The United States has taken a bold step in its efforts to restrict China’s advancement in advanced technology industries by adding semiconductor manufacturing equipment and high-performance computing (HPC) chips, particularly AI chips, to its Entity List. This move comes as part of a wider campaign to curb China’s growth in these sectors.
One significant development in this latest action is the formal inclusion of NXT:1980Di, which was previously in a regulatory grey area, into the list of controlled items. However, the immediate impact of this decision remains uncertain, as ASML, a key player in semiconductor manufacturing equipment, has already received approval for shipments following their application.
In the realm of HPC chips, the ban has been expanded to include the A800, H800, and L40S series. This expansion is poised to affect China’s tech giants, including ByteDance, Baidu, Alibaba, and Tencent (BBAT), which are expected to reduce their reliance on NVIDIA’s high-end AI servers. This could result in a shift from 5-6 percent of the global AI server market to a mere 3-4 percent.
Chinese cloud service providers (CSPs) are expected to accelerate stockpiling efforts in the short term and invest in independent AI chip development in the medium to long term.
As the clampdown affects Chinese CSPs’ demand for high-end AI servers, analysts anticipate a dash by BBAT to stockpile resources during this transitional phase. NVIDIA is also likely to allocate its limited resources, such as the H800 series, to meet the needs of Chinese customers.
Furthermore, since the US began imposing AI chip bans in 2022, there has been a growing trend among large Chinese firms to develop independent chips. Notably, Alibaba’s Pingtouge is entering the ASIC market, while Huawei is investing in its Ascend series to build a local ecosystem in China.
In the field of edge AI, where smaller models and inference chips dominate, Chinese chip manufacturers like Pingtouge and Hanergy are accelerating their development efforts.
To comply with the new US regulations, AI server chip suppliers are expected to seek AI solutions that align with these mandates. This is likely to drive the growth of computational power leasing models.
Analysts from TrendForce anticipate that AI chip makers, including NVIDIA and AMD, will diversify their product offerings to adapt to the constraints of global geopolitics. This may involve expanding product lines with more modest TFLOPS performance or shifting toward larger die sizes, all while maintaining market reach.
Additionally, these fresh sanctions could prompt a shift among China’s major tech companies and academic institutions, encouraging them to consider renting AI training resources from regions outside China. This shift may bolster NVIDIA’s DGX cloud subscription and leasing model, along with the adoption of more diverse cloud services.
In summary, the recent US sanctions are likely to have far-reaching implications in the global tech industry, reshaping the dynamics of AI chip production, usage, and international collaboration.