Microsoft launches Maia 200, a custom AI chip designed to compete with Nvidia in hyperscale cloud environments.

Microsoft has unveiled the Maia 200, the second generation of its in-house artificial intelligence chip, along with new software tools designed to reduce reliance on Nvidia. The Maia 200 comes online this week in a data center in Iowa, with a second location planned in Arizona. This represents the evolution of the Maia chip that Microsoft originally introduced in 2023.

The announcement reflects a broader trend among major cloud computing firms competing with Nvidia. Microsoft, Alphabet's Google, and Amazon Web Services—some of Nvidia's biggest customers—are all producing their own chips that increasingly compete with Nvidia. Google has particularly garnered interest from major Nvidia customers such as Meta Platforms, which is working closely with Google to close one of the biggest software gaps between Google and Nvidia's AI chip offerings.

To support the Maia 200, Microsoft is offering a package of software tools, including Triton, an open-source software tool with major contributions from OpenAI. Triton takes on the same tasks as Cuda, the Nvidia software that many Wall Street analysts say is Nvidia's biggest competitive advantage.

The Maia 200 is manufactured by Taiwan Semiconductor Manufacturing Co using 3-nanometer chipmaking technology and will use high-bandwidth memory chips, though an older and slower generation than Nvidia's forthcoming chips. Microsoft has also incorporated a significant amount of SRAM, a type of memory that can provide speed advantages for chatbots and other AI systems when they field requests from a large number of users. This approach follows the strategy of competitors such as Cerebras Systems, which recently inked a $10 billion deal with OpenAI to supply computing power, and Groq, the startup that Nvidia licensed technology from in a reported $20 billion deal.