Microsoft launched its latest chip Maia 200 as an update to the Maia 100 that was launched in 2023. It is a 3nm chip designed specifically for artificial intelligence (AI) inference tasks with a capacity of 10 petaflops (FP4) and 5 petaflops (FP8).
The inference chip is used to run models that have already been trained. As such, the Maia 200 is not a competitor to the Rubin series chips that NVIDIA just announced at CES earlier this month. Instead, the Maia 200 is a competitor to Google Tensor and Amazon Trainium used in cloud data centers.
Microsoft claims that the Maia 200 offers three times the FP4 performance of the third-generation Amazon Trainium and higher FP8 performance than Google's seventh-generation TPU.
Despite being just announced, the Maia 200 is already being used in their data center in Des Moines, Iowa to run Copilot 365, GPT-5.2 and by their AI Superintelligence team. This chip will then be used in a data center located in Phoenix, Arizona.
The Maia chip was developed in-house by Microsoft to reduce its dependence on NVIDIA, which in turn reduces the cost of developing data centers that require hundreds of billions.
