Microsoft has scaled back its data center projects in the U.S. and Europe, abandoning plans that would have utilized 2 gigawatts of electricity in the last six months, according to analysts at TD Cowen. The move is attributed to an oversupply relative to the company’s current demand forecasts and shifting strategies within the artificial intelligence landscape.
The decision, reported on Wednesday, was largely driven by Microsoft’s choice not to support additional training workloads for OpenAI, the company behind ChatGPT. Analyst Michael Elias and his team at TD Cowen noted this shift in a recent memo.
Investor skepticism regarding the magnitude of AI spending by U.S. tech firms has risen recently. This comes alongside the emerging capabilities of Chinese startup DeepSeek, which demonstrated AI technology at a significantly lower cost than its Western counterparts. This change in the AI landscape has spurred Microsoft to reevaluate its infrastructure investments.
Supply chain assessments by TD Cowen reveal that Alphabet’s Google is stepping in to fill the capacity gap in international markets, while Meta Platforms is doing the same within the U.S. market. This creates fluidity as tech giants continue to innovate.
Despite the pullback, Microsoft maintains its commitment to expansion. The company stated that while it may “strategically pace or adjust our infrastructure in some areas, we will continue to grow strongly in all regions.” Microsoft also confirmed that its planned spending of $80 billion on AI infrastructure for this fiscal year remains on track.
TD Cowen analysts had previously indicated, in February, that Microsoft had scrapped leases for “a couple of hundred megawatts” of capacity with at least two private data center operators.
Earlier in the month, AI cloud startup CoreWeave stated it hadn’t experienced any contract cancellations, despite a Financial Times report that Microsoft, its largest customer, was reducing some agreements.
Following the DeepSeek reveal in January, executives from Microsoft and Meta reaffirmed their AI spending, citing it as key to maintaining competitiveness. Alphabet has committed to spending $75 billion on AI this year, an increase of 29% over Wall Street’s expectations, while Meta has pledged up to $65 billion.
