According to a new report, Microsoft already has its own AI chips to use in training its products, to help reduce reliance on Nvidia’s own chips.
The new AI arms race in tech giants like Google and Microsoft seems to be forcing companies to go to new frontiers. In the latest, according to a report from the information (paywall)as monitored the edge Microsoft is already working on its own AI chips that can be used to train its products, in an effort to reduce reliance on Nvidia’s own server chips.
This Microsoft AI chip, codenamed “Athena,” appears to have been in the works since 2019 and is already in use by select Microsoft and OpenAI employees, with a broader internal rollout likely next year. However, it is not clear whether the chips will remain for internal use only or if they will also be shipped to their customers who use Azure cloud products. There are also many future versions of the chip planned the information And the edge Note that the chips are not something that will directly replace Nvidia’s H100 GPUs.
A report from a few years ago in 2020 indicated that Microsoft was considering its own custom ARM-based chips for servers, but it’s not clear if this is related to Athena at all. Microsoft already has its own ARM-based SQ series of chips in its Surface devices, which it’s worked with Qualcomm on. I also worked with AMD on Ryzen CPUs for the Surface Laptop series. So, Microsoft clearly has experience in this chip area as it continues to push AI across its core products.
This move shouldn’t be too shocking for those who follow the industry. Google uses its own AI supercomputers with more than 4,000 TPU and custom parts that can run AI models. Full documentation was It was recently revealed online And this has been in use since 2020. Amazon too It has its own Trainium chips Which are used to train machine learning models in natural language processing. Meanwhile, Meta uses Nvidia and AI chips in its supercomputer, the Super Computer.