AI Infrastructure DatabaseThe chips, power and companies behind AI
Saved
Menu
Home How It's Built Watchlist The Stack
Hardware ChipsCompare Chips Servers & SystemsNetworking MemoryFoundries & Packaging Cooling
Facilities Data CentersPower ProjectsCloud & GPU Compute
Intelligence AI ModelsSoftware AI ProductsBenchmarks Research PapersDatasets Concepts
Ecosystem CompaniesStocks PeopleTimeline TrendsFuture of AI
Learn News Ask the Data AI Agents
About About us Methodology Data & API Disclaimer
Options
DATA & API

Free read-only JSON access to the site's cached data.

Dark mode

Guided View
New to chips, data centers and model jargon? We explain key terms as you browse, in plain English. Same data, with the help built in.

Expert View
You already know the stack. Just the data: clean, fast and compact, with no extra explanations. This is the default view.

Interface language
‹ Back
Home › Chips › Microsoft Maia 300

Microsoft Maia 300

Chip AI Accelerator Announced

Microsoft's third Maia generation: HBM4, TSMC N2P chiplets and 96-chip racks from 2027.

You're looking at a chip record: what this processor is, its key specs, who designs, manufactures and supplies its parts, and where it gets used.

Announced October 12, 2026, Maia 300 is the third generation of Microsoft's in-house accelerator and is planned as the primary inference system for Microsoft products. It uses a multi-chiplet design with compute chiplets on TSMC N2P and eight HBM4 stacks giving 288 GB at 18 TB/s, within a 950 W TDP. Racks hold 96 accelerators in 167 kW. Microsoft projects up to 7x the rack-level token throughput of Maia 200 and plans multi-gigawatt production deployments starting in 2027. Figures are company targets until it ships.

Specifications & key facts

Each figure is labeled by basis: vendor spec, vendor claim, reported, estimate, or official. "Not publicly disclosed" means exactly that: we never invent values.

Overview

Designer Microsoft
Announced October 12, 2026Official Microsoft blog
Status Announced: production deployments from 2027Official Microsoft blog, 2026-10-12

Manufacturing

Process TSMC N2P (compute chiplets)Vendor spec Microsoft blog, Oct 2026

Memory

Memory 288 GB HBM4 (8 stacks)Vendor spec Microsoft, Oct 2026
Memory bandwidth 18.0 TB/sVendor spec Microsoft blog, Oct 2026

Power

TDP 950 WVendor spec Microsoft blog, Oct 2026

Compute

Peak FLOPS Not publicly disclosed Not disclosed in Oct 2026 announcement; supports MXFP4

Physical

Rack 96 accelerators, 167 kW, 27.6 TB HBM, 1.7 PB/sVendor spec Microsoft blog, Oct 2026

Performance

Vendor targets Up to 7x rack token throughput vs Maia 200; 40% more tokens per MW than leading GPU systemsVendor claim Microsoft blog, Oct 2026

How this connects to the rest of the stack

This record's links at a glance: click a group to jump to it.

Designed by

Manufactured by

Process node

Memory technology

Successor to

Related

Sources

Microsoft: Azure Maia to multi-gigawatt deployments (Oct 2026) ↗ · Azure: Deep dive into Azure Maia 300 architecture ↗ · DCD: Microsoft unveils Azure Maia 300 (Oct 2026) ↗

Record reviewed as of 2026-11-16; see Methodology.