Introduction:
As the U.S. sharpens its export restrictions on cutting-edge chips—especially the GPUs that power modern AI—China finds itself at a turning point. To stay in the global race, it must tap into its own tech muscle and engineering talent like never before. Rising to the challenge, two of China’s hottest GPU startups—Moore Threads and MetaX—are making bold moves.
They’ve just filed for a $1.65 billion (12 billion yuan) IPO on Shanghai’s STAR Market, aiming to fuel the next generation of homegrown AI chips.
It’s not just a funding play—it’s a statement: China is gearing up to build its own Nvidia rivals.
Follow us on Twitter here
5 Key Takeaways
IPO Size: Moore Threads and MetaX seek a combined ¥12B to fuel R&D and manufacturing.
Target Market: China’s AI and GPU sectors, now restricted from U.S. chip imports.
Loss-Making but Scaling: Both report large losses, mostly due to high R&D investment.
Geopolitical Advantage: U.S. sanctions are forcing Chinese firms to “buy local.”
Strategic Timing: IPOs come as China pushes to reduce chip import reliance by 70% by 2025.
Snapshot: IPO and Financials at a Glance
| Company | IPO Target | 2024 Revenue | 2024 Loss | Founders’ Background |
|---|---|---|---|---|
| Moore Threads | ¥8 billion | ¥438 million | ¥1.49 billion | Ex-Nvidia China General Manager |
| MetaX | ¥3.9 billion | ¥743 million | ¥1.4 billion | Ex-AMD GPU Global Design Leader |
Both companies were founded in 2020 and aim to become China’s answer to Nvidia, with domestic GPU solutions for AI, cloud computing, and high-performance graphics.
U.S. Export Curbs: Turning Sanctions into Strategy
In April 2025, Washington banned Nvidia’s H20 chips—its most powerful AI accelerators—from China.
These sanctions also restrict Chinese firms from using top foundries like TSMC for cutting-edge chips.
This triggered a shift. Domestic companies now scramble to replace U.S. chips with local alternatives.
For Moore Threads and MetaX, this isn’t just a challenge—it’s a market opportunity.
“U.S. restrictions are accelerating domestic substitution,” Moore Threads wrote in its filing.
“Geopolitical pressure is creating tighter ties with Chinese customers,” added MetaX.
Deep Dive: What Makes Moore Threads and MetaX Unique
- Moore Threads specializes in GPGPU (general-purpose GPUs), aiming to serve AI workloads, gaming, and data center acceleration.
- MetaX focuses on AI and data center GPUs, with a clear roadmap for 5nm chips and domestic ecosystem compatibility.
Both firms have one big thing in common: founders from U.S. chip giants.
Zhang Jianzhong (Moore Threads) led Nvidia China. Chen Weiliang (MetaX) was AMD’s GPU product head. This experience is crucial—these are people who’ve built GPUs at a global scale.
techovedas.com/67-6-market-share-how-tsmc-dominated-the-foundry-race-in-q1-2025/#google_vignette
Market Context: China’s GPU Demand Is Exploding
According to Omdia, China’s AI chip market is set to grow by 35% CAGR through 2027. The demand comes from:
- Chinese cloud service providers (Alibaba, Baidu, Tencent)
- Smart manufacturing and autonomous driving
- AI research and national infrastructure projects
With Nvidia out of reach, these companies must now turn to domestic options.
Strategic Analogy: Laying Fiber in the AI Desert
Think of AI chips as internet cables in a fast-growing city. If your connection gets cut by a foreign supplier, you can’t wait. You need your own fiber lines—and fast. Moore Threads and MetaX are laying China’s AI infrastructure with homegrown GPUs.
Their IPO money will buy the “shovels and cable”—funding new designs, hiring engineers, and expanding into production.
techovedas.com/wake-up-america-chinas-ai-threat-is-just-months-away
Technical Roadmap
Both firms are heavily R&D-focused. Here’s what their IPO funds will support:
- Moore Threads: Develop next-gen GPUs, expand middleware support (AI/graphics), hire 500+ engineers.
- MetaX: Launch 5nm-based chips by 2026, boost compiler stacks, optimize for local AI frameworks.
The future focus is on energy efficiency, parallel compute power, and domestic software stack integration—vital to run large language models (LLMs) inside China.
Risks and Challenges
These firms walk a tightrope:
- Losses are steep (each over ¥1.4 billion in 2024)
- Manufacturing constraints as they can’t access TSMC
- Fierce competition from Huawei, Biren, Cambricon, and others
But the reward? First-mover advantage in a closed but booming market with state support.
Conclusion : Will They Become China’s Nvidia?
U.S. sanctions may have slammed one door shut—but for these startups, it opened a tunnel to fast-track growth.
If they execute well, they might not just survive—they might define the next decade of Chinese semiconductor independence.
For more of such news and views choose Techovedas! Your semiconductor Guide and Mate!
