Ant Group is putting a new spotlight on Chinese chips in the AI race. The Chinese fintech giant, backed by Alibaba founder Jack Ma, says it found a way to reduce AI training costs while using chips made by Alibaba and Huawei.
The claim matters because Nvidia remains the most popular AI chip producer, and its position has been highly lucrative. If Chinese-made alternatives gain wider use, that could create new pressure on Nvidia’s role in AI hardware.
What Ant Group says it achieved
Bloomberg reported, citing sources familiar with the matter, that Ant Group used Chinese chips made by Alibaba and Huawei to create methods that cut AI training costs by 20%.
The same Bloomberg sources claimed that the Chinese-made chips performed about as well as Nvidia chips during Ant Group’s tests. That is the central point of the report: the claim is not only about lower cost, but also about whether domestic chips can get close enough to Nvidia’s hardware for some AI training work.
Ant Group and Nvidia didn’t immediately respond to requests for comment, so the public picture remains limited to the reported claims and the sources cited by Bloomberg.
Why the 20% figure gets attention
AI training is expensive because it depends on powerful computing hardware. A reported 20% cost reduction is therefore significant within the facts of this story, especially because the claim involves chips from Alibaba and Huawei rather than Nvidia.
The number also arrives in a market already sensitive to questions about chip demand. Earlier this year, DeepSeek briefly crashed Nvidia’s stock because of speculation that its models require far fewer chips. Ant Group’s reported work fits into that same broader question: how much specialized hardware is truly needed to train competitive AI models?
The source article does not say that Ant Group has replaced Nvidia across all AI work. It says Ant Group was able to use Chinese chips to create methods that reduced training costs by 20%, and that those chips performed about as well as Nvidia chips in its tests, according to Bloomberg sources.
What this could mean for Nvidia
Nvidia chips remain highly sought after, including in China. The source article notes that buyers in China are reportedly getting Nvidia’s latest Blackwell chip despite U.S. export controls.
That demand is why Ant Group’s claim matters. Nvidia’s current status as the leading AI chip producer is both popular and highly lucrative. If Chinese chips made by Alibaba and Huawei catch on, the source article says it could harm that position.
For now, the key word is if. A test result reported through sources is not the same as broad adoption. But the reported comparison with Nvidia chips is enough to make the claim commercially important, because AI hardware buyers care about both performance and cost.
The bigger signal from China’s AI market
The story points to a practical direction for Chinese AI companies: finding ways to train models with hardware available from Chinese chipmakers. In this case, the named suppliers are Alibaba and Huawei, and the company making the claim is Ant Group.
That does not mean Nvidia demand has disappeared. The opposite is also present in the source: Nvidia chips remain highly sought after, and buyers in China are reportedly still obtaining the latest Blackwell chip despite U.S. export controls.
The result is a market with two facts moving at the same time. Nvidia remains central to AI hardware demand, while Chinese companies are claiming progress with Chinese-made chips. Ant Group’s reported 20% training cost reduction is one example of that tension.
What to watch next
The next important question is whether Ant Group’s reported methods move beyond tests and become widely used. The source article does not provide a rollout plan, a product name, or a timeline.
It also does not include responses from Ant Group or Nvidia. That leaves several details unresolved, including how broadly the methods apply and whether other companies will see similar results.
Still, the claim is notable on its own terms. A major Chinese fintech company says Chinese chips from Alibaba and Huawei helped reduce AI training costs by 20%, while Bloomberg sources say the chips performed about as well as Nvidia chips in Ant Group’s tests. In a market where Nvidia’s AI hardware position is both dominant and profitable, that is a development worth watching.