Updated
Updated · The New York Times · Aug 27
Meta Projected $10 Billion Annual Spend on Anthropic Models Despite Public Attacks
Updated
Updated · The New York Times · Aug 27

Meta Projected $10 Billion Annual Spend on Anthropic Models Despite Public Attacks

3 articles · Updated · The New York Times · Aug 27

Summary

  • $10 billion a year was the internal spending level Meta at one point projected for Anthropic’s A.I. models, according to people familiar with the matter.
  • Meta had become such a heavy user of Anthropic’s products that it ranked among the startup’s largest customers, revealing a deeper commercial tie than previously known.
  • That projected outlay would have represented a major share of Anthropic’s business, which the company estimated in July would exceed $65 billion in annual revenue.
  • The relationship undercuts Mark Zuckerberg’s recent public criticism of leading A.I. labs, including rhetoric widely seen as aimed at Anthropic and CEO Dario Amodei.
  • The disclosure highlights Silicon Valley’s A.I. dynamic, where direct competitors can publicly clash while privately relying on each other for critical technology.

Insights

Why is Mark Zuckerberg secretly funneling billions into Anthropic's AI models while publicly slamming the startup?
What hidden limitations in Meta's own AI are forcing the tech giant to rely so heavily on a fierce rival?
Could Meta's quiet pivot to leasing AI compute ultimately make it the deadliest threat to traditional cloud giants?