Microsoft has reportedly ceased plans to proceed with certain data centre projects in the US and Europe, amounting to 2 gigawatts (GW) of electricity, over the past six months. This decision aligns with an oversupply relative to Redmond’s current demand forecasts, as per a report in Reuters, citing analysts from TD Cowen. The pullout is largely influenced by Microsoft’s decision not to support further training workloads for OpenAI, the developers behind ChatGPT.

The tech industry is witnessing increased investor scepticism regarding significant AI expenditures by US companies. This sentiment is partly driven by slow returns and competition from Chinese startup DeepSeek, which has introduced cost-effective AI technology compared to its Western counterparts.

TD Cowen’s supply chain assessments reveal that Google’s parent company, Alphabet, and Meta Platforms are stepping in to fill the capacity gap left by Microsoft’s retreat. Alphabet is expanding its presence in international markets, while Meta focuses on the US market.

Microsoft’s strategic adjustments and AI investment plans

Despite a more than 1% drop in Microsoft’s shares on Wednesday, the company stated its commitment to strategically adjust its infrastructure while continuing robust growth across all regions. Microsoft plans to invest $80bn in AI infrastructure this fiscal year. According to Brad Smith, vice chair and president, more than half of this investment will occur in the US.

Last month, TD Cowen analysts claimed that Microsoft had scrapped leases amounting to “a couple of hundred megawatts” of capacity with two or more private data centre operators. Meanwhile, AI cloud startup CoreWeave confirmed it had not experienced any contract cancellations, despite a Financial Times report suggesting that Microsoft is moving away from certain agreements.

Microsoft and Meta executives have defended their substantial AI investments following DeepSeek’s emergence, emphasising their importance for maintaining competitiveness in the evolving AI landscape. Alphabet has announced a $75bn investment in AI development this year, exceeding Wall Street expectations by 29%, while Meta has committed up to $65bn.

Earlier this year, Microsoft CEO Satya Nadella revealed plans for a $3bn investment in India over the next two years, focusing on cloud and AI infrastructure and skills development, including new data centres.

Recently, OpenAI awarded an $11.9bn contract to CoreWeave to enhance its computing capabilities for ChatGPT. This move highlights changes in OpenAI’s relationship with Microsoft, which remains a key supporter through revenue-sharing arrangements. However, Microsoft is no longer OpenAI’s exclusive cloud provider following agreements with SoftBank and Oracle. Redmond continues developing its own AI models under the MAI brand, intensifying competition with OpenAI.

Read more: Microsoft set to invest $80bn in AI-powered data infrastructure for FY25