Microsoft Corporation (NASDAQ:MSFT) showcased a pair of new chips at its Ignite event in Seattle on Wednesday, including one meant to decrease reliance on Nvidia Corporation (NASDAQ:NVDA) for AI GPUs.
The new Maia 100 artificial intelligence chip could rival the Nvidia graphics processing units currently dominating the market.
The other, a Cobalt 100 Arm-based chip, is designed for general computing and is more in line with Intel processors.
Microsoft is building increasingly robot chips to support its cloud market share, which ranked second behind only Amazon at the end of October, according to one estimate.
Notably, Microsoft and others in cloud computing buy servers containing their chips, which differs from chipmakers like Nvidia and Advanced Micro Devices, Inc. (NASDAQ:AMD).
The Redmond, Washington-based company is currently testing how Maia 100 handles the Bing search engine chatbot, known as Copilot, along with Microsoft-backed OpenAI’s GPT-3.5-Turbo.
The Cobalt CPU is being tested on Microsoft Teams and SQL servers, the company said. The plan is to make virtual machines available to customers for a variety of workloads starting next year.
The timeline for the Maia 100 has not been revealed.
Earlier this week, Nvidia unveiled the H200, a graphics processing unit designed for training large language models.
The H200 has 1.4x the memory bandwidth and 1.8x the memory capacity of the H100 chip, which OpenAI used to train CPT-4.
Contact Andrew Kessel at andrew.kessel@proactiveinvestors.com
Follow him on X @andrew_kessel