Login to Continue Learning
**Alibaba’s AI Chip Targeted for Inferencing Workloads**
Beijing is prioritizing the development of local AI technologies to reduce reliance on American counterparts. Companies like Huawei, Cambricon, and MetaX are investing heavily in domestic chips. Now, Alibaba is reportedly developing its own AI chip designed to fill the “NVIDIA void” and fully utilize homegrown technology.
**Key Aspects of China’s Domestic Chip Push**
China aims to enhance inferencing capabilities on domestic AI chips but still lags behind NVIDIA when it comes to model training. Huawei’s Ascend chips are seen as ideal for inferencing workloads, whereas they struggle in model training. This strategy is part of Beijing’s broader plan to increase domestic chip production and reduce reliance on foreign technology.
**Alibaba’s Chip Development**
The new AI chip is believed to have been developed by a Chinese company like SMIC, which has 7nm production lines operational. Given this, Alibaba likely used the 7nm process for its AI chip, putting it on par with solutions from Huawei and Cambricon. Despite advancements, China faces significant capacity issues due to outdated equipment and inefficient manufacturing techniques.
**NVIDIA’s Continued Dominance**
China still relies on NVIDIA’s AI chips for model training, as no domestic options can match the American tech stack in creating capable large language models (LLMs). Additionally, NVIDIA CEO Jensen Huang has been pushing towards a “Blackwell-based” product for China, highlighting the urgent need for a more advanced solution to maintain relevance in the region.
In summary, while Chinese firms are making strides, they still have a long way to go before challenging global tech giants like NVIDIA.
