⚡️ Microsoft делает ставку на свои чипы: Maia станет основной платформой инференса компании С 2027 года компания масштабирует семейство Maia до развёртываний на несколько гигаватт, и Maia станет основной системой инференса для её собственных продуктов и сервисов. Заодно представили Maia 300. По заявлению Microsoft, новый чип выдаёт на 40% больше токенов на ватт, чем ведущие системы на GPU. Предыдущее поколение уже в работе. Maia 200 развёрнута в США, Европе и Австралии, и в Teams и Copilot она даёт до 1,7 раза больше токенов на ватт. При гигаваттных масштабах главная метрика уже токены на ватт, а не пиковые терафлопсы: от неё зависит, во сколько обходится каждый ответ Copilot. Свой чип под свой инференс снижает зависимость от Nvidia и даёт контроль над всем стеком, от железа до модели. Тот же путь раньше прошли Google с TPU и Amazon с Trainium. Ранее сообщалось, что Microsoft ищет у TSMC мощности примерно под 300 тысяч чипов Maia 300. При этом от GPU компания не отказывается и обещает по-прежнему ставить в свой парк лучшие решения со всего рынка. https://blogs.microsoft.com/blog/2026/10/12/microsoft-to-expand-azure-maia-system-to-multi-gigawatt-deployments/ @ai_machinelearning_big_data
From the original source:
Telegram — AI Machine Learning Big Data
Read full article at Telegram — AI Machine Learning Big DataThis material was compiled from public sources. If its publication affects your rights → Request removal
Microsoft is betting on its chips: Maia will become the company's main inference platform
⚡️ Microsoft is betting on its chips: Maia will become the company's main inference platform From 2027, the company will scale the Maia family to multi-gigawatt deployments, and Maia will become the main inference system for its own products and services. At the same time...
AI Summary & Key Takeaways
- ⚡️ Microsoft is betting on its chips: Maia will become the company's main inference platform From 2027, the company will scale the Maia family to multi-gigawatt deployments, and Maia will become the main inference system for its own products and services
- At the same time...
Microsoft is betting on its chips: Maia will become the company's main inference platform
Related stories
Anthropic pitches AI safety bill that may block a model if 50+ people could be killed
The proposed policy seems to offer protections against AI run amok but actually forbids federal agencies from blocking a model's release without the permission of a judge.
SpaceX's AI business exceeds investor expectations
SpaceX Reports AI Profit Growth, Beating Expectations.
Stop paying for AI, run everything locally (here’s how)
Big labs are not going to talk about this but continual learning the idea of improving the models for your [music] own use cases for your own workflows. Big labs have no incentive of providing users with that [music] because they're training their models for the normal…
Anthropic’s cloud partnerships are getting messy
Anthropic is reportedly struggling to secure enough compute from cloud providers that increasingly view it as a competitor, even as its infrastructure spending balloons ahead of a planned November IPO.


