Navigating the Evolving Landscape of AI Investments
Future Trajectories of Cloud Spending
According to Baron Fung, a prominent analyst at Dell'Oro Group, following an unprecedented period of growth over the past three years, major US hyperscale cloud operators are now increasingly focusing on investment returns and the growing influence of depreciation costs on their profitability. While aggregate spending by the top five cloud computing enterprises is expected to approach $400 billion in 2025, some providers may reduce their outlays more substantially than others in the subsequent year. For instance, Meta Platforms' data center investment may remain elevated compared to that of Amazon, Microsoft, Alphabet, and Oracle, potentially benefiting Meta's AI cloud infrastructure suppliers.
Varying Forecasts for Major Players
Although estimates fluctuate depending on the companies included and the precise definitions of cloud infrastructure, a consistent trend toward reduced capital expenditure growth is evident across various analyses, as highlighted by Visible Alpha data. Incorporating Apple, Wall Street analysts foresee a deceleration in cloud capital spending growth for the top six firms from 54% in the current year to 19% in 2026, further slowing to 7% in 2027 and 5% in 2028. Goldman Sachs anticipates a slowdown to 26% growth in 2026 from 54% in 2025, while Morgan Stanley projects a decrease to 16% from 56%. Evercore ISI expects an 18% growth in 2026, down from 64% in 2025, and Dell'Oro estimates 21% growth next year, a notable drop from over 50% in 2025.
The Growing Influence of Depreciation
A recent report from Goldman Sachs emphasized that hundreds of billions of dollars in AI capital investment have consistently supported AI infrastructure equities. Specifically, public US AI hyperscalers such as Amazon, Google, Meta, and Microsoft have collectively invested $312 billion in capital expenditures over the last four quarters. However, the anticipated slowdown in capital expenditure growth presents a considerable risk to the valuation of these AI infrastructure stocks. Depreciation, an accounting principle, is becoming an increasingly critical factor for many technology giants. Cloud computing companies acquire new data centers, servers, storage devices, and networking equipment, which are classified as long-term assets. As the useful life of this AI infrastructure diminishes, depreciation expenses rise, directly affecting profit margins.
Market Reaction and Future Outlook
For some market observers, depreciation trends are a significant concern that investors in AI stocks must closely monitor. In the June quarter, both Microsoft and Google reported better-than-expected cloud revenue growth, while Amazon showed a lag. Praetorian Capital, in a blog post, argued that hyperscalers would need to generate ten times their current revenues from data centers merely to offset the annual depreciation costs of AI infrastructure. They contended that, at the current rate, a saturation point is imminent, as there simply isn't enough revenue potential to cover the existing capital expenditure. Short-seller Jim Chanos echoed these concerns in a recent social media post, critiquing CoreWeave's assessment of Nvidia GPU lifespan and its implications for data center economics and depreciation. Despite the emergence of low-cost AI model developers like DeepSeek, which initially caused a sell-off in AI stocks due to fears of reduced data center spending, the long-term outlook remains optimistic for key players. Nvidia, a bellwether in the AI sector, maintains a positive long-term perspective. Their Chief Financial Officer, Colette Kress, projected $3 trillion to $4 trillion in AI infrastructure spending by the end of the decade, driven by the increasing demands of reasoning agentic AI, global sovereign AI initiatives, enterprise AI adoption, and the advent of physical AI and robotics.