Huawei founder and CEO Ren Zhengfei recently acknowledged that the company’s AI chips remain one generation behind US competitors, particularly Nvidia, amid escalating export controls and US–China tensions. Despite the gap, Huawei aims to close it with innovative workarounds such as cluster computing. This admission highlights both the challenge and ambition of Chinese semiconductor ambitions.
What Ren Zhengfei Said
In an interview with People’s Daily, Ren emphasized:
- Huawei’s flagship AI chip, Ascend, “still lags behind the US by one generation.”
- He downplayed overstated claims, saying “the US has exaggerated Huawei’s capabilities.”
Workarounds to Fight the Gap
Ren outlined key strategies:
- Cluster computing — connecting multiple Ascend chips to match performance.
- Mathematical and theoretical research to complement physical limitations (“non-Moore’s Law”).
These approaches help Huawei’s CloudMatrix 384 server match Nvidia’s systems in practical task
R&D Effort & Investment
- Huawei invests ¥180 billion (~$25 billion) annually in R&D—around ¥60 billion in basic research.
- These efforts drive future innovations needed to catch up with US chip leaders.
Technical Barriers & Sanctions
- US export controls since 2019 limit access to cutting-edge US technology and equipment. asiafinancial.com
- Huawei’s chips remain on 7 nm or 6 nm production nodes, while US firms like TSMC are advancing to 2 nm, a gap of multiple generations
- ASML’s CEO warns this could mean a 10–15 year lag in EUV-enabled chipmaking
Industry Perspective
- Nvidia CEO Jensen Huang counters that China isn’t far behind in AI, calling Huawei “formidable”
- However, analysts note Huawei still trails 2–3 years behind and faces yield issues at 7 nm
Why This Matters
- The delay affects Huawei’s ability to compete globally in AI and smartphone markets.
- Cluster computing offers a short-term fix, but hardware limitations persist.
- Huawei’s investment in research signals long-term ambition to build independent chip capabilities.
