Microsoft unveils Maia 300, its third AI chip, and says Maia 200 already runs Copilot
On 12 October 2026, Microsoft unveiled Maia 300, the third chip in its Azure Maia family, built to run AI models. Production deployments start in 2027. Microsoft says its earlier chip, Maia 200, already serves Microsoft Copilot, and that Maia will become the main system running AI across its products and services.
The short version
- 01Microsoft unveiled Maia 300 on 12 October 2026. Production deployments start in 2027.
- 02Maia 200, the chip before it, is serving Microsoft Copilot now.
- 03Microsoft expects 40% more tokens per megawatt than leading GPU-based systems.
- 04Those figures are Microsoft's own, and the chip is not in use yet.
At a glance
| Announced | 12 October 2026 |
|---|---|
| By | Microsoft |
| Chip | Azure Maia 300, third generation |
| Production deployments | From 2027 |
| Memory per chip | 288 GB, 18 TB/s of bandwidth |
| Chips per rack | 96 |
| Microsoft's efficiency claim | 40% more tokens per megawatt than leading GPU-based systems |
| Status | Confirmed |
What is Maia 300?
Maia 300 is a new AI chip designed by Microsoft, announced by Scott Guthrie, Microsoft's head of cloud and AI, in a post on the Official Microsoft Blog on 12 October 2026. It is the third generation of the Azure Maia family.
The chip is built for inference, which means running an AI model to answer questions, rather than training it. Microsoft says it is designed for long-context reasoning and multi-step agent tasks.
Is Maia already in use?
Yes, the earlier chip is. Microsoft says Maia 200 is now serving its production workloads, including Microsoft Copilot, with GitHub Copilot and Microsoft Foundry to follow by the end of the year.
Microsoft says Maia 200 runs OpenAI models, its own MAI models and leading open-source model families. It also says Maia 200 delivers up to 1.7x more tokens per watt for built-in AI features in Teams, Copilot and Cowork than leading GPU infrastructure. That is Microsoft's own figure.
What does Maia 300 add?
Microsoft says it expects Maia 300 to deliver 40% more tokens per megawatt than leading GPU-based systems. A token is a small piece of text that an AI reads or writes. In a full rack of 96 chips, Microsoft says it delivers up to 7x the throughput of Maia 200.
Each chip has eight HBM4 memory stacks, which give 288 GB of memory and 18 TB/s of bandwidth, within a 950-watt power limit. Its compute chiplets are built on TSMC's N2P process. A rack of 96 chips draws 167 kilowatts.
These are Microsoft's own figures and expectations. The chip is not in production yet.
When will Maia 300 be available?
Microsoft says Maia 300 will begin production deployments in 2027. It says it will scale the Maia family to multi-gigawatt deployments starting in 2027, making Maia the primary inference system for its products and services.
Microsoft did not say that outside customers can buy or rent Maia 300 chips directly.
Why does it matter?
Every answer from an AI assistant uses a chip somewhere. Microsoft is moving the AI in Copilot onto chips it designed itself, and plans to do far more of that from 2027.
What's next?
Microsoft says it will share more at its Ignite conference, which runs 17 to 20 November 2026. Watch for independent tests of the 40% claim once the chip is deployed.
Questions people ask
What did Microsoft announce on 12 October 2026?
It unveiled Maia 300, its third-generation Azure Maia AI chip, with production deployments starting in 2027.
Does Copilot already run on a Microsoft chip?
Microsoft says Maia 200, the previous generation, is now serving production workloads including Microsoft Copilot.
How much better is Maia 300?
Microsoft expects 40% more tokens per megawatt than leading GPU-based systems, and up to 7x the rack throughput of Maia 200. Both are Microsoft's own figures.
What is Maia 300 made on?
Its compute chiplets use TSMC's N2P process, with eight HBM4 memory stacks per chip.
When can people use Maia 300?
Production deployments begin in 2027. Microsoft has not given a date for anything earlier.
Sources
- PrimaryOfficial Microsoft Blog: Microsoft to expand Azure Maia system to multi-gigawatt deployments (12 Oct 2026)
- PrimaryAzure Blog: Deep dive into Azure Maia 300 architecture (12 Oct 2026)
Why "Confirmed"? We label a story "Confirmed" when the company or government announced it itself. Spotted an error? Email hello@openloopnews.com and we will correct it.
