Ackman's Pershing Square says Wall Street estimates penalize hyperscalers for AI spending

What happened

Bill Ackman's Pershing Square said analyst estimates underestimate hyperscalers' growth potential, arguing consensus forecasts count high AI infrastructure spending while discounting its returns. It cited Meta, Microsoft and Amazon Web Services, and estimated one gigawatt of AI data center capacity costs about $40 billion to $50 billion.

Confirmed

Global impact / market context

If Pershing Square is right, today's spending builds future revenue that analyst estimates miss, so companies like Meta, Microsoft and Amazon could earn more than forecasts imply. That gap could change how investors value these stocks relative to current expectations.

Analyst inference

The debate centers on heavy capital spending, meaning money spent on long-lived assets like data centers. Consensus models see capital expenditures rising as a share of revenue while revenue growth slows, which pressures near-term profit and shapes how investors position in these stocks.

Analyst inference

What to watch

1. Pershing Square estimated one gigawatt of AI data center capacity using the latest chips costs roughly $40 billion to $50 billion, and projected $15 billion to $20 billion in annual revenue per gigawatt. Confirmed
2. Watch whether hyperscalers report actual revenue and operating income from new AI capacity in line with Pershing Square's estimates, which would test the firm's argument that consensus forecasts understate returns. Proposed
3. Watch whether capital spending as a share of revenue keeps rising while revenue growth slows, because that combination would pressure profit per sale and could weigh on investor positioning in these stocks. Analyst inference

Evidence

• seekingalpha.com