Amazon Web Services (AWS) has reportedly decided to pause some of its negotiations for international data center leases, hinting at a potential shift in its plans for expanding hyperscale infrastructure. This news, revealed by Wells Fargo analysts recently, raises concerns about major cloud providers being more cautious with their capital expenditures due to economic uncertainties and the changing landscape of generative AI technology.
The pause in negotiations, especially focused on international colocation discussions, seems to be following a similar trend seen with Microsoft. Just a week prior, Microsoft also announced delays or halts in some of its early-stage data center construction projects. Both companies have been increasing their investments in recent times to meet the growing demands for AI, particularly for power-intensive NVIDIA chips and large-scale compute infrastructure. However, with the surge in AI investments and economic challenges, a more cautious approach is emerging.
According to Wells Fargo analysts, AWS is not canceling any agreements that have already been signed but is rather reassessing some of the aggressive leasing activities from the past few months. They described this move as a way for the company to digest previously committed deals instead of completely pulling back. It is worth noting that major hyperscale operators, including AWS and Microsoft, are now being more selective in securing power clusters and are tightening their lease windows for facilities expected to be delivered before the end of 2026.
While Amazon has downplayed the significance of this report by calling it “routine capacity management,” it does contribute to a larger trend of caution among key players in cloud infrastructure. TD Cowen analysts previously disclosed that Microsoft had postponed several data center projects in the U.S. and Europe, totaling up to two gigawatts of power capacity, over the past six months due to a discrepancy between available capacity and projected demand.
This trend of adjusting build-out timelines reflects the growing tension between the need to lead in AI and the financial responsibility required by shareholders. Analysts believe that hyperscale companies are now being more strategic in how they deploy their capital, especially with increased competition from new players like Chinese firm DeepSeek, which offers AI capabilities at a lower cost compared to Western companies.
Despite this cautious approach by some hyperscalers, not all are slowing down. Meta, Alphabet (Google), and Oracle are reportedly actively pursuing new data center leases. Amazon CEO Andy Jassy has also defended the company’s ongoing AI investments, highlighting their strategic importance in maintaining AWS’s market position.
These developments highlight the challenging balancing act that cloud leaders face as they strive to meet the future infrastructure needs for AI while navigating a complex and volatile global economy.
