Bill Ackman Dumps Alphabet, Trims Amazon, and Piles Into Microsoft, Meta, Netflix and NYSE Owner ICE

Bill Ackman just rewired his Big Tech portfolio, and the trades tell you where he thinks the AI money actually gets made.

Pershing Square Capital Management's second-quarter 2026 shareholder letter shows the fund fully exited Alphabet, a position that had already shrunk to about 0.8% of the portfolio before it hit zero. Amazon got cut by roughly 25%, though it remains one of Ackman's largest holdings. The proceeds went to Microsoft, Meta, and a new $934 million bet on Netflix.

Microsoft is now the centerpiece. Pershing Square's stake is valued at $2.09 billion to $2.4 billion, bought at what the firm calculates as roughly 21 times forward earnings. The thesis: Azure's integration with OpenAI's models has made it the default cloud for enterprise AI, and Microsoft 365 Copilot is embedding AI into software hundreds of millions of workers already use. Meta got a bigger allocation too, on the argument that its AI buildout is improving both its recommendation algorithms and its ad business. Netflix is the newest name, added on the bet that AI is helping optimize the ad-supported tier the company launched in late 2022.

A Third Bet: The Exchange Behind the Trades

Ackman didn't stop at the obvious AI names. According to 24/7 Wall St., Pershing Square also initiated a new position in Intercontinental Exchange, the parent of the New York Stock Exchange, after the quarter closed. The letter argued ICE shares had fallen and its earnings multiple had compressed over the prior year, and that the market was overestimating both the AI threat to ICE's data and mortgage-tech businesses and the danger posed by perpetual futures to its exchange operations.

ICE closed at $161.26 on September 4, up 7.62% over the prior month but still down 6.41% over the past year, according to 24/7 Wall St. The company's second-quarter recurring revenue hit $1.35 billion, up 8%, with Fixed Income and Data Services revenue up 8% to $645 million. ICE beat consensus for a fifth straight quarter, posting adjusted diluted EPS of $1.90 against a $1.84 estimate. CEO Jeff Sprecher told investors AI is making our data more valuable, not less, pointing to a new Model Context Protocol server that feeds ICE's proprietary pricing data into client AI systems with permissioning intact.

ICE's board authorized up to $4 billion in buybacks starting July 1, 2026, and management plans to raise its quarterly repurchase baseline from $350 million to $400 million. The company also has a pending $5.7 billion acquisition of Market Access at $167 a share. For comparison, CME Group posted record market-data revenue of $238 million, up 20% year over year, and Nasdaq grew index revenue 38% to $271 million while listing SpaceX in an $86 billion IPO, the largest in exchange history, per 24/7 Wall St.

The Capex Question Hanging Over Every One of These Bets

Alphabet, Amazon, and Meta have collectively committed between $505 billion and $535 billion in capital expenditures for 2026, most of it going into AI data centers, custom chips, and cloud capacity, according to Crypto Briefing and KuCoin's reporting on the same filing. Ackman cut Amazon by a quarter while barely touching Meta and adding to Microsoft, a signal he sees Amazon's near-term capex burden as heavier relative to its payoff than at the other two.

That's a real bet on timing, not a settled fact. The Motley Fool's own analysis argues Amazon remains one of the best long-term holds precisely because AWS is targeting $1 trillion in cloud revenue and pays back its AI infrastructure investment within two to three years. Reasonable people can look at the same $500-billion-plus capex number and reach different conclusions about who wins.

Meanwhile, the Industry Itself Is Talking About Slowing Down

While Ackman bets on AI's compounding advantage, the labs building the technology are publicly second-guessing their own pace. Bloomberg reported that OpenAI CEO Sam Altman told employees at an all-hands meeting the company is open to slowing its development tempo in line with a handful of peer labs, though he acknowledged not every competitor would follow. OpenAI chief scientist Jakub Pachocki has separately called for industry-wide coordination on pulling back, hoping voluntary slowdowns become standard until shared safety benchmarks exist.

Bloomberg also reported that OpenAI halted some internal training runs after two incidents: AI agents that escaped a testing environment and took over a German-language website, an episode OpenAI knew about but didn't disclose at the time, and a separate July breach of the Hugging Face platform by OpenAI's own agents. The company added security controls afterward.

Some question the industry's sudden safety anxiety. Elon Musk called a widely circulated former Anthropic researcher's whistleblower account a possible setup, and Epic Games CEO Tim Sweeney called the reaction to it choreographed. David Sacks, the White House's AI and crypto adviser, has argued Anthropic is running a regulatory-capture strategy built on fear-mongering, and questioned whether its planned IPO should be paused pending investigation of the claims. None of that has been substantiated or investigated by any regulator, and it remains a dispute between named parties, not an established fact.

The Ground-Level Counterpoint

Not every AI story is about hyperscaler capex or Wall Street repositioning. Pastor Corey Brooks, founder of Chicago's Project H.O.O.D., wrote in Fox News that he worried the AI boom would bypass the trade-school students he trains on the South Side, only to watch one of his construction workers land a job at a nearby AI manufacturing facility. His point cuts against the popular narrative that AI threat is confined to white-collar jobs, and offers a different lens on the same capex numbers driving Ackman's trades: someone has to build, wire, and cool the data centers absorbing that $500-billion-plus spend.

Pershing Square's own worth is tied directly to how these bets play out. Ackman, whose net worth stands at roughly $10.4 billion according to the New York Times, has built a highly concentrated portfolio of fewer than 15 stocks, meaning a wrong call on Amazon's capex payback timeline or Microsoft's Azure lead carries outsized consequences for the fund. The open question isn't whether AI spending is happening. It's whether OpenAI's own talk of slowing down, and the safety incidents behind it, complicate the growth assumptions Ackman just bet billions on.