.
JPMorgan estimates hedge-fund deleveraging — the forced selling that occurs when funds dump assets to repay debt — is roughly 90% complete. Foreign investors were net buyers of about ₩3 trillion on August 14.
In plain terms = panic selling is largely over. Institutional money has shifted from "forced to sell" to "re-risking," and that pivot opened the window for second-tier names to run.
Why are they rallying? What is order visibility worth?
Morgan Stanley Investment Management senior fund manager Gitanjali Kandhari says US hyperscalers have committed a cumulative ~$2.4 trillion to AI compute infrastructure.
Suppliers of rack hardware, cooling and power distribution "hold robust order backlogs that provide multi-year revenue visibility" — and the market is pricing that visibility in.
Goldman Sachs projects global AI capex rising from $765 billion per year in 2026 to $1.6 trillion per year by 2031, a six-year cumulative total of roughly $7.6 trillion.
This means → more than 80% of the spending is still ahead. Second-tier order books are far from peaking.
Can the second tier hedge a downturn in the leaders?
Julius Baer equity strategist Nenad Dinic expects EM performance in H2 to be more evenly distributed across AI supply-chain segments.
But he warns explicitly: second-tier AI infrastructure firms "depend on the same underlying AI capex and infrastructure cycle" and should not be treated as a hedge against weakness in AI mega-caps.
In plain terms = if TSMC falls because hyperscalers cut AI budgets, rack stocks fall too. They are not a safe haven — just the last to get wet in the same rainstorm.
How big is the concentration risk?
By weight, Asia accounts for 82% of the MSCI Emerging Markets equity index. Eastern Europe, the Middle East, Africa and Latin America together make up just 18%.
The rise of AI infrastructure companies could further entrench China and Korea's dominance in the index, pushing tech-sector weight even higher.
This reflects a structural paradox: the more the rally spreads, the more the index concentrates in Asian tech — and any cut to North American hyperscaler AI capex could drag the entire EM index into a sell-off.
Content is for reference only, not financial advice.