Microsoft added 24 data centres across five continents in the September quarter, while reporting faster GPU deployment and higher Copilot workload throughput. The figures show how quickly the company is expanding AI capacity, and how much it is spending to keep that capacity productive.
Microsoft AI Watch analysis
What happened
The Hindu reports that CEO Satya Nadella said Microsoft brought the 24 facilities online in the September quarter. The company already had more than 500 data centres across 34 countries, though it did not disclose a new total after the additions.
Microsoft said the time to deploy NVIDIA GB300 GPUs in its largest regions fell by about 20% quarter-on-quarter. Nadella also said throughput for Copilot workloads rose by more than 50% over the same period. Microsoft attributed further efficiency gains to its GPU fleet and better use of existing CPU capacity.
The company spent $45.2 billion on capital expenditure in the quarter, about three-quarters of it on short-lived assets, primarily CPUs and GPUs. It expects capital expenditure to exceed $50 billion in the December quarter. Microsoft also cited its Maia systems: Maia 200 is delivering up to 1.7 times more tokens per watt across Teams and Copilot AI experiences, while Maia 300 is expected to deliver 40% more tokens per watt than leading GPU-based systems.
Why it matters
More data centres are only part of the story. Faster deployment and higher workload throughput suggest Microsoft is also trying to get more usable capacity from the infrastructure it already has. That matters to customers waiting for cloud and AI services, and to anyone trying to judge whether this enormous build-out can serve demand efficiently.
The spending is substantial and still rising. Microsoft’s figures describe its own reported results and performance claims, not an independent comparison of its systems with competitors. But the combination of new facilities, deployment improvements and rising Copilot throughput offers a clearer picture of what the investment is meant to achieve.
Our read
The headline number is 24 new data centres; the more revealing detail may be that Microsoft says it is bringing GPUs online faster and pushing more Copilot work through its fleet. Infrastructure is not glamorous, but a faster route from installed chip to useful service is the sort of detail customers may actually notice. The next test is whether those efficiency gains keep pace with spending above $50 billion in the coming quarter.
What to watch
- Whether Microsoft discloses an updated data-centre total or more detail on regional capacity.
- Whether GPU deployment times and Copilot throughput continue improving.
- How the next quarter’s capital expenditure compares with the company’s forecast.
- Whether Microsoft provides comparable performance figures for its Maia systems in wider use.
Discussion spark: Should Microsoft’s expanding AI infrastructure be judged mainly by how much capacity it adds, or by how much useful workload it says each new investment can deliver?
Sources and evidence
- Microsoft expands AI infrastructure, adds 24 data centres to its 500-plus global kitty in Q1 FY27 – The Hindu (29 October 2026, 16:27 UTC)
not affiliated with or endorsed by Microsoft