Salesforce Is Paying Anthropic $300 Million While Anthropic Hires Away Its Sales Team. Marc Benioff Calls Anthropic 'Awesome.'
Anthropic and OpenAI have hired ~100 Salesforce people — ~45 and ~40 — almost all enterprise sales, solutions engineering, and customer success, the relationship-holders who know the Fortune 500. Meanwhile Salesforce's CEO says he'll 'probably use $300 million of Anthropic' this year. So Salesforce is at once Anthropic's customer, its talent farm, and its competitor — and seems fine with all three. The unstable part isn't the poaching; it's the coopetition bind. The skeptic's question: do those Fortune 500 relationships actually travel, or stay with the platform contract?
Marc Benioff has a vendor he can't stop praising. Anthropic is "awesome," he says, and Salesforce will "probably use $300 million" of it this year. Anthropic has also hired roughly 45 of Salesforce's own people lately — enterprise salespeople, solutions engineers, customer-success staff, the relationship-holders who know the Fortune 500 buyers Salesforce spent two decades cultivating. OpenAI took another 40 or so. So Salesforce is, simultaneously, Anthropic's paying customer, Anthropic's talent farm, and Anthropic's competitor. And the CEO is cheering. This is either three-dimensional chess or a man complimenting the tide as it carries off his furniture.
The framing everyone reached for — "AI labs poach Salesforce talent" — undersells what's actually strange here. That two cash-rich companies heading for IPOs are hiring enterprise salespeople is not news; of course they are. The interesting part is that the company they're hiring from is bankrolling them and competing with them at the same time, and appears untroubled by any of the three facts. That's not a hiring story. It's a case study in what coopetition looks like when one side has the leverage and the other has the checkbook.
The substance underneath it is that OpenAI and Anthropic have reached the enterprise-revenue phase, where go-to-market execution matters as much as model quality. You can have the best model on the leaderboard and still lose the Fortune 500 deal to whoever has the relationship with the CIO and the procurement team — and those relationships take years to build or about a quarter to hire. Buying roughly a hundred of Salesforce's relationship-rich people, concentrated entirely in sales and customer-facing roles rather than research, is the fast path to the enterprise-revenue growth both labs need to show before they list. The talent didn't go to the model teams. It went to the revenue story.
Which sharpens the Salesforce bind into something genuinely unstable. Salesforce sells Agentforce and Einstein into the same enterprise-AI budgets Anthropic and OpenAI are chasing — they are competitors. Salesforce is also, by Benioff's own mouth, about to spend a third of a billion dollars consuming Claude — it's a customer. And it's watching its sales bench walk to the company it's paying. A firm can survive being a supplier to a rival, or a customer of one, but being talent feeder, customer, and competitor to the same company at once is a position that resolves, not a position that holds.
Now the part the breathless version skips, because it matters to whether any of this is as dramatic as it sounds. "Anthropic is dismantling Salesforce's sales org" overstates it: a hundred people across two buyers, over a timeframe the sources can't even agree on — "this year" in some accounts, eighteen months in others — is a meaningful go-to-market build, not a teardown of a company with tens of thousands of sellers. The $300 million is Benioff's "probably use," a spending intention, not a disclosed contract with a floor. And the load-bearing assumption — that these hires bring their Fortune 500 relationships with them — is exactly the thing that usually isn't true. Enterprise deals are tied to multi-year platform contracts that survive the individual salesperson's exit; the rep leaves, the renewal stays with Salesforce. If the relationships don't travel, the labs bought credibility and a playbook, not a book of business, and the headline is louder than the revenue.
Here's where it goes, and Benioff's cheerfulness is the tell. The tripartite bind breaks one of two ways: Salesforce curbs the Anthropic spend and hardens retention to defend its turf, or it concedes the frontier-model layer entirely and leans into being the application and distribution layer that sits on top of Claude and GPT — reseller, not rival. A CEO calling the company that's draining his sales team "awesome" while pledging it $300 million is not a man preparing to fight; he's a man who has quietly decided which layer of the stack he wants to own, and it isn't the model. If that read is right, the sales-talent loss isn't an accident Salesforce failed to stop. It's the cost of a strategy it hasn't announced. What would change my mind is simple and visible: Salesforce curbing the Anthropic spend, hardening its non-solicits, or otherwise acting like a company that intends to keep competing at the model layer. So far it's doing the opposite, with a smile.