Search
On January 26 (local time), Microsoft introduced its second-generation, in-house developed AI chip, the Maia 200. Leveraging TSMC's cutting-edge 3-nanometer process technology, this chip boasts an impressive integration of over 140 billion transistors. It delivers computational capabilities exceeding 10 PetaFLOPS at FP4 precision and surpassing 5 PetaFLOPS at FP8 precision, all while maintaining a thermal design power consumption below 750W. The Maia 200 has already been integrated into data centers in Iowa and is set to expand its presence to data centers in Phoenix. This deployment will provide crucial computational support for services like OpenAI GPT-5.2 and Microsoft 365 Copilot, offering a remarkable 30% increase in performance per dollar when compared to current hardware offerings.
-
C114 Communication Network -
Communication Home
