Salesforce Dents Profit Margins with Aggressive AI Adoption

Salesforce's aggressive adoption of AI technology, specifically its partnership with Anthropic and use of the Claude model, has put pressure on the company's profit margins. In a recent investor conference, Deputy CFO Mike Spencer revealed that the CRM giant's inability to improve its margin forecasts was due in part to the dollars spent on Claude tokens.

According to Spencer, Salesforce unleashed Claude in its R&D cycle about six months ago as a way to accelerate product development and break new ground. However, this decision also meant that the company had to cover some of the token spend, which contributed to its lower margin guidance for the year.

Now, Salesforce is shifting into 'refinement mode,' focusing on getting greater value from its AI spending by re-evaluating which models are used for specific tasks. For many tasks, the latest and greatest model may not be necessary, and the company can opt for second- or third-generation models instead.

Spencer noted that Salesforce is experimenting with different vendors, including OpenAI, Cursor, and Claude, each with its own cost structure. This move aligns with what the company sees in its customer base, which is also looking to optimize AI spending.