4 ms·
I would assume instead of Google and Microsoft designing their own ARM Processor on the server using ARM's IP with custom design. Nvidia could do that for them.
by ksec 3y ago
I would assume instead of Google and Microsoft designing their own ARM Processor on the server using ARM's IP with custom design. Nvidia could do that for them. After all Microsoft and Google dont have the economy of scale, nor do they have the ( or as much ) expertise. Nvidia could also provide other IPs such as Network, Ethernet and other GPGPU integration. I guess that is why they are in talks with Ericsson. Ericsson has a history of working with Intel and Intel failed to deliver.
Basically I think Nvidia will now move all the sunk cost of New Node to every new generation of AI chip. While the rest of the business unit, from GPU, Network, SoC, and now Custom Chips benefits from it. I am even wondering if Nvidia will come back to Smartphone Mobile SoC.