According to Bloomberg, citing informed sources, Ant Group has made a significant breakthrough in artificial intelligence, reducing the cost of AI model training by 20% using Chinese chips produced by Alibaba and Huawei. This news undoubtedly injects new momentum into China's pursuit of self-reliance in key technologies.
The report further indicates that Ant Group's internal tests show that these Chinese-made AI chips are comparable in performance to those of industry leader Nvidia. If this result is widely verified and applied, it could significantly alter the global AI chip market landscape.
Image Source Note: Image generated by AI, licensed through Midjourney.
Previously, market rumors suggested that DeepSeek, a Chinese AI startup, required far fewer chips than expected for its new model, causing Nvidia's stock price to fluctuate. Ant Group's announcement provides more compelling evidence of the capabilities of domestically produced Chinese AI chips.
Despite US export controls, Nvidia's AI chips remain in high demand in the Chinese market. Reports even suggest Chinese buyers are actively seeking its latest Blackwell series chips. However, if Ant Group's technology is widely adopted and encourages more Chinese companies to use domestically produced AI chips, it could weaken Nvidia's profitability as the most popular AI chip supplier.
As of now, neither Ant Group nor Nvidia has responded to requests for comment.
Ant Group's progress not only showcases China's advancements in AI chip design and manufacturing but also foreshadows the increasing importance of a domestic supply chain in the future development of AI technology. Reducing AI training costs is crucial for promoting the widespread adoption and application of AI, especially in the training of large models with massive computational demands.
