How Huawei Plans to Bridge the Chip Gap: Ren Zhengfei on Mathematics, Clusters, and More than Moores Law

While top diplomats from the U.S. and China are discussing trade disagreements, including the controversial technological restrictions in London, Huawei’s founder Ren Zhengfei has made a rare and candid comment about the situation within the company, specifically regarding chip production. The news is… complex, yet filled with hope.

In an interview with the *Renmin Ribao* (published on the front page), Ren openly acknowledged: **“Our standalone chip is still one generation behind the American one.”** It’s surprising to hear such honesty from a leader surrounded by buzz about technological breakthroughs in defiance of sanctions. However, he immediately followed up with an important caveat: Huawei is actively exploring and discovering ways to overcome this gap at a systemic level.

How exactly? **Here’s the formula straight from Ren himself:**

*“We are using mathematics to complement physics.”* – this sounds mysterious and foundational.

*“We are augmenting Moore’s Law with an ‘anti-Moore’s Law’”* – likely referring to alternative methods of enhancing performance when traditional scaling becomes challenging.

*“We are utilizing clustered computing to boost the capabilities of individual chips.”* – and this is already a tangible and powerful advantage.

Everyone remembers the April announcement of the “AI CloudMatrix 384” system? This is precisely the case: 384 Ascend 910C chips combined into a single potent cluster for training AI models. Analysts from SemiAnalysis boldly claimed at the time that this system **could surpass Nvidia’s flagship GB200 NVL72 cluster in certain tasks!** While it may not exceed in all categories, the mere possibility provokes thoughts about the pace of development.

*“Software is not our bottleneck,”* Ren confidently stated in response to concerns raised by American export restrictions. The sanctions imposed since 2019, which have cut Huawei and other Chinese companies off from advanced chips and manufacturing equipment, clearly have a painful impact, but they haven’t brought progress to a halt.

*“The United States has overestimated Huawei’s achievements: Huawei is not that large. We need to work hard to meet their assessment,”* asserted the CEO. He emphasized that Huawei is just one of many Chinese chipmakers, investing a staggering 180 billion yuan (about $25 billion) annually in R&D. Furthermore, a third of that budget is allocated to theoretical research – *“Without theory, there will be no breakthroughs, and we won’t catch up with the U.S.”*

Ascend competes directly with Nvidia’s AI accelerators in the Chinese market. Although Nvidia’s chips are more powerful, Washington’s restrictions prevent the company from selling its most advanced solutions in China, leaving a vast opportunity window for Huawei. Interestingly, in May, the U.S. Department of Commerce explicitly warned that the use of Ascend chips by Chinese companies could violate export regulations. The tension is palpable.

Ren expressed **great hopes for composite chips** (made from multiple elements) as a promising direction. Given the rate of investment and focus on fundamental science, it’s unlikely that Huawei will rest on its current achievements.

Yes, Huawei admits to lagging behind at the individual chip level. However, instead of accepting defeat, the company is channeling enormous resources into R&D and finding ingeniously sophisticated methods (clusters!) to ensure competitive performance at the systemic level. The technological race, especially in the AI sector, is entering an even more complex and fascinating stage. Laws of physics are one thing, but perseverance and mathematics can work wonders. What do you think – will the Chinese manage not only to catch up but to overtake?

Want to stay updated on AI news? Subscribe to our **Telegram channel** [**BotHub AI News**](https://t.me/bothub)!