Somewhere in Anthropic’s careers system, a recruiter typed “Mega Account Executive, Meta“ and hit publish. The role offered $380,000 to $450,000 in total compensation — putting it squarely at the high end of enterprise AI sales. Then Business Insider called to ask about it. The listing vanished. Both companies declined to comment. For a relationship reportedly worth hundreds of millions of dollars per month, neither side had a single word to offer.
That’s the whole incident. It’s also almost perfectly revealing.
Every other account executive role Anthropic posts covers a geography or vertical — EMEA, Southeast Asia public sector, that kind of thing. This one named a single company in the title. “Mega account,” for context, is enterprise sales shorthand for an exceptionally large, strategically critical customer. The job body text never mentioned Meta at all — just “win new business and drive revenue within a book of strategic digital native accounts.” Only the title gave it away.
Someone typed the quiet part loud. Then someone else noticed.
The Billion-Dollar Elephant in the Room
Meta publicly criticizes AI concentration while privately modeling billions in annual spend at Anthropic.
The financial reality underneath that deleted title is striking. According to NYT sources:
- Meta reportedly pays Anthropic hundreds of millions of dollars per month for access to its AI tools.
- Internal Meta planning documents modeled spending as high as $10 billion annually on Anthropic’s models — though that figure represents a ceiling projection, not a signed contract.
- Meta engineers reportedly used Claude Code, Anthropic’s AI coding assistant, widely earlier this year.
- Anthropic’s models also helped power development and testing of Hatch, Meta’s upcoming AI agent, even though the public version of Hatch is expected to run primarily on Meta’s own model.
Meta’s head of AI product, Nat Friedman, reportedly told employees that reducing Anthropic reliance could materially affect Anthropic’s revenue ahead of its IPO.
That IPO context matters. A lot.
Meanwhile, Mark Zuckerberg has made a habit of publicly criticizing AI concentration — specifically the outsized power of labs like Anthropic. It’s the corporate equivalent of a restaurant owner denouncing a supplier at a trade conference while quietly renewing the supply contract backstage. In June, Meta restricted internal Claude Code use and redirected engineers toward in-house tools. The spending, per available reporting, continued anyway.
This is what coopetition looks like at scale: your rival is also your biggest customer, and both sides are pretending that tension is manageable.
The deleted listing signals that Anthropic is betting Meta stays a mega customer even as Meta publicly works to reduce that dependence. Pulling the post after media questions only confirms how exposed that bet is — particularly for a company trying to tell a clean revenue story before going public.
