At Microsoft (MSFT)’s annual developer conference, Microsoft Ignite 2023, this week, CEO Satya Nadella unveiled the company’s first in-house artificial intelligence chip and a cloud-computing processor, marking the tech giant’s further vertical integration in A.I. and foray into an A.I. infrastructure market currently dominated by Nvidia and AMD.
“We are entering this exciting new phase of A.I. where we are not just talking about it as technology that’s new and interesting,” Nadella said during his opening keynote address. “But we are getting into the details of product making, deployment, safety, real productivity gain. All the real-world issues.”
He then showcased the Azure Maia 100 AI Accelerator, a custom A.I. chip similar to Nvidia (NVDA)’s A100 and H100 GPUs (graphics processing units) that are widely used to train large language models (LLMs) and generative image applications.
The Maia 100, whose name suggests there will be subsequent generations to come, will be used to first power Microsoft’s Bing and Office A.I. products and then be available to partners and customers next year, Nadella said. The chip is currently being tested on GPT 3.5 Turbo, the LLM behind ChatGPT, Bing A.I. workloads and GitHub Copilot, an A.I. app codeveloped by GitHub and OpenAI.
A major beneficiary of the Maia 100 would be OpenAI, which struggles to source enough Nvidia chips to power its increasingly large A.I. models and has a long-term partnership with Microsoft. “We were excited when Microsoft first shared their designs for the Maia chip, and we’ve worked together to refine and test it with our models,” OpenAI CEO Sam Altman said in a statement on Nov. 15. Today (Nov.17) OpenAI announced Altman will step down as CEO after a “deliberative” board review.
At Ignite 2023, Microsoft also announced the Azure Cobalt 100 CPU, a custom processor designed to power general cloud services on Microsoft Azure. The Cobalt 100 is currently being tested on workloads like Microsoft Teams and SQL server, the company said.
Microsoft, best known for its software, actually has a long history in silicon development, Rani Borkar, head of Azure hardware systems and infrastructure at Microsoft, told The Verge. “These efforts are built on that experience. In 2017, we began architecting the cloud hardware stack and we began on that journey putting us on track to build our new custom chips,” Borkar said.