Huawei's Ambitious Three-Year Plan to Challenge Nvidia's AI Chip Dominance

Huawei Plans Three-Year Campaign to Overtake Nvidia in AI Chips

Huawei Technologies Co. openly admits its silicon can't match Nvidia Corp.'s in raw power and speed. So to pack the same punch, China's national champion is counting on its traditional strengths: brute force, networking, and policy support. Huawei on Thursday took the rare step of publicizing a three-year vision for eroding Nvidia's dominance in the AI boom. Rotating Chairman Eric Xu outlined the technology the Shenzhen-based company envisions in painstaking detail during its annual Huawei Connect conference, triggering wall-to-wall media coverage.

Huawei Bets On Brute Force To Outpace Nvidia - NVIDIA (NASDAQ:NVDA)

Nvidia NVDA shares pulled back in premarket trading on Tuesday, surrendering some of Monday's gains following reports that Huawei is aggressively pursuing an alternative path to challenge the U.S. chip designer's dominance. At its annual Huawei Connect conference, Huawei's rotating Chairman Eric Xu unveiled a three-year plan to erode Nvidia's leadership in artificial intelligence. The strategy centers on clustering a massive number of Ascend processors via a new UnifiedBus interconnect system. Also Read: Nvidia's Bold Move Catapults Intel To The Heart Of AI Innovation According to a Bloomberg report on Tuesday, the company claimed this design could move data up to 62 times faster than Nvidia's upcoming NVLink144 technology. Last week, Huawei unveiled a multiyear roadmap to challenge Nvidia by clustering thousands of its Ascend AI chips into powerful SuperPod systems. Eric Xu said the new design will connect up to 15,488 processors and scale into super clusters with about 1 million cards. The company also introduced self-designed high-bandwidth memory and outlined future Ascend 950, 960, and 970 chips. Huawei argues that while its processors trail Nvidia in single-chip performance, massive clustering and faster interconnects can deliver competitive results, Bloomberg reported. Interestingly, so far, Huawei has voiced that its chips lag Nvidia's in raw power and speed. This unusually high-profile rollout marks a significant shift for Huawei, which has maintained a low profile since U.S. sanctions in 2020 cut it off from Taiwan Semiconductor Manufacturing Co. TSM. The announcement also aligns with Washington's tightening curbs on Chinese access to advanced semiconductors and Beijing's push for national champions like Huawei to develop domestic alternatives. Nvidia's stock had climbed on Monday after the company deepened its collaboration with OpenAI, signing a letter of intent for a strategic partnership. Nvidia committed up to $100 billion to supply at least 10 gigawatts of systems to power OpenAI's next-generation AI infrastructure, beginning with the Vera Rubin platform in 2026. This deal establishes Nvidia as OpenAI's preferred compute and networking partner, ensuring the companies align their hardware and software roadmaps. The rally pushed Nvidia shares near a 52-week high, extending year-to-date gains of more than 36% and outperforming the Nasdaq 100. This highlights Nvidia's pivotal role in Big Tech's AI buildout, despite rivals like Broadcom AVGO also securing chip orders from OpenAI. However, analysts from Bernstein and Jefferies told Bloomberg that Huawei still lags technologically, with a next-generation Ascend chip delivering only a fraction of Nvidia's performance. The company has also not progressed beyond 7-nanometer designs due to a lack of advanced chipmaking equipment. Nevertheless, Huawei insists that its brute force, networking know-how, and government support can narrow the gap. Price Action: NVDA stock was trading lower by 1.00% to $181.77 at last check Tuesday. Read Next: Meta Teams Up With US Government To Bring Llama AI Models To Every Federal Agency Image via Shutterstock NVDANVIDIA Corp$181.77-1.00%OverviewAVGOBroadcom Inc$340.630.54%TSMTaiwan Semiconductor Manufacturing Co Ltd$282.333.56%Market News and Data brought to you by Benzinga APIs