Microsoft is set to reveal its own AI chips in order to challenge NVIDIA's monopoly.

According to sources close to The Information, Microsoft is set to reveal details about its upcoming custom silicon design for accelerating AI workloads. The announcement of the new chip is expected to take place in November during Microsoft's annual Ignite conference, which will be held in Seattle from November 14 to 17. The conference will showcase Microsoft's advancements in the field of AI, with the launch of an AI chip being the highlight of the announcement. The demand for AI accelerators has been so high that companies are struggling to obtain GPUs, which are currently dominated by NVIDIA's H100 and A100 models that power most of the AI infrastructure worldwide.

With the introduction of a custom AI chip codenamed Athena, Microsoft aims to match or surpass the performance of NVIDIA's offerings while reducing the cost of AI infrastructure. The price of an H100 GPU can reach up to $30,000, making the construction of a data center filled with H100s cost hundreds of millions of dollars. By using their own chips, Microsoft can potentially lower these costs and become less reliant on NVIDIA for the AI servers needed in the future. We are eagerly anticipating the announcement at Microsoft Ignite in November and will provide updates on the event.