Pershing Square Defends Big Tech AI Spending as Offensive Growth
In a Q3 shareholder letter, Bill Ackman's Pershing Square defended massive AI capital expenditures by tech giants, arguing that spending by Microsoft, Meta, Amazon, and Google drives offensive growth rather than defensive maintenance.
In a third-quarter shareholder letter released on November 2, 2026, Bill Ackman and his partners at Pershing Square presented an optimistic stance regarding artificial intelligence capital expenditures, asserting that cloud giants are engaging in offensive rather than defensive spending.
Market Skepticism vs. Growth Potential
Pershing Square noted that the contraction in P/E multiples among major cloud providers largely stems from investor anxiety over potential returns on soaring capital investments. Total capital expenditures for the four primary cloud giants are projected to reach approximately $740 billion in 2026, up nearly fivefold from roughly $140 billion in 2023.
"Rather than merely defending existing businesses, we believe the cloud giants are investing to accelerate growth, allocating capital to game-changing technology that integrates directly with their competitive advantages," the letter stated. These advantages include unmatched distribution, long-standing customer relationships, proprietary data and context, world-class security, and balance sheet resilience.
Infrastructure Providers Versus Cloud Giants
According to the letter, skepticism is not distributed evenly across the sector. While semiconductor and hardware infrastructure providers have surged—with S&P 500 sub-indexes climbing 44% and 36% year-to-date—the share prices of Microsoft, Meta, Amazon, and Google have advanced by an average of just 8%, despite a 27% increase in their earnings per share.
Pershing Square argues that investor enthusiasm for a relatively small group of hardware suppliers has caused the market to overlook established tech giants exhibiting robust operational performance.
Economics of Data Centers and Servers
Citing Amazon CEO Andy Jassy, the letter explained that data center structural shells require capital commitment roughly two years prior to utilization but boast a lifespan of 30 years or more with minimal ongoing maintenance costs.
Servers and networking equipment, conversely, are acquired just months before deployment, matching customer demand closely. These servers typically recoup their investment in under three years, boast a useful life of at least five to six years, and generate substantial free cash flow thereafter.
Monetization Across Microsoft and Meta
At Meta, core digital advertising operations are successfully leveraging AI to refine ad targeting, generate copy, and improve content recommendations, driving accelerated revenue growth from 22% in 2025 to an estimated 27% in 2026. Meanwhile, Microsoft reported that paid seats for Microsoft 365 Copilot doubled over the past six months to exceed 30 million.
Pershing Square estimates that establishing one gigawatt of AI data center capacity currently costs between $40 billion and $50 billion using cutting-edge chips, while generating annual revenues of $15 billion to $20 billion per gigawatt.
