Microsoft (MSFT) unveiled Maia 300, the next generation of its in-house AI chip, on Monday, October 12. Production deployments start in 2027, and Microsoft says it will scale the Maia family to multi-gigawatt deployments from there. Reuters also reported that Microsoft plans to deploy potentially more than one million of the chips next year to run its own AI services.
Maia 300 follows Maia 200, which launched in January and already runs Microsoft Copilot. The new chip is a much bigger bet.
The 40% Claim
Microsoft expects Maia 300 to deliver about 40% more tokens per megawatt than leading GPU-based systems, though it didn’t identify the comparison systems. That framing matters because power availability can limit how much AI computing a data center supports. More output from the same megawatt means more Copilot answers from the same building.
The deployment figures hang together. Each rack holds 96 chips and draws 167 kW, so one million chips works out to about 10,400 racks and roughly 1.7 gigawatts of power.
Why Microsoft Needs a Cheaper Chip
Microsoft’s cash spending on property and equipment reached $35.80 billion in the quarter ended June, more than double the $17.08 billion a year earlier. Over the past four quarters, that spending totaled $115.95 billion, up 80% from the prior four.
Gross margin is moving the other way. It fell from 69.05% in the quarter ended September 2025 to 67.20% in the quarter ended June, three straight quarterly declines. Spending doubled, and each dollar of revenue now keeps less gross profit.
What Maia Doesn’t Change Yet
The 40% figure is Microsoft’s own estimate with no named benchmark, and none of these chips run in production until 2027. Maia also serves Microsoft’s internal services, not Azure customers renting compute, so Microsoft stays a major buyer of Nvidia (NVDA) hardware for now. Microsoft is also playing catch-up: Amazon (AMZN) and Alphabet (GOOGL) already rent their own chips, Trainium and TPUs, to cloud customers.
The clearer winner is Taiwan Semiconductor Manufacturing (TSM), which builds Maia 300’s compute chiplets on its N2P process. TSMC gets paid whether Microsoft’s AI runs on Nvidia GPUs or its own silicon.
Conclusion
Microsoft has the right problem in its sights. Gross margin has slipped almost two points in three quarters while capex doubled, and Maia 300 goes straight at the cost of running AI on Microsoft’s own services. The chip won’t help the numbers before 2027, though, and the 40% claim hasn’t been tested outside Microsoft. Gross margin is worth tracking, but depreciation from all that capex could keep it under pressure even if the chip saves money. The clearer signal will be how much of its own AI work Microsoft says has moved onto Maia once the first racks go live in 2027.
Want to watch whether that margin turn actually comes? Pull up Microsoft’s capex and gross margin charts on TIKR for free and check them against each new quarter. Learn more here.
So what is Microsoft stock actually worth?
TIKR lets you forecast the future price of any stock in less than a minute. Just enter a few assumptions into TIKR’s valuation model and see what MSFT stock could be worth. Start from Wall Street consensus estimates, or adjust the inputs to reflect your own view of the business. It’s free to use.
Value Microsoft Corporation for free→
Disclaimer:
Please note that the articles on TIKR are not intended to serve as investment or financial advice from TIKR or our content team, nor are they recommendations to buy or sell any stocks. We create our content based on TIKR Terminal’s investment data and analysts’ estimates. Our analysis might not include recent company news or important updates. TIKR has no position in any of the stocks mentioned. Thank you for reading, and happy investing!
