Salesforce plans to spend $300 million on Anthropic AI tokens this year while simultaneously freezing engineering hiring, CEO Marc Benioff said on the All-In podcast published May 15, a direct trade-off between AI spending and human headcount.
"Coding. Everything's going to be cheaper to make. These coding agents are awesome. Anthropic is awesome." The $300 million goes toward tokens, discrete text units that Anthropic's models process when completing tasks. AI companies track and bill based on token volume, making this effectively a consumption-based spend on Claude models for Salesforce's engineering workflows.
Salesforce announced in 2024 that it would freeze engineering hiring for the first time in 2025. AI tools had boosted engineering productivity by more than 30%. The company's roughly 15,000 engineers are shifting into supervisory roles overseeing AI-generated code rather than writing it from scratch. "We're not adding any more software engineers next year because we have increased the productivity this year with Agentforce."
The AI business unit, Agentforce, has crossed approximately $800 million in annual recurring revenue. AI now accounts for 30 to 50% of the company's overall workload. The spending doesn't stop with Anthropic. Salesforce engineering teams work alongside multiple AI tools including OpenAI Codex, Cursor AI coding tools, and its own Agentforce platform. The company also launched Headless 360, an API-first platform with 60-plus MCP tools giving agents like Claude Code direct access to Salesforce's enterprise stack.
He is also working on bringing AI-powered coding directly into Slack, the workplace messaging platform Salesforce bought for $27.7 billion in 2021. "You're going to see some cool stuff with Slack and code I'm not ready to talk about yet."
"But there's no question that we are in a new moment in coding."
Not every token will hit Anthropic's frontier models. Salesforce needs an "intermediate layer" to route simpler inputs to smaller, cheaper models while reserving Anthropic's Claude for complex tasks, a cost optimization strategy for a company burning through $300 million in AI compute annually. The AI ROI is already showing in headcount reductions. Last August, the company noted that AI agents let Salesforce cut support staff from 9,000 to 5,000. While engineering hiring is frozen, Salesforce is adding 1,000 to 2,000 sales employees to explain AI products to customers.
"We're not at that level yet of AI," acknowledging that human engineers remain necessary. But the spending direction is clear: $300 million to Anthropic, zero new engineering hires.
