What's Actually Happening
Anthropicand OpenAI have collectively poached roughly 100 Salesforce employees since early 2026 — 45+ to Anthropic, ~40 to OpenAI — and almost none of them are engineers. They're hiring account executives, sales leaders, and GTM operators with deep Fortune 500 relationships. Sales and revenue roles now represent approximately 20% of open headcount at both firms.
This is not a coincidence. AI labs have solved the hard technical problem. Now they're solving the harder commercial one: how to sell directly into large enterprises without going through a middleware layer like Salesforce, HubSpot, or any other CRM-anchored platform.
The Structural Conflict Your Vendor Won't Mention
Salesforce is spending $300M on Anthropic tokens in 2026 — while Anthropic uses that revenue to hire away the salespeople who built Salesforce's enterprise relationships and use those relationships to go direct. That's not irony, it's a business model stress test playing out in real time.
For any revenue organization whose GTM motion is anchored to a CRM platform, the implication is straightforward: the AI layer that currently lives inside your platform vendor's roadmap is being actively recruited out of that vendor and pointed at your account. The question isn't whether AI labs will have enterprise sales capacity. They're building it now, with the exact talent that knows how to navigate your procurement process.
What Operators Should Do With This
This doesn't mean rip out your CRM. It means get specific about where your current platform actually creates lock-in versus where it's just a layer of convenience that an AI-native alternative could undercut.
Three things worth stress-testing now:
Audit your integration depth. If your CRM is deeply embedded in your revenue workflows — forecasting logic, territory rules, comp calculations — you have real switching cost. If it's mostly a contact database with dashboards, you don't.
Map where AI capability is coming from. Is your AI roadmap dependent on your platform vendor shipping features, or have you built model-agnostic workflows that could swap providers? Vendor-dependent AI is the highest-risk position as labs move direct.
Watch the mid-market sales motion. Enterprise deals have long cycles and procurement inertia. The faster threat is in the mid-market, where an AI-native revenue tool with ex-Salesforce sellers and a lower price point could close before your renewal conversation starts.
The labs aren't just building better models. They're building the commercial infrastructure to sell those models directly into the accounts your current stack serves. That's worth a conversation with your RevOps lead before your next platform renewal.