China’s tech giants are facing challenges in the development of artificial intelligence due to a shortage of Nvidia processors and tightening US export controls. As a result, companies like Alibaba, Tencent, and Baidu are shifting their focus towards using domestic chips for AI-related projects.
The recent clampdown on sales of Nvidia’s H20 chip by the Trump administration has pushed Chinese tech companies to explore alternative semiconductor options. With their existing stockpile of Nvidia chips expected to run out by early next year, these companies are looking for new solutions to meet their AI demands.
Baidu’s Shen Dou mentioned that they are considering various chip options to replace Nvidia’s products, especially for inference processing. Alibaba’s Eddie Wu and Tencent’s Martin Lau also highlighted the importance of exploring alternative chip products to meet growing demands.
The export controls imposed by the US have spurred Chinese companies to invest in domestic high-end AI chips, with Huawei’s Ascend chip series gaining traction. State-owned companies and those in sensitive industries have been the primary buyers of Huawei’s chips, but now, a broader range of tech companies are expected to adopt them.
While some companies are cautious about using Huawei’s chips due to potential legal implications, analysts predict that Nvidia may release new chips compliant with US export rules by early July. However, details about these new processors, such as the inclusion of high-bandwidth memory, remain unclear.
Switching from Nvidia to domestic alternatives like Huawei can be a costly and time-consuming process for tech companies. Porting training code from Nvidia’s CUDA software framework to Huawei’s CANN requires significant support and may cause disruptions in AI development.
To mitigate these challenges, companies are considering a hybrid approach where AI training continues on Nvidia chips while local processors are used for inference tasks. Other Chinese chipmakers like Cambricon and Hygon are also being tested, and companies like Baidu and Alibaba are developing their own processors to meet the growing demand for AI technology.
In conclusion, Chinese tech giants are adapting to the changing landscape of AI development by exploring domestic chip options and diversifying their supply chains to ensure continuity in their projects. The shift towards using homegrown chips reflects a broader trend of self-sufficiency and innovation in China’s AI ecosystem.
