Science & Tech · AI · 1 hr ago
Microsoft plans to deploy Maia 300 AI chip from 2027
Microsoft has announced Maia 300, the third generation of its in-house AI chip.
The company plans to start using it in production in 2027 and expand the Maia family at gigawatt scale.
Microsoft says Maia 300 is designed for tasks such as long-context AI, fast conversations and multi-step agent work.
It expects the chip to process 40% more tokens per megawatt than major GPU-based systems.
In a rack of 96 chips, Microsoft says throughput will be up to seven times that of the previous Maia 200 generation.
Maia 200 is already running in production, supporting workloads including Microsoft Copilot and the company’s AI model development.
Microsoft plans to give more details at its Ignite event in San Francisco from November 17 to 20, 2026.
Microsoft announced its third-generation Maia AI accelerator, Maia 300, on October 12.
The company plans to begin deploying the chip in production in 2027.
Microsoft intends to expand the Maia family to several gigawatts from 2027 and use it mainly for inference across its products and services.
Microsoft says Maia 300 is expected to deliver 40% more tokens per megawatt than major GPU-based systems.
A 96-accelerator rack is expected to provide up to seven times the token throughput of the previous-generation Maia 200.
- Who
- Microsoft announced the Maia 300 AI accelerator.
- What
- The company plans to start deploying the chip in production and expand its Maia family from 2027.
- When
- Microsoft announced it on October 12, 2026; production deployment is planned for 2027.
- Where
- Not stated
- Why
- Microsoft designed Maia 300 for newer AI workloads, including long-context inference, latency-sensitive interactions and multi-step agent tasks.
This story does not have two clearly opposing sides.
Microsoft announced the Maia 200, which is now operating in production.
Microsoft announced its third-generation Maia 300 accelerator.
Microsoft plans to begin deploying Maia 300 in production and expand the Maia family to several gigawatts.
Microsoft plans to discuss Maia 300 at Microsoft Ignite 2026 in San Francisco.
- Product
- Maia 300, Microsoft’s third-generation AI accelerator
- Production deployment
- Planned to begin in 2027
- Power efficiency
- Expected 40% more tokens per megawatt than major GPU-based systems
- Rack configuration
- 96 accelerators within a 167-kilowatt power limit
- HBM4 memory
- 288 GB per accelerator
- Process technology
- TSMC 2nm-generation N2P
What people are saying
自社シリコンへのこれまでで最大の賭け
Sources
Related news
Microsoft president calls for global AI safety rules
Meta launches Muse Wonder models to compete with leading AI systems
Meta unveils two AI models it says rival leading competitors
Scholar urges Global South to move beyond supplying AI raw materials
Meta launches new AI models, with faster development planned
Robotics competitions in Hangzhou to feature embodied AI challenges
Samsung outlines AI memory strategy beyond faster HBM
Anthropic AI submits false murder tip to Philadelphia police
Meta launches Muse models, with Muse Wonder ranked second in performance evaluation
Anthropic says new safeguards sharply reduce AI distillation attacks
Test taker caught trying to use AI smart glasses in Yonsei exam
