NVIDIA can shortly reach the limits of chip scalability

Unlike AMD and Intel, Nvidia is mainly oriented toward the usage of monolithic chips. However, experts from Huawei are convinced that such an approach will soon become irrelevant when compared with multi-die configurations.   

Huawei's Long-Term Strategic Intentions

Such a warning relates not only to Nvidia but to most Western developers, according to Huawei Technologies. Huawei is planning to rely on the “tau scaling” law, which is based on a more innovative approach that does not require the use of advanced lithography. The primary explanation for this choice may be related to U.S. sanctions and the lack of access to advanced Western equipment resulting from this political background.

Liao Heng is criticizing the approach of Western partners and their scaling, especially when talking about the physical limits. Whereas Huawei is trying to accelerate data transfer rather than the physical scaling of components.

Geopolitical tensions are already impacting the development of independent semiconductor industries. Despite the obvious limitations, Chinese experts demonstrated their capability of creating AI models even under resource-constrained conditions.  

Hope you found this article helpful - what do you think? Like and subscribe to our blog to get more practical tips and the latest tech news from HostZealot.