US Races to Widen AI Lead as Huawei Presses Ahead

US AI industry leaders have called for slowing the pace of development, but reality tells a different story. Executives at major American artificial intelligence (AI) firms, including Anthropic and OpenAI, have stepped forward to urge a slowdown in technology development. Yet with China's Huawei accelerating its AI semiconductor launches and rapidly closing the gap, intensifying competition between nations makes any slowdown unlikely, observers say.

Huawei announced it will bring forward the release of two AI semiconductors ahead of schedule. Its training chip, the Ascend 960DT, will now ship in the first quarter of 2027, nine months earlier than planned, while its inference chip, the Ascend 960PR, will arrive in the third quarter of 2027, three months early. The company plans to roll out the Ascend 970 and 980 in 2028 and 2029 respectively, launching a new AI chip every year and doubling computing performance with each generation.

Huawei also benefits from a shift in demand: Chinese customers, cut off from AI semiconductor imports by export controls, are increasingly turning to domestic products. Huawei Rotating Chairman Xu Zhijun spoke positively of this changing market landscape.

Guardedness in US political circles runs just as deep. On Monday (local time), President Donald Trump criticized on social media that AI risk-related controversies were nothing but a "hoax," one that would only benefit China. House Speaker Mike Johnson also told reporters on Tuesday, "Slowing the pace of AI technology development would cost us our advantage over China," adding, "This could become a serious national security problem."

Meanwhile, the US government is restricting China's access to high-performance AI chips through advanced semiconductor export controls. Nvidia, for its part, continues to maintain its stronghold on the CUDA platform, a cornerstone of the AI development ecosystem.