Recent analysis suggests that Amazon Web Services (AWS), a leading cloud provider, may be slowing down its data center expansion, prompting concerns among analysts. Jim Chanos, a well-known short-seller, highlighted these concerns on X, referencing an analyst note that indicated AWS is exercising caution regarding its data center plans. Wells Fargo analysts reported that AWS has paused discussions on new colocation data center deals, especially those located internationally. While the extent of this pause is not yet clear, analysts are apprehensive, noting that hyperscalers are becoming more selective with leasing large power clusters and are tightening pre-lease windows for capacity slated for delivery before the end of 2026.

TD Cowen analysts corroborated these findings with their own research, pointing to a reduction in colocation deals from Amazon in the US. They indicated that Amazon has withdrawn from certain colocation agreements and has chosen not to pursue expansion options. Colocation deals involve multiple companies sharing space within the same data center. The analysts also suggested that Amazon's AI ambitions in Europe may be slowing down, a trend they are closely monitoring.

Cowen's analysts have identified a broader cooling trend in the data center market compared to the rapid growth of previous years. They observed a decrease in the exuberance surrounding the outlook for hyperscale demand. This moderation is evident in several signs, including a slight decrease in data center demand in Europe, a reduction in the urgency among cloud companies to secure data center capacity, and a decrease in the number of large deals in the market.

The AI data center market experienced significant growth following the emergence of OpenAI's ChatGPT in late 2022. The current signs of moderation are relatively small compared to this larger trend; however, given the trillions of dollars invested in generative AI, any indication of slowing growth is causing concern.

These developments mirror similar actions taken by Microsoft, which also recently paused some data center projects. Wells Fargo analysts noted that AWS, like Microsoft, appears to be absorbing recent leasing activity. This does not imply canceled deals but rather a pullback from preliminary agreements such as Letters of Intent or Statements of Qualifications, which are common in the early stages of data center projects.

In response to these concerns, Kevin Miller, Vice President of Global Data Centers at AWS, clarified on LinkedIn that AWS continues to see strong demand for both Generative AI and foundational workloads. He explained that AWS serves thousands of global customers and must consider various solutions to provide the right capacity at the right time. Miller stated that some options may be too expensive or may not deliver capacity when needed, while others may require adjustments in location. He characterized these actions as routine capacity management, asserting that there have been no fundamental changes to AWS's expansion plans.

Amazon has not provided further comments to Business Insider regarding these matters.

Miller's statements suggest that the pause is a normal part of data center growth rather than a cause for alarm. Wells Fargo analysts noted that historically, these digestion periods, characterized by slower leasing activity or deferred builds, can last 6 to 12 months before a rebound. They cited Google's pullback from leasing in the latter half of 2024, followed by an aggressive return in early 2025, as an example.

Cowen analysts also suggested that Amazon's recent caution in colocation deals may be related to efforts to improve efficiency across its data center operations. They noted that AWS typically prefers building its own data centers over engaging in colocation deals. Other tech giants like Meta and Google are reportedly still actively pursuing new capacity.

In conclusion, while AWS seems to be pausing to reassess, the AI cloud race is ongoing. Analysts and investors will closely monitor whether this pause is a brief adjustment or a more significant shift in AI strategy.