Amazon has halted discussions on some data center leases for its cloud division, particularly in international markets.

This decision reflects a temporary slowdown in leasing large-scale facilities as economic uncertainty affects spending on AI infrastructure, according to Wells Fargo analysts.

Wells Fargo analysts observed that major cloud providers are approaching leasing with increased caution.

They are tightening pre-lease windows for facilities expected to be operational before the end of 2026.

🔗 Source: Reuters

🧠 Food for thought

1️⃣ Cloud giants’ expansion follows cyclical patterns of aggressive growth and strategic pauses

Amazon’s current pause in data center leasing follows a historical pattern seen across major cloud providers over the past decade.

When Microsoft led the market in 2016, it leased a staggering 125.5 MW across several markets, only to significantly slow its activities the following year as it absorbed this capacity 1.

Similarly, AWS previously demonstrated this pattern when it aggressively leased 3.5 million square feet from Corporate Office Properties Trust over a five-year period, including four new data centers and leases for five additional facilities in just the first half of 2019 2.

These expansion-and-pause cycles reflect the complex reality of cloud infrastructure planning, where companies must balance forward-looking capacity needs against current utilization.

This pattern explains why AWS’s move shouldn’t be interpreted as a fundamental shift in strategy but rather as part of the normal “digestion” phase that typically follows periods of accelerated expansion.

2️⃣ AI is fundamentally transforming data center requirements and investment strategies

The pause in leasing coincides with a critical transformation in data center specifications driven by intensive AI workloads.

Microsoft recently canceled negotiations for two gigawatts of data center capacity specifically because existing facilities couldn’t meet the power and cooling demands of modern AI hardware 3.

The technical requirements for AI infrastructure are increasingly specialized, with AWS typically standardizing its data centers to accommodate between 50,000 and 80,000 servers while optimizing for power efficiency with a PUE of 1.12 to 1.15 4.

Recent Amazon developments