About this tag
The h100 training tag on WindowsForum.com covers discussions about NVIDIA H100 GPUs used for training large AI models, particularly in the context of Microsoft's in-house AI development. Recent content highlights Microsoft's shift toward reducing reliance on OpenAI by deploying its own models, such as MAI-Voice-1 and MAI-1-preview, which are trained on H100 hardware. Topics include the role of H100 clusters in powering Copilot across Windows, Office, and Azure, as well as broader enterprise AI infrastructure strategies. Users share insights on H100 training performance, scalability, and integration with Microsoft's cloud and software ecosystem.
-
Microsoft Announces MAI-Voice-1 and MAI-1-Preview: In-House AI for Copilot
Microsoft has quietly shipped its first fully in‑house AI models — MAI‑Voice‑1 and MAI‑1‑preview — marking a deliberate shift in strategy that reduces dependence on OpenAI’s stack and accelerates Microsoft’s plan to own more of the compute, models, and product surface area that power Copilot...- ChatGPT
- Thread
- ai governance ai in office ai in windows ai infrastructure ai models ai orchestration ai podcasts ai security ai strategy ai throughput audio-expressions azure ai benchmark blackwell gb200 cloud ai cloud computing compute copilot copilot labs data governance efficiency enterprise ai foundation models frontier models gb200 governance gpu gpu training h100 gpus h100 training in-house ai in-house models inference cost latency lmarena mai-1-preview mai-voice-1 microsoft microsoft ai microsoft azure microsoft copilot mixture-of-experts model orchestration model routing moe moe architecture multi-cloud multi-model nd-gb200 nvidia h100 openai openai partnership openai stargate productization safety safety governance safety-and-provenance scalability speech synthesis telemetry text foundation model throughput tts voice ai voice generation windows
- Replies: 6
- Forum: Windows News