EM tech's leadership has moved to the plumbing, and cap-weighted exposure does not own it

Emerging-market technology returns are increasingly coming from companies few allocators can place: King Slide Works, Henan Shijia Photons Technology and EverProX Technologies have posted stock gains of as much as 90% inside the MSCI Emerging Markets Index, according to a Reuters report carried by Cryptobriefing ^[4]. The index itself is up roughly 22% year to date ^[3], so the leadership is running at something like four times the benchmark ^[15], and investors are reported to be rotating out of Taiwan Semiconductor and Samsung Electronics to fund it ^[2].

What these firms sell is unglamorous. They make server components, cooling systems and optical connectivity equipment, the wiring and plumbing that keeps AI data centres running ^[5]. That is a different exposure from a wafer fab, and it is not one you get by owning the two names that carried the previous phase of the rally. Earlier in the year the MSCI EM IT sector surged 62% in a concentrated run dominated almost entirely by large-cap semiconductor stocks ^[6], roughly 40 percentage points ahead of the broad index's year-to-date move ^[16]. The report frames the broadening into smaller infrastructure names as a sign of healthier, more sustainable momentum ^[7], on the argument that when two or three mega-caps drive an index, one bad earnings report can drag down an entire region ^[9].

The mechanical point matters more than the narrative one. These winners are described as sitting within the MSCI EM index ^[4], so an index buyer technically holds them; on a capitalisation-weighted basis, a 90% move in a component supplier does not move the same needle that a mega-cap does, which is why the reported index return is 22% and not something closer to the leaders ^[3][4]. The source does not publish weights, so treat that as arithmetic logic rather than a disclosed figure. An operator or allocator who expressed the AI theme through TSMC and Samsung, on the report's own account, was positioned for the phase that has just been rotated out of ^[2].

Two cautions on the data itself. "As much as 90%" is a maximum across three named companies, not an average for a basket, and the report gives no start dates, float or liquidity for any of them ^[4]. And this is one report, republished from Reuters by a secondary outlet, rather than a set of index attribution numbers ^[1].

The fuel is global AI capital expenditure, which the report identifies as the primary driver of the EM tech rally through 2026, with little sign so far this year of pullback from major cloud and enterprise buyers ^[8][10]. That is also the fragility. Any belt-tightening by hyperscalers would hit the smaller suppliers hardest, because they lack the diversified revenue of a Taiwan Semiconductor ^[11]. Thinly covered small-caps move sharply in both directions once expectations are priced ^[12], and slide rails, cooling technology and optical components are not winner-take-all markets ^[13].

Watch three things. Whether hyperscaler capex commentary stays firm, since the suppliers have no cushion if it does not ^[11]. Whether breadth persists or returns collapse back into the chipmakers, given that 22% is already described as one of the strongest first-half performances in recent memory ^[14]. And whether any of these niche suppliers can show pricing power rather than a cyclical order book ^[13].