Dario Amodei, CEO of Anthropic, said companies are currently capturing only 5–10 percent of the economic value available from artificial intelligence. He made the remark while speaking at the Dreamforce conference in San Francisco alongside Salesforce leadership; the Bloomberg reported the comments.
What the executives said
Amodei argued that the business impact of AI has only begun to materialize and that enterprise adoption lags the future vision many leaders have described. Marc Benioff, CEO of Salesforce, highlighted uneven AI adoption among customers and urged his company to do more to accelerate clients’ digital transitions.
The two firms have strengthened their relationship in recent months: about a month earlier they announced an expansion of their partnership to integrate Anthropic’s Claude language model into Salesforce products. Salesforce says roughly 7,000 of its employees already use Claude-connected tools.
Investment and market context
Salesforce holds a significant stake in Anthropic; the investment was estimated at about $5 billion in early June. On Wall Street, the deal was viewed by some as a kind of détente, since investors have worried for months that leading AI providers might launch products that siphon customers away from traditional software vendors.
Safety, calls to slow development and industry debate
Amodei’s comments came days after he published an essay warning about AI risks and suggesting a slowdown in the pace of development. His piece received support from some leaders of rival AI companies, and it has sparked active debate among policymakers and industry participants about how to avoid loss of control over the technology.
By contrast, Jensen Huang, CEO of Nvidia, told the Dreamforce audience he believes market forces will regulate AI safety without the need for new laws.
Patrick Stokes, president of Salesforce, said at a pre-conference press event that the company’s strategy focuses on customer-oriented adaptation of existing models. Stokes added that a slower development pace could be beneficial to developers, since there is still much to catch up on, and that their current task is to mature the infrastructure and capabilities surrounding AI models.
Why this matters
The executives’ remarks underline that, although generative AI is spreading quickly in business, its full economic benefits and deep integration into enterprise processes remain incomplete. Partnerships and financial commitments—such as the Salesforce–Anthropic collaboration—are important steps toward broader adoption, but development pace, safety risks and regulatory questions remain central issues.
An AI assistant contributed to preparing this article; the final content was edited and verified by our journalist.
