Nvidia CEO Jensen Huang thinks the technology giant and the U.S. have a China problem. He lamented on the B2G podcast that the Asian country was only "nanoseconds behind" U.S. chipmakers when it comes to building hardware crucial to cashing in on the AI boom. Yes, Nvidia dominates AI, but growth has slowed and Nvidia stock has had an uneven ride.
Now comes a bigger challenge: Beijing appears to be squeezing Nvidia even more, as Chinese companies look to challenge the U.S. behemoth on the AI front.
The same day Huang's podcast appearance dropped, Bloomberg reported that Chinese AI company iFlytek was training large language models on Huawei's Ascend chips. That's a major shift that underscores how China's biggest tech firms are adapting to U.S. sanctions by replacing Nvidia hardware with domestic alternatives.
It's one of the clearest signs yet that China's push to decouple from U.S. chip dependence is gaining traction, an important shift for investors tracking Nvidia stock's exposure risk.
Nvidia Stock's Bumpy Year
Nvidia is the world's most valuable company, the first to top a $4 trillion market value. It dominates AI training GPUs. But Chinese companies are pivoting to homegrown hardware, a potential threat to Nvidia's position in the Asian country.
Nvidia also faces problems in the U.S. In July, Washington announced a ban on sales of Nvidia's H20 processors to China before walking the decision back. Those export restrictions would hamper Nvidia's ability to generate revenue in one of the company's largest markets.
The company reported zero sales from its China H2O chip in its fiscal second-quarter earnings released on Aug. 27. Management did not include H20 sales to China in its fiscal third-quarter revenue outlook of $54 billion.
In August, Nvidia struck a deal that secured a license to sell its H20 AI chips in China, but, in exchange, the company will give 15% of its revenue from chip sales in China to the U.S. government.
Back-and-forth trade talks and concern over slowdowns in AI spending have sent Nvidia stock on a bumpy ride in 2025. Shares have rallied off April lows, climbing more than 115% and breaking to new highs.
Nvidia stock rose to a new high Thursday, closing at 192.57. Shares remain in a buy zone above an 184.48 entry out of a flat base. Nvidia stock is back above its 10-week moving average after testing it in the stage two base. Shares rallied Thursday, topping an all-time high of 191.05 reached on Oct. 2.
This week, Huang reassured investors that demand for artificial intelligence computing has gone up "substantially" in the last six months.
"I think we're at the beginning of a new buildout, beginning of a new industrial revolution," Huang told CNBC on Wednesday. Demand is "really, really high" for Nvidia's Blackwell graphics processing units, he said.
Beijing Tightens Grip On AI Chips
Soon after Trump's export bans, China unveiled its own moves against Nvidia. Chinese regulators cautioned companies against buying the H20 and pressed them to explain why they would pick it over local alternatives, the Financial Times reported.
That push has since extended to Nvidia's newest China-only chip, the RTX Pro 6000D. The Cyberspace Administration of China ordered firms including Alibaba and TikTok parent ByteDance to call off testing and suspend planned purchases, effectively shutting down what had been Nvidia's last significant product offering in the second largest economy in the world, according to the Financial Times.
Huang said in September he was disappointed by reports that Beijing had barred major Chinese tech firms from buying the company's processors designed specifically for their market.
Nvidia's China Struggle
It is not yet clear why China is targeting Nvidia. Chris Miller, author of "Chip War: The Fight for the World's Most Critical Technology" and a professor at Tufts University's Fletcher School, told IBD he believes Beijing is seeking leverage ahead of trade talks with the U.S. while also pursuing a longer-term goal of reducing reliance on foreign-made semiconductors.
"What we've seen over the last couple of decades is that China's learned to produce almost everything domestically," Miller told IBD. "Chips are the one product that China still imports at scale, because at least right now, China can't produce the most cutting edge chips at the large volumes that are needed to meet its own domestic demand."
In mid-September, Chinese regulators said a preliminary probe found Nvidia in breach of the country's anti-monopoly law and that the investigation remains open. The comments came after the United States confirmed in August that Nvidia would pay Washington 15% of revenue from certain AI chip sales in China.
Huang has described China as a $50 billion opportunity with 50% annual growth, a vision now facing fresh obstacles.
Miller said the booming AI demand in the U.S. from companies like OpenAI, Google and Amazon has already reduced China's importance in Nvidia's sales.
At the same time, U.S. export limits have kept Nvidia from selling its most advanced chips into China, leaving the company less dependent on that market than it was five years ago.
"The moment China can produce cutting-edge products at home, it dramatically reduces, in some cases just stops, buying foreign-made products," Millers said. "That's the risk that Nvidia faces, just like every other manufacturer. If China can do what you can do, it stops buying your products."
Nvidia Stock's China Competitors
Nvidia's graphics processors are still considered the industry standard. But Beijing is moving to cut reliance on the company and boost domestic alternatives.
The Financial Times also reported that China regulators recently called in Huawei, Cambricon, Alibaba and Baidu to assess how their processors stack up against Nvidia's China-only products.
Regulators concluded that Chinese AI chips now perform at a similar level, and in some cases better, than the models allowed under U.S. export controls, the FT reported, citing an unnamed source.
The battle over chips carries global weight because China is the world's second-largest economy and AI is one of its fastest-growing industries. For Nvidia, losing ground in that market could mean giving rivals a foothold in one of the biggest arenas for AI adoption. And could impact Nvidia stock's future growth.
"I think the key question is when will China be able to produce enough of the high end chips it needs domestically?" says Miller.
