Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Microsoft’s vast cloud division posted slower growth than Wall Street forecast as the tech group struggled to keep pace with customer demand for artificial intelligence-related services.
The Seattle-based software group beat analysts’ expectations for revenue and net income in the quarter ending in December, but its cloud division — its biggest revenue driver, which includes its Azure cloud computing platform — narrowly missed expectations.
Despite Azure’s AI services revenue growing by 157 per cent year on year, chief financial officer Amy Hood said Microsoft expected the “ongoing impact” of capacity constraints to persist in 2025 as the company worked to “address execution challenges”. She said this would lift by the end of the fiscal year.
Microsoft’s overall cloud division notched up a 21 per cent rise in revenues from a year ago to $40.9bn, missing expectations in a Bloomberg survey of $41.1bn.
The disappointing results from the cloud unit sent shares sliding more than 4 per cent in after-hours trading in New York, leaving the group set to shed about $150bn in market value.
Group revenue for the quarter rose 12 per cent from the previous year to a record of $69.6bn. Net income was up 10 per cent to $24.1bn, ahead of the average estimate in a Bloomberg survey of $23.5bn.
Microsoft spent $22.6bn on capital expenditure in its second quarter, doubling from the previous year. Earlier this month, the company said it would spend roughly $80bn in this fiscal year ending on June 30 to build out the data centre infrastructure necessary to train AI models and deploy applications.
The bumper spending comes as concerns swirl on Wall Street over China’s DeepSeek, which claims it can perform AI tasks as well as big US companies such as Microsoft-backed OpenAI but at a fraction of the cost.
Chief executive Satya Nadella said DeepSeek had demonstrated “real innovation”, including matching the performance of OpenAI’s o1 model, which was touted for its ability to answer more advanced and scientific questions.
“We are going to see that all get commoditised and it’s going to get broadly used. And the big beneficiaries of any software cycle like that is the customer,” he said.
Nadella added one of the lessons from the transition to cloud computing — when companies outsource their computer processing to external servers — was that as the cost of resources fell, demand climbed. “Optimisation means it will be much more ubiquitous,” he added.
The software giant has been one of the main beneficiaries of the AI boom, with surging demand for its cloud services and enthusiasm about its $13bn partnership with OpenAI propelling it into a tiny club of companies with market values exceeding $3tn.
OpenAI’s models underpin Microsoft’s Copilot chat bot and its large language model is available to customers via Azure.
Last week, the group said it would change the structure of its deal with OpenAI to enable the start-up to use rivals’ cloud computing services. But it reserved the right of first refusal.
The move came as the start-up laid out plans with cloud provider Oracle and Japan’s SoftBank to build at least $100bn of AI infrastructure in the US as part of a project dubbed Stargate. Microsoft is only named as a technical partner for the project.
“Their success is our success because of all the other commercial arrangements,” Nadella said, referring to Microsoft’s profit sharing arrangement with OpenAI as well as application and intellectual property rights, before he cautioned the company was working to optimise costs. “You don’t want to buy too much of anything at one time.”
Microsoft posted a 67 per cent uplift in commercial bookings growth in the quarter largely driven by commitments from OpenAI, the company added.
