NVIDIA’s top acientist says Huawei is gaining ground in AI

NVIDIA’s top scientist issues a stark warning about Huawei

NVIDIA’s chief scientist, Bill Dall, says Huawei is quickly closing the AI gap, fueled by US export restrictions. Once trailing in AI chip and software development, Huawei is accelerating, thanks partly to an influx of former NVIDIA researchers.

Dally’s comments reflect growing concern inside NVIDIA that America’s restrictions may backfire, pushing talent and innovation toward China instead of holding it back.

From 1/3 to 1/2, China’s share of AI researchers surges

In 2019, around one-third of global AI researchers were in China; today, that figure is nearly half, according to NVIDIA chief scientist Bill Dally.

With fewer options to access NVIDIA’s tech, China has invested in building its AI talent and ecosystems, catalyzing a brain gain that may have lasting global implications.

US export controls, a double-edged sword?

Export restrictions were designed to limit China’s access to high-end AI hardware, but NVIDIA argues they have unintended consequences.

Dally and CEO Jensen Huang warned that the policy inadvertently fosters domestic innovation in China. Instead of stalling progress, it’s forcing Chinese firms to go all-in on their silicon and in some cases, succeed faster than expected.

Huawei poaching NVIDIA talent in China

Dally alleges that Huawei has hired ex-NVIDIA researchers to bolster its AI efforts. This quiet talent migration is helping Huawei mimic NVIDIA’s capabilities and develop competing alternatives.

It’s a significant concern for Team Green, whose dominance in AI has historically relied on its talent pipeline as much as its chips. Losing that intellectual capital could shift the balance in the global AI race.

NVIDIA’s financial pain is also driving its message

This warning isn’t purely patriotic; it’s financial. NVIDIA reported a $4–5.5 billion Q1 write-down and warned of up to $8 billion in Q2 revenue losses due to H20 ban.

As CEO Jensen Huang puts it, “If they don’t have enough NVIDIA, they will use their own.” Translation: China will move on, with or without them. NVIDIA’s bottom line could suffer in the long term.

Huawei’s ascend chips are rising alternatives

Huawei’s Ascend 910C and 920 chips, built in partnership with China’s SMI, are fast becoming viable options for AI developers in China.

Though not as powerful as NVIDIA’s top-tier silicon, these chips are homegrown, locally available, and politically safe. For Chinese companies facing sanctions, reliability matters more than raw power. And Huawei is stepping in to fill the void with increasing confidence.

NVIDIA’s CUDA vs. Huawei’s CANN, software showdown

NVIDIA’s CUDA ecosystem has been the backbone of modern AI development. However, Huawei is developing CANN (Compute Architecture for Neural Networks), a rival framework designed to do much of the same work.

While not yet equivalent in scope or adoption, CANN is rapidly maturing. If Huawei closes the gap in software, it could reduce global dependence on CUDA, which would be a seismic shift.

“Formidable” Huawei is making NVIDIA nervous

At Computex 2025, Jensen Huang didn’t mince words. He called Huawei “quite formidable” and acknowledged that Chinese AI firms are finding ways to compete, even without NVIDIA’s help.

His comments suggest that what was once dismissed as an underdog is now a genuine rival, armed with domestic talent, growing hardware, and government backing.

The geopolitics behind the AI chip war

This is more than a business rivalry; it’s geopolitical. The US sees AI as a strategic domain and has placed export restrictions accordingly. China sees AI as a path to technological independence.

Companies like Huawei are now being positioned as national champions. And firms like NVIDIA are stuck in the middle, losing access to one of the largest markets on Earth while watching competitors gain ground.

NVIDIA’s China strategy is being redrawn

With its Hopper chips no longer exportable and Blackwell architecture facing scrutiny, NVIDIA is forced to reconsider its roadmap.

It may look to alternative channels, like cloud-hosted inference in Hong Kong or chip-leasing deals that skirt direct sales. However, none of these options are easy, and China’s appetite for native solutions is growing.

Losing the hardware race isn’t the only threat

Even if NVIDIA maintains an edge in chip performance, its global influence could still shrink. Why? China is building entire AI stacks from silicon to frameworks without US help.

That means less reliance on NVIDIA, AMD, or Intel in the long run. In the AI age, platform independence is power, and Huawei is closer to achieving it than many thought possible.

Chinese tech’s rise comes with global implications

As Huawei and others catch up, a bifurcated AI world is emerging. Western AI will be built on NVIDIA, OpenAI, and CUDA. Chinese AI will run on Ascend chips, CANN, and homegrown models.

This division may mirror what happened with internet platforms, cloud infrastructure, and app ecosystems. For global developers, it could mean building and deploying in two separate AI environments.

Huawei’s AI surge is accelerating R&D cycles

Driven by urgency, Huawei is accelerating its chip and software development cycles. Reports suggest it’s already planning successors to its current Ascend models, aiming to close the performance gap by 2026.

Combined with AI model development from firms like Baidu, Alibaba, and iFlyTek, China is forming a full-stack AI industrial base, without needing to lean on the West.

The export ban may have backfired for the U.S

Instead of slowing China down, the export restrictions might have simply rerouted its trajectory. Bill Dally argues that U.S. policies have given Chinese companies the incentive and necessity to become self-reliant. And in some ways, it’s working.

Huawei is stronger, Chinese AI firms are more innovative, and global influence may shift faster than policymakers anticipated.

Huawei isn’t slowing down anytime soon

Despite the headwinds, Huawei is doubling down. From chip development and AI talent acquisition to partnerships with state-backed labs and universities, the firm is expanding at speed.

Whether its next chip rivals Hopper or Blackwell doesn’t matter; it’s becoming less dependent, more autonomous, and more attractive to developers looking for freedom from U.S. export politics.

Want to see what Huawei’s up against? Here’s a look at NVIDIA’s latest RTX 50-series specs and pricing.

NVIDIA’s wake-up call, the race is far from over

Bill Dally’s message is clear: Huawei is gaining, and the AI race is tightening. U.S. restrictions may have set this in motion, but it’s up to industry and policymakers to decide what happens next.

Will America double down, recalibrate, or fall behind? One thing is sure: the era of unquestioned U.S. dominance in AI hardware is ending, and NVIDIA knows it.

Wondering who else is eyeing NVIDIA’s throne? Intel’s Arc B770 might just be the next contender.

Do you think Huawei will gain control over AI in most of the globe? Please share your thoughts and drop a comment.

Read More From This Brand:

• Nvidia Debuts AI Assistant To Optimize Your Gaming Rig
• Nvidia Plans to Make AI Chips in the US
• Nvidia Wants AI to Solve Its Own Power Issues

Don’t forget to follow us for more exclusive content on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.