Microsoft forecasts show that the data center crunch continues in 2026

(Bloomberg) -Microsoft Corp. ‘s data center crunch will continue longer than the company previously set out, which emphasizes the struggles of the software giant to keep up with the demand for cloud. Many of Microsoft’s US data center regions experience shortages of physical space or servers, according to people who are familiar with the internal forecasts of the business. New subscriptions for Azure Cloud Services are limited in some important ministry farms, including North Virginia and Texas, through the first half of next year, said the people who requested anonymity to discuss internal predictions. This is a longer timeframe than the company previously set out. In July, Amy Hood, chief financial officer, said the current restrictions will continue at the end of 2025. The lack of capacity affects machines that use graphic processing units that are usually used for artificial intelligence, as well as data centers dominated by central processing units that have long been the work pair chips for traditional cloud services, people said. Azure is Microsoft’s most important growth engine – The cloud unit generated more than $ 75 billion during the financial year 2025. Its expansion has its biggest opponents, Amazon.com Inc. and Alphabet Inc. BOOGE surpasses. A lack of servers to rent to customers has been a recurring concern for cloud suppliers over the past few years. During the past six quarterly earnings calls, Microsoft said he was unable to meet all customer cloud question. Amazon and Google have described similar restrictions. A Microsoft spokeswoman said a majority of Azure services and regions in the US have “available capacity so that existing customers can grow with deployed workload.” In some cases of unplanned demand spikes, the company will introduce “capacity conservation methods” to balance the client’s demand in his data center fleet, the spokesman said. Azure customers choose data center regions based on physical proximity and which software is available. If a preferred facility does not have space, Microsoft sellers point clients to others with capacity, according to internal guidance. But these solutions can increase the complexity and the amount of time it takes to travel data between a server farm and the client, people who are familiar with the work said. In some cases, clients who experience Azure capacity issues take their business elsewhere, according to Apurva Kadakia, which helps businesses to set up cloud workload as global cloud head and partnerships at Hexaware Technologies. Some tap on several Azure regions or only send critical workload to the cloud until more capacity is available, he said. “Our teams work with large clients periodically to plan around demand, such as holiday periods, to lead them to the most suitable regions” and products, the Microsoft spokeswoman said. “In unusual cases where customers face increased cost or delay, Microsoft will compensate them for extra expense.” Microsoft was on a historic building to get data centers online – and has added more than two gigawatts capacity over the past year, approximately equivalent to the power delivery of the Hoover Dam. “It was almost impossible to build capacity fast enough since Chatgpt and GPT-4 was launched,” Chief Technology Officer Kevin Scott said in early October, referring to Openai’s popular chatbot and the AI ​​model it manages. “Even our most ambitious predictions just appear to be inadequate regularly.” For more information: Microsoft is $ 33 billion on Neoclouds such as Nebius to facilitate AI Crunch AI’s intense computer requirements, much of the need for new data centers fueled. But Microsoft is also staring at a crunch in demand for its traditional cloud infrastructure, which underlies the applications and websites across the Internet. For this CPU-based workload, Openai is the largest customer of Microsoft, according to people who are familiar with the company’s operations. Microsoft also uses significant computer sources to offer its own workload and applications such as the Office Suite. According to people who are familiar with the matter, some Microsoft employees are told to include internal projects in the impact regions to save capacity. It may take years to bring a data center online from initial planning to turn on the servers. Several important components for data centers, ranging from semiconductors themselves to electricity infrastructure such as transformers, are subject to long delivery times, according to people familiar with the process. The people who are known have made exceptions to key customers who want extra capacity in Azure regions with supply buttons. Availability is much better outside the US. For example, many of Microsoft’s European regions can handle new customer subscriptions without limitation. During an income in July, Finance’s chief Hood said that ongoing supply deficits are due to increasing demand. “I talked about it-Gosh-in January and said I thought we would be in a better supply question by June,” she said during the July call. “And now I say – I hope I’m in a better shape by December.” -With help from Matt Day and Dina Bass. More stories like these are available on Bloomberg.com © 2025 Bloomberg LP