Salesforce to – Marc Benioff says Salesforce expects to spend $300 million on Anthropic tokens this year, betting that AI coding agents are boosting efficiency across service, support, distribution and marketing—and pushing new efforts to make coding easier inside Slack.
When Marc Benioff talks about AI coding agents, he doesn’t sound cautious—he sounds like someone ready to spend.
Speaking on the All-In podcast published on Friday, the Salesforce CEO said his company is projected to spend $300 million on tokens from Anthropic this year. Benioff described the tokens as “bits of information, like parts of a word,” that users type and an AI model processes.
“These coding agents are awesome. Anthropic is awesome,” Benioff said. “I am going to probably use $300 million of Anthropic (tokens) this year at Salesforce. Coding. Everything’s going to be cheaper to make.”
The money matters because tokens are how AI usage is measured and billed—companies track token usage to charge users, and large organizations to understand and manage compute costs.
Benioff says the payoff is already showing up inside Salesforce. He pointed to “unprecedented” levels of efficiency tied to AI agents, naming areas including service, support, distribution, and marketing.
He also connected the agent-driven push to a major staffing shift last year: in August, he announced that AI agents enabled Salesforce to cut its support staff from 9,000 to 5,000.
As part of the next push, Benioff said Salesforce is trying to make coding easier for employees inside Slack, the productivity platform Salesforce has owned since 2021.
“We’re even working on technology inside Slack to make it easier for everybody to code,” he said. “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.”
Underpinning the $300 million bet is a question of how companies decide where their AI spending goes. While Benioff expects a $300 million spend on Anthropic tokens this year, he said he doesn’t believe every token used by an employee has to be routed to a frontier model like Anthropic’s Claude.
In his view, the future should include an “intermediary layer”—a system that can decide which token inputs should go to Anthropic and which could be handled by smaller models, balancing performance with cost as AI coding agents spread.
For Salesforce, the direction is clear: agents are already reshaping how work gets done, and Benioff is signaling that AI consumption—especially for coding—will remain a top budget priority as the company extends the push through Slack.
Marc Benioff Salesforce Anthropic tokens AI coding agents Slack Claude productivity support staff