Ackman Flags A 27% Earnings Gap
Bill Ackman argues that Wall Street is looking at only half of Big Tech’s AI spreadsheet. In Pershing Square Capital Management’s third-quarter 2026 shareholder letter, the hedge fund manager said analysts are mispricing the four largest AI hyperscalers: Amazon, Microsoft, Meta Platforms and Alphabet.
His complaint, as reported on 13 November 2026, is straightforward. Forecasts fully reflect the enormous cost of building AI infrastructure but give little credit to the revenue those investments could generate. The result is a valuation gap between what the companies earn and what their shares are doing.
Capex Bill Climbs From $140 Billion To $740 Billion
The numbers driving investor anxiety are hard to ignore. According to the letter, combined capital expenditures for the four companies are expected to rise from approximately $140 billion in 2023 to an estimated $740 billion in 2026. Pershing Square also pointed to projections suggesting the group’s capex may top $1 trillion in 2027.
Ackman described the spending surge as aggressive positioning rather than a reluctant cost of staying in the game. His phrase for it: “Offensive rather than defensive.”
Shares Lag As Chip Stocks Outrun Hyperscalers
The disconnect shows up clearly in performance data cited in the letter. Earnings at the hyperscalers have grown at around 27% on average, yet the group posted average gains of only about 8% year to date through early November 2026.
Meanwhile, the companies selling the picks and shovels have thrived. Semiconductor stocks rose 44% over the same period, and technology hardware gained 36%. That split suggests the market has rewarded the suppliers of AI infrastructure more than the operators paying for it.
Why Capex Is The Swing Factor In AI Economics
Ackman’s core thesis rests on the economics of AI data-center capacity. The letter estimates that each gigawatt of new capacity carries a defined cost and a potential revenue stream, and the gap between the two is where the mispricing sits.
If analysts are capitalizing the spending but discounting the future revenue, the shares look expensive on near-term numbers. If the revenue arrives on schedule, the same spending becomes the engine of the next earnings leg. The letter’s argument is that the second half of that equation is being ignored.
Pershing Square’s Broader Bet On Dominant Compounders
Speaking at the Invest 2026 Conference in an interview published on 13 November 2026, Ackman said markets are heading back toward “a little more sanity,” with more focus on cash flows and business quality and less on which stock will go up the most in the next week.
He said some AI momentum trades, including so-called neo clouds, have been cut in half, and he called that “sane.” Pershing Square’s bias, he said, is to own dominant long-term compounder businesses with a low probability of disruption in a highly disruptible world. He noted the firm has made about seven new investments in the last six months, which he said is a lot for Pershing Square and reflects companies getting “really cheap” that it had wanted to own for years.
What The 2027 Capex Line Would Confirm
The thesis hinges on whether hyperscaler earnings keep compounding at roughly 27% while the capex bill climbs toward $740 billion in 2026 and potentially past $1 trillion in 2027. Watch the four companies’ next quarterly reports for capex guidance and any commentary on AI revenue conversion.
A confirmed 2027 capex figure above $1 trillion without a matching acceleration in earnings would undercut Ackman’s argument. A quarter in which hyperscaler shares close the 44-point performance gap with semiconductor stocks would be the first real evidence the disconnect is closing.
Sources: rss_autopost