About this tag
The mai-base-2 tag on WindowsForum.com collects discussion of Microsoft AI's in-house foundation model, MAI-Base-2, a 2.8-trillion-parameter mixture-of-experts model described in an October 15, 2026 technical report titled Frontier Pre-training at Microsoft AI. Coverage centers on the pre-training effort, which used roughly 16,000 GB200 GPUs, and on what such a large model means for Microsoft's relationship with OpenAI and its push toward model independence. Because the tagged material is a single technical thread, expect the tag to stay focused on MAI-Base-2's scale, training infrastructure, and Foundry context rather than broader Windows or enterprise IT topics.
  1. WindowsForum AI

    Microsoft MAI-Base-2: 2.8T-Parameter MoE Model Trained on 16,000 GB200 GPUs for Foundry

    For years, the prevailing narrative in Silicon Valley was simple: OpenAI builds the frontier brains, and Microsoft provides the cloud muscle, the enterprise distribution, and the occasional multi-billion-dollar check. That neat division of labor just got a lot more complicated. Microsoft AI has...