Microsoft earnings reflect efficiency gains from faster GPU deployment | JP

Microsoft has reported an 18% increase in revenue for the quarter to 30 June 2026. Total revenue for the quarter was $90bn, with its Productivity and Business Processes business contributing $38bn. The Microsoft Intelligent Cloud business grew by 32% to $39bn.

Overall, Microsoft Cloud posted revenue of $59bn and grew by 27%, which, according to chief financial officer Amy Hood, reflects “strong demand across Azure and our first-party AI [artificial intelligence] applications and services”. For the full year, she said Microsoft Cloud revenue surpassed $214bn, with nearly 90% from customers outside of frontier model companies. The company’s PC business, More Personal Computing, declined by 4%.

Discussing the results, Microsoft CEO Satya Nadella said: “We are advancing the frontier on the cost-to-outcome curve, ensuring every customer can turn tokens into business results. This year, Azure revenue surpassed $100bn for the first time, and Microsoft 365 Copilot reached over 30 million paid seats, reflecting the confidence customers are placing in us to power their AI transformation.”

The company reported capital expenditure (capex) of $41bn, which it said included the impact from higher component pricing. “Roughly two-thirds of our capex was for short-lived assets, primarily CPUs [central processing units] and GPUs [graphics processing units], as customers increasingly build solutions that leverage both AI and non-AI infrastructure,” said Hood.

When asked about Microsoft’s growth strategy, Hood admitted there are still constraints: “For a number of quarters, demand continued to exceed available supply, and that certainly remains true.” 

What the company has tried to focus on is efficiency, which, according to Hood, means getting more out of everything Microsoft has deployed in its datacentre fleet. “That applies to efficiency gains in the CPU fleet and efficiency gains in the GPU fleet,” she said.

The efficiency gains had a positive impact on the company’s results during the quarter, according to Hood. “Because of the supply-demand imbalance, when we can make efficiency gains, they are quickly monetised in the quarter,” she said.

Microsoft also made process improvements to ensure that CPUs and GPUs are installed quickly when shipments arrive. Nadella said that over the past fiscal year, it has reduced dock-to-live times for new GPUs in its largest regions by nearly 50%, implying that the time between GPUs arriving in a country where an Azure datacentre region is located and being installed in a physical server is now much shorter.

This means there is potentially a quicker turnaround time from when the company pays for the processors and when they start contributing to its earnings. Given the scale of Microsoft’s datacentre operations, Hood said: “Making efficiency improvements that can be quickly monetised results in acceleration in the quarter.”

She added that the company was extending the estimated useful life of its datacentres and office buildings from 15 to 25 years, reflecting “operating history and expected use of these assets”.

One area of interest among enterprises is the use of open models over proprietary AI models like OpenAI. When asked about model choice, Nadella said organisations want to be in control of their own destiny, in terms of building their human capital and what he calls “token capital”. He added: “Every [organisation] is going to evaluate who are the providers who are helping them with their outcomes and their knowledge creation.”

Commenting on the results, Tracy Woo, principal analyst at Forrester, said: “Microsoft’s position heading into its earnings call reflects a real tension between extraordinary AI demand and the challenge of delivering on it. That dynamic is shaped by Microsoft’s dependence on OpenAI at a time when OpenAI is no longer required to sell its frontier models exclusively through Azure. Microsoft now relies on a partner that is in talks to run its models on its biggest competitor’s infrastructure – all while Azure remains capacity‑constrained and the company pushes to expand infrastructure fast enough to meet customer expectations.”