Here are some of the companies are racing to fill that gap in China.
Huawei
Miller called Huawei "the most important by far" of the Chinese companies challenging Nvidia.
The telecom and electronics giant now produces its own Ascend line of AI chips. Huawei's customers include AI tech company iFlytek which has said its LLMs are trained entirely on Huawei's computing platform.
However, Miller said Huawei's biggest challenge is manufacturing.
"It's a key bottleneck right now," he said. "The leading Chinese companies can't access the most advanced chipmaking tools, equipment that's produced in Europe or the United States or Japan. As a result, they can't produce large enough volumes of high end chips, at least not yet, to meet China's domestic demand."
That's why a key challenge for Huawei is how to help its partners, the "actual chip manufacturers in China" find ways to "scale up their production so that they can produce thousands and eventually millions of these."
Huawei is rolling out new models in its Ascend chip line aimed at handling AI workloads, the Wall Street Journal reported in September. The next chip is expected early next year.
While its processors still lag Nvidia's on advanced features, Huawei is betting that its networking technology can link chips together to boost overall computing power. The company is "making some progress alongside its partners in the Chinese semiconductor industry," Miller said. "We should expect that it produces larger volumes of AI chips, going into the future".
Alibaba
Huawei faces another challenge. Because of U.S. sanctions, the company's chips are not built to run on Nvidia's platform. This is where Alibaba, another Nvidia challenger, enjoys an advantage. Alibaba is developing a new chip that will be compatible, letting engineers reuse programs written for Nvidia hardware.
And Alibaba is investing heavily to boost its AI arsenal. This fall, the company announced plans to exceed its $53 billion AI infrastructure budget after CEO Eddie Wu said demand had "far exceeded expectations."
Alibaba earns much of its profit from cloud computing, where it competes in Asia with Amazon Web Services, Microsoft and Google, even as e-commerce remains its largest business globally.
Though best known worldwide for its e-commerce business, the company was once one of Nvidia's largest customers.
Now it has developed a new chip it says is more versatile than earlier versions.
Alibaba also introduced its biggest AI language model yet, the Qwen3-Max in September.
The company said it will bring a suite of Nvidia development tools into Alibaba Cloud to support "physical AI," a category that includes robotics and self-driving cars. In a move that stands apart from Beijing's recent curbs, Alibaba called the deal a "milestone collaboration" with Nvidia.
Its cloud division will adopt the U.S. firm's full physical AI stack to help drive humanoid robot development.
BABA stock charged higher on the news. Shares powered out of a long consolidation on Sept. 11 and quickly reached the 20% profit taking zone a couple of weeks later.
MetaX
MetaX, a Shanghai-based AI startup founded by former AMD engineers, counts Chairman Chen Weiliang among its leaders. Chen previously served as AMD's global head of GPU product design.
In July, MetaX introduced a chip it says could replace Nvidia's H20. The processor comes with more memory, giving it an advantage on some AI tasks, though it also consumes more power.
By August, the company said it was preparing to move into mass production.
Reuters reported that MetaX has posted steep losses over the past three years, largely due to heavy spending on research and development.
Moore Threads
Moore Threads was founded in 2020 by former Nvidia employees.
Its chairman, Zhang Jianzhong, previously served as Nvidia's general manager for China operations. The company focuses on GPU design and has attracted investment from major firms including ByteDance and Tencent.
In October 2023, the U.S. placed Moore Threads on its Entity List, restricting its access to American technology.
Cambricon
Cambricon Technologies, a Beijing-based AI chipmaker, posted $247 million in revenue for the April-to-June quarter on strong demand for its Siyuan 590 processor.
The company's shares have more than doubled in the past month, outpacing peers and major indexes.
In late August, Cambricon issued a filing cautioning investors about trading risks, saying its stock price had run far ahead of fundamentals, with a price-to-earnings ratio above 5,000 times the industry average.
The warning sent its shares down 5.6% the next day, though the stock remains one of the most valuable in China.
Baidu
Baidu, the Chinese search and AI services giant, develops its own chips for servers and self-driving cars under the Kunlun brand.
Regulators called in Baidu, along with Huawei, Cambricon and Alibaba, to compare their chips with Nvidia's, placing the company among China's main domestic chip developers.
In July, the company unveiled its latest "digital human" technology aimed at businesses. The tool can create virtual livestream hosts using cloning features that mimic a person's voice, tone and body language from just 10 minutes of sample video.
Biren Technology
Reuters reported in July that Biren Technology, a Chinese AI chipmaker, secured about 1.5 billion yuan in new funding and is preparing for a Hong Kong listing.
The company develops general-purpose GPUs along with a software platform that lets developers build applications on its hardware.
Biren's Bili series is aimed at data centers and used for AI training.
The company was also added to the U.S. Entity List last year.
Tencent
Tencent has also built in-house AI accelerators for its cloud business, part of the effort to lessen dependence on Nvidia. At a Shanghai AI conference this year, it introduced a new AI model.
What's Ahead For Nvidia
The question is whether China's chipmakers can truly close the gap with Nvidia. Miller says it may happen someday, but not soon.
Analysts say Beijing's broader aim is AI sovereignty, the ability to train and deploy advanced models entirely on Chinese hardware.
"The question is just how much can China begin to meet its domestic demand? Can it begin to produce chips for exports? Right now, it's a long way away," said Miller. "And the key goal of both Chinese policymakers as well as Huawei and all of its partners are to try to find ways to ramp-up chip production."
