Microsoft has deployed the first batch of its self designed artificial intelligence chips in one of its data centers, marking a major step in its AI strategy. The company said more of the chips will be rolled out in the coming months as demand for AI computing continues to grow.
The new chip, known as Maia 200, is built to handle AI inference, the heavy computing work involved in running AI models at scale. Microsoft described the chip as a high performance solution and said its processing speed surpasses that of Amazon’s latest Trainium chips and Google’s most recent Tensor Processing Units.
The move places Microsoft alongside other major cloud companies that are developing their own AI chips. One key reason for this shift is the difficulty and high cost of accessing advanced chips from Nvidia, whose products remain in short supply globally with no immediate relief in sight.
Despite the breakthrough, Microsoft Chief Executive Officer, Satya Nadella, said the company will continue to buy chips from partners such as Nvidia and AMD. He explained that while Microsoft is innovating internally, it does not intend to rely solely on its own hardware, stressing the importance of collaboration and long term competitiveness.
The Maia 200 chip will be used by Microsoft’s Superintelligence team, which is responsible for developing the company’s most advanced AI models. The team is led by Mustafa Suleyman, former co founder of Google DeepMind, and is working on frontier models that could eventually reduce Microsoft’s dependence on external AI providers.
The chip will also power OpenAI’s models running on Microsoft’s Azure cloud platform. Commenting on the launch, Suleyman said his team would be the first to use Maia 200, describing the deployment as a major milestone. However, Microsoft acknowledged that access to cutting edge AI hardware remains a challenge across the industry, both for customers and internal teams.