Salesforce is investing heavily in AI tokens, with CEO Marc Benioff confirming a $300 million spend this year. While other tech giants grapple with soaring token costs, Salesforce’s CMO Patrick Stokes anticipates a new wave of AI usage in knowledge work and stresses the company's focus on evolving pricing structures to fit this changing landscape.
- Salesforce commits $300 million to Anthropic tokens focusing on AI innovation.
- Token consumption is highest in coding but knowledge work is the next frontier.
- New pricing models under development aim to balance legacy seat-based licenses with emerging agent use cases.
What happened
Salesforce CEO Marc Benioff revealed the company’s plan to spend $300 million on AI tokens through its partnership with Anthropic in 2026. This significant investment sets Salesforce apart from other major enterprises like Microsoft and Uber, which have experienced unexpectedly large token expenses and are actively working to reduce their costs.
Patrick Stokes, Salesforce’s CMO, highlighted during the Mizuho Technology Conference that while much of the token use remains concentrated in coding, a growing portion is devoted to knowledge workers in marketing, sales, and forecasting. These users leverage tokenized AI to enhance decision-making and daily workflows, signaling a broadening of token applications beyond engineering.
Why it matters
The rapid rise in token usage across enterprises has put tokenomics, or the economic models underpinning token consumption and pricing, at the forefront of corporate strategies. Salesforce recognizes that current token spending patterns, notably in coding, are steep, with individual engineers potentially generating monthly token bills in the six-figure range.
To remain competitive and customer-friendly, Salesforce aims to pioneer new pricing approaches that reflect the diverse ways AI agents will be utilized. This is critical as the company transitions from a traditional 25-year seat-based pricing paradigm toward licensing structures accommodating autonomous AI agents, ensuring governance and proper billing for these new digital participants.
What to watch next
Salesforce’s forthcoming Headless 360 product exemplifies its intent to reduce friction in AI integration by enabling external AI agents to interact seamlessly with its platform. A key part of this innovation is how Salesforce will implement agent identification and licensing to maintain control and accountability over AI resource consumption.
In the coming months, Salesforce will engage customers and partners to refine its pricing models, balancing legacy licensing habits with fresh, forward-thinking frameworks. Observers should monitor how these new agent licenses, potentially similar to human user licenses, evolve and influence enterprise AI adoption and economics.