Microsoft · Chips and hardware · Data centers and energy
Microsoft unveils Maia 300, the AI chip it will use across its own products starting in 2027
On October 12, Microsoft unveiled Maia 300, the third generation of its in-house AI accelerator, designed for inference. Its compute chiplets are built on TSMC’s N2P process, and eight HBM4 stacks provide 288 GB of memory and 18 TB/s of bandwidth. Microsoft says a full rack of 96 chips will deliver up to 7x the token throughput of Maia 200, and it expects 40% more tokens per megawatt than leading GPU-based systems. Those are the company’s own projections; no independent testing has been published. Maia 300 begins production deployments in 2027, as Microsoft makes Maia the primary inference system across its products and services. Maia 200, introduced in January, already serves Microsoft Copilot.
Sources
- Microsoft (Official Microsoft Blog), “Microsoft to expand Azure Maia system to multi-gigawatt deployments”, Announcement: N2P, 8 HBM4 / 18 TB/s, 96-chip rack, 7x Maia 200, 40% per megawatt, 2027, (blogs.microsoft.com)
- Microsoft Azure Blog, “Deep dive into Azure Maia 300 architecture”, Architecture details: 288 GB HBM4 per chip, (azure.microsoft.com)
- Microsoft Source, “Microsoft’s latest custom chip: Maia 300”, Press page and image gallery (news.microsoft.com)
- UPI, “Microsoft announces new AI chip Azure Maia 300”, No third-party testing yet, (upi.com)
About this story
This story was posted on Instagram by @jarrus.tech on Oct. 13, 2026.
Spotted an error in this story? [email protected] · Instagram
Short link: thejarrus.com/en/maia-300
This story in Turkish: Microsoft, 2027’den itibaren kendi ürünlerinde kullanacağı Maia 300 yapay zeka çipini tanıttı
On the same topic
Intel unveils GPU D270, an AI card with up to 480 GB of memory
Intel’s inference-only GPU D270 carries 160 GB of LPDDR5X memory, with partner versions reaching 480 GB. No price or launch date yet.
Microsoft calls about 50 Copilot Studio employees in Israel to pre-dismissal hearings
About 50 employees were called to pre-dismissal hearings, Calcalist reported; around half are expected to stay in other roles.
Chinese AI infrastructure company Infinigence AI raises about $344 million and prepares for a Hong Kong IPO
Infinigence AI closed a 2.3 billion yuan Series C round, and Bloomberg reported a confidential Hong Kong IPO filing.