Washington’s strategy of using chip controls to constrain China’s AI progress appears to be losing force as China moves to expand domestic production and U.S. companies use AI to design new chips faster. Chinese officials have set goals for 70% AI chip self-sufficiency by 2027, while Beijing has declared a goal of full self-reliance.
Huawei’s Ascend 910C is at the center of that push. After U.S. sanctions cut off TSMC production, Huawei shifted manufacturing to SMIC and built its own supply chain. The chip is now being mass-produced in large volumes and sampled to major customers such as Alibaba and ByteDance, though its performance still trails Nvidia’s H200.
The article also points to a more fundamental shift in the United States: OpenAI’s first in-house inference chip, Jalapeño, was co-developed with Broadcom from initial design to tape-out in just nine months. That rapid timeline reflects a new era in which AI tools are helping design the next generation of AI chips.
Taken together, the developments suggest that U.S. export controls alone may not be enough to slow China’s chip progress for long. The broader race now includes both semiconductor manufacturing and the use of AI itself to accelerate chip design.
