MetaX confidentially files for Hong Kong IPO with Huatai

Shanghai GPU designer MetaX has confidentially filed for a Hong Kong listing, working with Huatai International Financial Holdings and targeting a share sale by the end of 2026, according to reporting from the South China Morning Post. The move comes barely half a year after MetaX raised about 4.2 billion yuan, roughly $600 million, on Shanghai's STAR Market on December 17, 2025, a debut where the stock jumped close to 700% on the first day.

The subtext is straightforward. A dual listing widens the pool of capital MetaX can pull from, especially the international investors who cannot easily buy A-shares. The company's own list of uses for the proceeds is telling: next-generation GPU research and development, software ecosystem expansion, supply-chain investments, and potential acquisitions. Building silicon that customers actually deploy at scale takes years of sustained spending on the chip itself, on high-bandwidth memory supply, and on the software layer underneath, and equity funding this deep is the realistic way to keep that going.

MetaX is not alone. Moore Threads, sometimes tagged "China's Nvidia," debuted on the mainland roughly two weeks before MetaX in a listing of about $1.1 billion and closed its first day up more than 400%, per CNBC. Biren Technology, founded in 2019, has since kicked off its own Hong Kong IPO, Tom's Hardware reported. Beijing has been stepping up IPO approvals for semiconductor firms as part of a tech self-sufficiency push against US export curbs, and investors are clearly willing to bid these names hard.

The honest caveat is that the SCMP piece is sourced to unnamed people familiar with the matter, and MetaX did not respond to a comment request, so timing and deal size should be treated as reported rather than settled. What the reporting does not spell out is how much MetaX intends to raise in Hong Kong, at what valuation, or how much of the offering might be taken by mainland-linked strategic buyers versus genuinely fresh international money. It also does not break down what share of the proceeds funds silicon versus the messier and arguably harder problem of building software that can substitute for CUDA.

For anyone selling into, or building on, Chinese cloud infrastructure, the direction of travel is the part worth watching. Even if none of these startups matches Nvidia on the frontier soon, the collectively well-funded domestic field they are building is probably enough to shift more Chinese workloads onto local accelerators over the next couple of years.