← Back to live feed1 story
Internal projections at Meta showed potential annual spending of as much as $10 billion on AI models from Anthropic. The company currently spends hundreds of millions of dollars a month on those tools to help power and test Hatch, an upcoming personal AI agent.
This reliance on a competitor follows a delay of Meta’s own frontier model, Watermelon, which is now scheduled for October after pretraining paused in July. Nat Friedman, Meta’s head of AI product, told staff that shifting to internal coding tools or those from OpenAI could reduce Anthropic's revenue before its public offering.
Sign in to suggest edits
Key sources
- SOURCE@firstsquawk“META PROJECTED COULD SPEND UP TO $10B/YEAR ON ANTHROPIC AI”x.com
- SUPPORT@wallstengine“Meta has become one of Anthropic’s largest customers”x.com
- SUPPORT@techmeme“Zuckerberg publicly criticized Anthropic”x.com
- SUPPORT@negligible_cap“if Meta uses only its own coding tools or ones from OpenAI, it could lower Anthropic’s revenue before its public offering”x.com
- SUPPORT@andrewcurran_“pretraining run of Watermelon was paused for unknown reasons back in July”x.com
- SUPPORT@stocksavvyshay“paying a rival to preserve product velocity which shows how expensive it is to fall behind at the frontier”x.com
- SOURCE@wallstengine“can "do anything you can do online,"”x.com
- SUPPORT@mtslive“Meta internally projected spending as much as $10 billion a year on Anthropic's models”x.com