Microsoft is making "significant investments" in training its own AI models

"We should have the capacity to build world class frontier models in house of all sizes, but we should be very pragmatic and use other models where we need to,"

"We're also going to be making significant investments in our own cluster, so today MAI-1-preview was only trained on 15,000 H100s, a tiny cluster in the grand scheme of things."

"looking forward to us building model capability, so that we can build model-forward products."

"partly powered by Anthropic models,"

Microsoft launched its first in-house AI models last month and is investing significantly in compute capacity to develop future frontier models. MAI-1-preview was trained on 15,000 H100 GPUs, characterized as a small cluster relative to planned expansions. Microsoft aims to build clusters six to ten times larger to train models comparable to Meta, Google, and xAI. Microsoft intends to have the capacity to build world-class frontier models of all sizes while pragmatically using external models when beneficial. Microsoft will support multiple models in products and plans to partly power Microsoft 365 Copilot with Anthropic models after superior performance in Excel and PowerPoint.

Read at The Verge