Microsoft has stirred speculation once again by scrapping several planned data center projects across the United States and Europe. According to analysts at TD Cowen, as reported by Bloomberg, the abandoned facilities would have provided around 2 Gigawatts (GW) of capacity—significant when compared to a typical large data center, which operates between 20 to 100 Megawatts (MW), according to Dgtl Infra.
This move follows a similar decision just a month earlier when Microsoft canceled U.S. data center leases amounting to roughly 200 MW. TD Cowen analysts suggested that these project cancellations indicate an oversupply of data centers compared to Microsoft’s projected demand.
The news has fueled debate about whether this signals an impending burst in the AI investment bubble, strategic cost-cutting, or broader challenges within the AI sector. Bloomberg cited Alibaba chairman Joe Tsai, who recently pointed out signs of an AI infrastructure bubble, noting that many projects are being launched without confirmed demand. He also expressed concerns about the massive capital investments being discussed, referencing the $500 billion Stargate Project as a potential example of overestimation.
Some industry experts believe Microsoft’s decision is driven by the immense power and cooling needs of advanced AI hardware, prompting a reconsideration of facility design and location. Others suggest that Microsoft’s evolving relationship with OpenAI, into which it has invested $13 billion, may be influencing its cloud infrastructure strategy. OpenAI can now work with other cloud providers, reducing Microsoft’s obligations to meet all of its computing needs.
Additionally, the growing number of competitive large language models, particularly from China, could lead to price pressures and lower returns on investment, making large-scale AI infrastructure projects less financially viable.
Despite these cancellations, Microsoft reaffirmed its commitment to AI investments, maintaining its plan to spend $80 billion on data centers by June. However, the strong reaction to these decisions highlights uncertainty in the AI market and raises questions about its near-term stability.