- Microsoft unveils Maia 200 AI hardware
- Maia 200 would offer more performance and efficiency than its competitors AWS and GCP
- Microsoft will use it to improve Copilot internally, but it will also be available to customers.
Microsoft has unveiled Maia 200, its “next major step” in supporting the next generation of AI and inference technologies.
The company’s new hardware, the successor to the Maia 100, will “radically change the economics of large-scale AI”, providing a significant improvement in performance and efficiency to gain a foothold in the market.
The launch will also aim to make Microsoft Azure a great place to run AI models faster and more efficiently, as it looks to take on larger rivals Amazon Web Services and Google Cloud.
Microsoft Maia 200
Microsoft claims that Maia 200 contains more than 100 billion transistors built on the TSMC 3nm process with native FP8/FP4 tensor cores, a redesigned memory system with 216 GB HBM3e at 7 TB/s and 272 MB of on-chip SRAM.
All of this contributes to the ability to deliver over 10 PFLOPS in 4-bit precision (FP4) and approximately 5 PFLOPS in 8-bit performance (FP8) – easily enough to run even today’s largest AI models, and with room to grow as technology evolves.
Microsoft says Maia 200 delivers 3x better FP4 performance than third-generation Amazon Trainium hardware and better FP8 performance than Google’s seventh-generation TPU, making it the company’s most efficient inference system to date.
And thanks to its optimized design, which sees the memory subsystem focused on narrow precision data types, a specialized DMA engine, on-chip SRAM and a specialized NoC fabric for high-bandwidth data movement, Maia 200 is able to keep more of a model’s weights and data locally, meaning fewer devices are needed to run a model.
Microsoft is already using the new hardware to power its AI workloads in Microsoft Foundry and Microsoft 365 Copilot, with wider customer availability to come.
It is also deploying Maia 200 in its Central US data center region, with further deployments coming soon to its US West 3 data center region, near Phoenix, Arizona, with other regions expected to follow.
For those interested in getting a sneak peek, Microsoft invites academics, developers, cutting-edge AI labs, and open source model project contributors to register now for a preview of the new Maia 200 software development kit (SDK).
Follow TechRadar on Google News And add us as your favorite source to get our news, reviews and expert opinions in your feeds. Make sure to click the Follow button!
And of course you can too follow TechRadar on TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp Also.
