Salesforce Plans to Spend US$300m on Anthropic AI Tokens

In 2025, Salesforce said it planned to cease the hiring of software engineers as AI tools become increasingly more commonplace in software development.

CEO Marc Benioff said he expects the company to spend hundreds of millions on Anthropic tokens to increase the capabilities of its AI coding-tools.

While Marc said the company had no plans to hire more software engineers due to the productivity gains from AI in 2025, he now says coding agents and AI models are changing how engineers work, rather than eliminating them. 

This approach of AI improving efficiency rather than replacing human workers is growing traction throughout the tech industry, with large companies like Anthropic and NVIDIA agreeing that AI should serve as an enhancement to existing workflows, not a tool to upend them.

Speaking on the All-In podcast, Marc said Salesforce plans to spend US$300m on Anthropic tokens in 2026, with plans for most of the tokens to go toward coding-related work. 

He describes AI coding agents as “awesome” and says the investment would lower software development costs while increasing output.

Increasing productivity with AI integration

Marc first announced in 2024 that Salesforce planned to stop hiring software developers the following year, citing productivity gains of more than 30% from Agentforce and other AI technologies.

“We’re not adding any more software engineers next year because we have increased the productivity this year with Agentforce and with other AI technology that we’re using for engineering teams by more than 30%,” Marc said at the time in an earnings call. 

He also added that engineering teams had achieved higher velocity due to AI integration.

Despite this hiring freeze, Salesforce increased hiring in other company areas. Marc previously stated that the company planned to add between 1,000 and 2,000 salespeople to help explain AI products their business value to customers.

He later added that AI is yet to reach a stage where it can replace engineers completely and that Salesforce’s approximately 15,000 engineers already work alongside several AI tools, including Anthropic models, OpenAI Codex and Cursor.

“When they start to use these models, they’re now working not only with the AI, but agents to help them code – and they can even become somewhat supervisory over these agents. But still, those engineers are needed. The model still cannot operate autonomously,” Marc said.

“We’re not at that level yet of AI,” he added, while saying that Salesforce’s engineering organisation had become around 30% more productive.

A new horizon for business 

The plans for spending tokens with Anthropic comes as Salesforce expands its AI offerings across its product line, including Slack and Agentforce. 

Marc also suggested the company plans to develop systems to route AI requests between larger and smaller models based on complexity, potentially reducing operational costs.

In addition to its US$300m investment into Anthropic, Salesforce reportedly owns a 1% stake in the AI startup, equating to a share worth US$1bn.

Regarding the company’s product offerings, Agentforce – Salesforce’s AI-focused business – has reached $800 million in annual recurring revenue, while Slack has added several new AI capabilities powered by Anthropic’s Claude models.

In a company statement, Marc says agentic AI serves as “a new labour model, new productivity model and a new economic model”.

Salesforce reports that AI currently accounts for between 30 and 50% of its overall workload.

“Digital labour is a new horizon for business. How we architect our businesses and run our businesses and staff our businesses and think about our businesses will never be the same,” Marc adds.