China's domestic AI chip market just hit 41% share and nobody here seems to be talking about it
Everyone's watching Nvidia hit new highs and debating whether AI capex is sustainable, but there's a parallel story unfolding in China that I think deserves more attention from an allocation perspective.
Jensen Huang himself said it at Citadel Securities back in October: "We went from 95% market share to 0%." That's the Nvidia CEO publicly admitting the company is completely locked out of China's AI accelerator market after years of rolling export bans. And the IDC data from 2025 confirms what filled the gap. Chinese domestic chipmakers now control 41% of the local AI server market, shipping 1.65 million AI GPUs. Huawei alone shipped around 812,000 chips, roughly 20% of the entire market. Names like Cambricon, Baidu's Kunlunxin, and Alibaba's T Head are all scaling up.
What makes this more than just a geopolitics headline is the capital formation happening around it. In December alone, Moore Threads listed on the STAR Market and closed up 425% on day one after raising $1.13 billion. MetaX debuted the same month, up 693%. Biren Technology became the first Chinese GPU company to list in Hong Kong in January, raising HK$5.58 billion. Baidu spun out its chip unit Kunlunxin and filed for a Hong Kong IPO the same day. This is not speculative froth like a shoe company pivoting to AI. These are companies with real chips shipping to real customers like Alibaba, Tencent, and DeepSeek.
On the hardware side, Huawei is planning to produce about 600,000 Ascend 910C chips in 2026, doubling output, with total Ascend production reaching 1.6 million dies according to Bloomberg. They launched their first chip with in house HBM memory in Q1 and have a three year roadmap through 2028 that aims to double performance each generation. The supply chain is increasingly domestic too, with SMIC fabricating on an enhanced 7nm process.
The investment question I keep running into is accessibility. Most of the direct beneficiaries are either private (Huawei), listed on Shanghai's STAR Market (Moore Threads, MetaX), or still pre IPO. If you're a US based investor trying to get exposure, the options are limited. KWEB and CQQQ are mostly internet stocks. KWEB has literally zero A share exposure and no semiconductor weight to speak of.
The one ETF I found that's structurally different is CNQQ. It allocates about 8.5% to semiconductors and 10.2% to telecom equipment, with holdings like Cambricon at 2.3%, ZTE at 1.1%, NAURA Technology at 1.3%, and Zhongji Innolight at 3.5%. It's roughly 50/50 A shares and Hong Kong listed names, about 100 constituents, and the underlying index returned 39.7% in USD in 2025. One structural point worth flagging is ChiNext exposure. The ChiNext Index, which is essentially China's Nasdaq, just hit a ten year high on April 15 and gained another 3.17% on April 16. About 40% of CNQQ's A share weighting sits in ChiNext constituents like CATL, Mindray, Inovance, and Zhongji Innolight. KWEB has zero exposure to this because it holds no A shares. CQQQ has some A share names but applies a 25% inclusion factor that dilutes the weighting significantly. So when ChiNext rallies hard like it has this week, CNQQ captures that in a way its peers structurally cannot. Not a pure semiconductor play by any means, but it's the closest thing I've found to a broad China tech basket that actually includes the hard tech and chip ecosystem names rather than just Tencent, Meituan, and PDD.
I'm not saying any of this is a slam dunk. China tech carries real risk including regulatory, geopolitical, and currency. But when I see an entire semiconductor ecosystem being rebuilt from scratch with government backing, record IPO demand, and a $30 to $35 billion addressable market in 2026 alone, it feels like something worth understanding even if you decide not to act on it.
Curious how others here are thinking about China tech exposure in general. Are you avoiding it entirely, using broad EM funds, or have you found specific vehicles that give you the exposure you want? Not advice, just trying to figure out where the opportunities are that most people aren't looking at.