Huawei’s founder, Ren Zhengfei, has publicly dismissed claims that his company is closing the technology gap with the U.S., saying its AI chips still trail behind American rivals — a comment seen as a strategic move amid sensitive trade negotiations between Washington and Beijing. His remarks, reported in the Financial Times and originally delivered during a rare interview with China’s state-run People’s Daily, appear designed to temper U.S. fears over Huawei’s technological rise.
“Still lagging behind the U.S. by one generation,” Ren said of Huawei’s Ascend chip, which has become the primary domestic alternative to Nvidia’s powerful AI hardware in China. “The U.S. has exaggerated Huawei’s capabilities — we’re not that strong yet,” he added.
Ren’s comments come just days after Nvidia CEO Jensen Huang warned that Huawei’s advances in artificial intelligence pose a serious threat to American dominance in the field. Huang blamed U.S. export restrictions — which bar Nvidia from selling its most advanced chips to China — for inadvertently creating a “formidable” Chinese rival.
The issue of export controls has now taken center stage in the latest round of U.S.-China trade talks, which resumed Monday in London. While earlier negotiations in Geneva did not address tech restrictions, rising tensions over access to rare earth materials — crucial for electric vehicle manufacturing and weapon systems — have forced the topic onto the agenda.
Washington’s chip sanctions, while targeting Huawei and others, have inadvertently boosted domestic alternatives. Huawei has seen a surge in orders for its Ascend chips as Chinese companies pivot to homegrown technology. However, most leading Chinese AI firms, such as DeepSeek, still rely on Nvidia chips for the computationally intensive training of large language models (LLMs).
Ren acknowledged those challenges, noting technical problems in using Huawei chips for training LLMs, such as difficulties in inter-chip communication and balancing workloads. But he also hinted at significant progress: “Using clustering and stacking, our computing results are comparable to the world’s best.”
Huawei’s latest effort to rival Nvidia centers around its CloudMatrix 384 AI server, which connects 384 Ascend processors using proprietary optical technology. While several clients are currently testing the system, issues remain with heat dissipation and the server’s considerable physical weight.
According to insiders cited by the Financial Times, Huawei’s greatest challenge is building a developer ecosystem that can rival Nvidia’s. Nvidia’s Cuda software framework has become the industry standard due to its ease of use and widespread adoption — something Huawei is racing to replicate.
Ren said the company is investing ¥180 billion ($25 billion) annually in R&D, with one-third allocated to fundamental research rather than commercial products. This long-term commitment, he argued, is key to China’s technological independence.
“AI depends on abundant electricity and advanced network infrastructure,” Ren said. “China’s power generation and grid systems are world-class. Our telecoms infrastructure is the most advanced in the world.”
Despite these statements of confidence, Ren’s tone remained calibrated — more focused on strategic reassurance than nationalistic bravado — a sign that Huawei is carefully positioning itself amid mounting geopolitical scrutiny. With the U.S. closely watching Chinese advancements, downplaying strengths may be Huawei’s most powerful tactic for keeping trade channels open and innovation alive.
