Intel and AMD shares fell sharply after Nvidia announced its entry into processor manufacturers, revealing a high-performance CPU for data center applications.
Although it does not have the necessary licenses to manufacture CPUs capable of natively running X86 applications for Windows PCs, Nvidia can compete without any restrictions in the ARM processor market. Known especially for its offer of GPUs, the American giant certainly has the expertise and financial resources needed to develop CPUs with very competitive performance. And in a real show of strength, Nvidia did just that, revealing Grace, a CPU for large server systems, which emphasizes the widespread use of AI technologies.
Developed with an architecture optimized for large-scale neural processing networks, Nvidia Grace is not the kind of CPU you would find in home PCs. In fact, the product is so specialized that Nvidia will probably only offer it in supercomputer configurations, where it will be responsible for coordinating NVIDIA P100 accelerator clusters, based on GPUs.
For now, Nvidia’s plan seems to be to exclude Xeon / EPYC processors from the configuration of AI-focused supercomputers, providing customers with a superior performance solution. But in the longer term, the new family of processors is likely to become a serious threat to AMD and Intel, with NVIDIA attracting very valuable customers from the market for server systems, data centers and large supercomputers.
Not really a surprise, the news of entering the competition with AMD and Intel comes less than a year after NVIDIA announced the start of the steps for the acquisition of the giant Arm Holdings, with the amount of $ 40 billion. Instead, Nvidia has come up with a vast collection of patented technologies in the field of ARM processing solutions. By adding its own expertise in the field of GPUs, NVIDIA can develop CPUs specialized in certain applications and usage scenarios, delivering performance that the competition will not be able to match too easily.