Microsoft extends data center lifespan to 25 years as Azure hits $33.2bn

Microsoft now expects its data centers to remain useful assets for up to 25 years, doubling the previous 15-year assumption and departing sharply from an industry average of 11.3 years.

The company disclosed the change on its Q1 2027 earnings call. CFO Amy Hood said the extension reflects Microsoft's "operating history and expected use of these assets." The revised depreciation schedule also lets Microsoft spread the cost of its multi-billion-dollar facilities across an additional decade of accounting life.

For comparison, Uptime Institute research published in 2024 found data centers typically operate for 10 to 15 years, with the average mean lifespan at 11.3 years. Microsoft is thought to have access to roughly 12GW of data center capacity today.

What did Azure disclose for the first time?

Microsoft revealed Azure's revenue figure for the first time: $33.2 billion for the quarter, a 48 percent year-on-year increase. That made Azure responsible for 36 percent of total company revenue in the three months to the end of September.

The wider Agents and Infra division, which houses Azure and Microsoft's AI tools, posted $76.8 billion in revenue, up 25 percent year-on-year. Microsoft introduced the division as a reporting segment in September, a move it framed as "full transparency of quarterly revenue across each of our key businesses."

How is Microsoft getting more from existing capacity?

CEO Satya Nadella told investors the company continues "to get more from the infrastructure we already have by optimizing across the data center design, silicon systems, software, and models."

He cited one concrete result: "We increased the throughput for Copilot workloads by over 50 percent quarter-over-quarter. We continue to scale our fleet with the latest innovations from Nvidia, AMD, alongside our own Maia systems."

The quarter also saw 24 new data centers come online across five continents. Nadella said Microsoft is reducing dock-to-live times by about 20 percent quarter-over-quarter for new GB300 systems in its biggest regions, which account for the majority of GPU deployments.

He described the strategy as "building against a growing book of business with a fungible fleet that allows us to dynamically allocate capacity to the highest value opportunities across compute, data, Foundry, and Copilot."

Why did capex come in below expectations?

Capex for the quarter totaled $45.2 billion, below the expected $50.7 billion. Microsoft attributed the shortfall to delays among neocloud partners that failed to deliver contracted data center capacity on time.

As a result, the company cut its annual capex estimate for 2027 to $170 billion, down from the previous $175 billion forecast. Microsoft has signed deals with a range of neocloud providers, including CoreWeave, Nscale, Lambda, Iren, and Nebius.

What does the 25-year assumption signal?

Combining longer asset lifespans with throughput gains and a fungible fleet points to a strategy of extracting more value from capital already deployed, rather than simply building faster. If Microsoft's facilities genuinely last 25 years against an industry norm of 11.3, the company holds a structural cost advantage over rivals depreciating plants twice as fast.

Longer books, faster dock-to-live times, and a $5 billion capex cut landed in the same quarter.