Nvidia’s Huang and Trump Downplay AI Safety Concerns as Salesforce Debuts Enterprise Reasoning Model

Nvidia CEO Jensen Huang received a live phone call from President Donald Trump during Monday’s All-In Summit in Los Angeles, prompting an onstage exchange about AI safety. The conversation followed recent warnings from Anthropic CEO Dario Amodei calling for a slower pace of AI capability development, a position also echoed by OpenAI’s Sam Altman and SpaceX’s Elon Musk. Trump dismissed growing concerns about AI risks as a “hoax,” suggesting they served the interests of political opponents or China, while Huang expressed agreement. The exchange came despite Gallup polling showing widespread public opposition to data center construction, driven largely by environmental and cost-of-living concerns.

As well as this, Salesforce unveiled Koa, its first reasoning model, built in partnership with Nvidia using its open-weight Nemotron model and tailored specifically for sales, marketing, and customer-support tasks. Salesforce AI executive Jayesh Govindarajan said the company previously relied on frontier models like Claude or ChatGPT for complex reasoning tasks but chose to build its own model once a viable, transparent open-weight base became available. He noted that Koa was trained using synthetic data simulating customer interactions rather than actual customer data, aiming to reduce data exposure risk while lowering token costs. Nvidia’s Kari Ann Briski said the collaboration prioritized efficient, sovereign AI architecture.

Despite launching Koa, Salesforce also announced a new partnership with Anthropic, called Claudeforce, allowing customers to use Claude within Salesforce’s secured infrastructure, indicating the company is pursuing multiple AI strategies rather than moving away from frontier model providers entirely.