Salesforce CEO Marc Benioff said his company will likely spend $300M on Anthropic tokens

"“These coding agents are awesome. Anthropic is awesome,” he 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.” Benioff said in the podcast that AI agents have brought “unprecedented” levels of efficiency at his company, pointing to areas such as service, support, distribution, and marketing."

"Last August, the CEO announced that AI agents enabled Salesforce to cut its support staff from 9,000 to 5,000. With the productivity boosts he's seen with coding agents, Benioff revealed in the podcast that Salesforce is also working on ways to make coding easier in 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.” Benioff said that while he projects a $300 million spend on Anthropic this year, he doesn't believe every token a company employee use has to go to a frontier model like Anthropic's Claude."

"The CEO called for an “intermediary layer” in the future that could determine which token inputs should go to Anthropic and which ones may be suitable for smaller models."

Salesforce is investing heavily in AI coding agents and expects to spend $300 million on Anthropic tokens in a year. Tokens represent small pieces of information processed by AI models, and token tracking supports billing and usage measurement. The company credits AI agents with unprecedented efficiency gains across service, support, distribution, and marketing. Salesforce previously reduced support staff from 9,000 to 5,000 after adopting AI agents. Salesforce is also developing technology inside Slack to make coding easier for users. Benioff expects a new era in coding and argues that not all token usage must go to frontier models, calling for an intermediary layer to route inputs to the most appropriate model size.

Read at www.businessinsider.com