Bill Ackman sold a quarter of his Amazon position in the second quarter while adding to Microsoft and Meta Platforms, rotating a concentrated portfolio of fewer than fifteen names toward what the billionaire views as stronger artificial intelligence contenders. The moves show a manager willing to shrink a winner to fund bets on companies he believes have clearer AI monetization paths.

The Amazon trim

Amazon remained one of Ackman’s largest holdings after the reduction, and the source notes no signal on whether selling will continue. The cloud business is targeting a path to one trillion dollars in revenue and sees a two-to-three-year payback on AI server and networking spend, a timeline the company expects to shorten as it leans more on its own chips. E-commerce growth and a high-margin sponsored-ad business that has become one of the world’s largest digital-ad platforms remain intact.

The Microsoft addition

Ackman increased his Microsoft stake before the stock surged in late July on strong fiscal fourth-quarter results, though shares have been roughly flat over the past year and year to date. The company’s 27 percent ownership stake in OpenAI comes with massive cloud commitments over the next several years. Custom silicon and proprietary model development lag rivals, but the OpenAI relationship buys time to close the gap. Enterprise software growth, led by Copilot adoption, has shifted the narrative from AI laggard to software winner.

The Meta bet

Ackman also built his Meta position in the second quarter, citing the platform as an AI flywheel. In the same quarter, ad impressions rose 14 percent and ad prices jumped 12 percent as AI improved content recommendations and advertiser tools. The new Muse agent aims to make Meta the leader in consumer agentic AI, a bet that could extend the flywheel beyond the feed.

What to watch

The portfolio remains highly concentrated, so any further rotation will move prices. Whether Ackman continues to fund the Microsoft and Meta additions by trimming Amazon, or lets the current weights ride, will signal his conviction on the relative speed of AI payback across cloud, enterprise software, and consumer advertising.