Microsoft (MSFT) is expanding its data centers on an unprecedented scale, attempting to reverse the constraints that computing power shortages have placed on its cloud and AI businesses. The leap from 12 gigawatts to 38 gigawatts demonstrates both its determination to seize the high ground in AI infrastructure and reflects the difficult balance the entire industry faces between surging demand and construction cycles. Whether it can timely convert plans into usable computing power will determine Microsoft’s final position in this trillion-dollar race.
According to people familiar with the matter, by 2032, Microsoft’s global data center network capacity will exceed 38 gigawatts, compared with about 12 gigawatts currently. This figure will surpass the electricity consumption of New York State during peak demand periods, and it also rarely reveals the scale of computing power Microsoft expects to obtain after its massive investment in the data center sector.
Computing power shortages have become Microsoft’s biggest recent obstacle. The problem was exacerbated by the company’s decision around early 2025 to pause some data center development. The lack of available servers prompted some customers to move new business elsewhere, frustrating Microsoft salespeople eager to sell cloud and artificial intelligence services, and also dissatisfying investors who hope to see returns from massive AI investments. Microsoft’s capital expenditure in its most recent fiscal year reached $145 billion, and analysts expect this spending to continue growing in the coming years.
Competitors such as Google (GOOGL), owned by Alphabet Inc., are also working hard to keep up with demand and are investing heavily in an attempt to change the situation. In this race for computing power, four major participants—also including Amazon (AMZN) and Meta Platforms Inc. (META)—have collectively committed nearly $2.4 trillion in the coming years, mainly for data center equipment and leases.
A growing number of leaders within Microsoft say they hope the company is finally beginning to emerge from its computing power constraints. Chief Executive Satya Nadella said in July that the company is “bringing computing power online faster than ever before” and is finding ways to improve the efficiency of existing hardware, thereby increasing profitability.
The most concrete symbol of Microsoft’s optimism is a new data center cluster under construction in the Atlanta area. Cloud infrastructure executive Alistair Speirs said the project, called East US 3, should help alleviate computing power shortages in Virginia, Microsoft’s largest data center hub. According to people familiar with the matter, about 300 megawatts of data center power will come online in the region this year and expand to more than 1 gigawatt in the coming years. The people said that of Microsoft’s current 12 gigawatts of capacity, only about 2 gigawatts are concentrated in AI-specific chips. They said that by 2032, this proportion is expected to grow to about one-third of the 38 gigawatts of total capacity Microsoft plans to bring online.
The big question facing Microsoft is whether it can build enough data centers in time to meet accelerating demand for cloud and AI services. It is worth noting that some of the capacity the company is bringing online today was originally planned in the early days of the pandemic, when work-from-home demand pushed Microsoft to expand its network of server farms.
