Nvidia CEO Jensen Huang urges AI companies to stop scaring people with doomsday talk

Nvidia CEO Jensen Huang has come out publicly against apocalyptic rhetoric around AI. At the GTC 2026 conference, he said tech companies should warn about risks but should not turn the conversation about generative models into horror stories, especially amid the conflict between Anthropic and the Pentagon.

Why Huang stepped in

The trigger was a drawn-out dispute over how to use AI safely and ethically in the military sphere. The Anthropic-Pentagon conflict has made the topic too loud for one of the biggest beneficiaries of the AI boom to ignore.

Huang did not argue that warning society about the technology's capabilities is useful. But, he said, there is a clear line between discussing risks and deliberately stoking fear, and it matters that the industry does not cross it.

According to Huang, the problem is no longer just how people feel about individual models, but how fear affects the pace of AI adoption as a whole. He tied this directly to US national security: if the country fears the new technology more than it embraces it, it risks slowing itself down in the face of global competition. In this logic, the harm may come less from AI itself than from the panic surrounding it.

"It's not an alien.

It has no consciousness. It's a computer program."

Where the line falls

Huang is effectively proposing a more pragmatic vocabulary for talking about AI. He does not dismiss safety questions and does not mock those who discuss limits on generative models. His point is different: when AI company leaders describe the technology as something almost alive, uncontrollable and catastrophic, it distorts the real picture. In his view, extreme scenarios with no clear basis can harm society and the industry more than is commonly assumed.

• Warning about risks is fine
• Scaring people with apocalyptic scenarios is already harmful
• AI, according to Huang, is not a "biological" entity
• Today's systems have no consciousness or will of their own
• Catastrophic rhetoric can slow technology adoption

Notably, Huang did not attack Anthropic despite its public conflict with US authorities. On the contrary, he suggested the company would be fine financially even if the dispute with the defense department drags on. What's more, the Nvidia chief said Anthropic's revenue could exceed $1 trillion by 2030, and he considers Dario Amodei's forecasts on this front too cautious. To an outside observer, this sounds like criticism of the rhetoric, not a loss of faith in the business itself.

Taiwan and chips

Huang did not stop there and moved the conversation from AI ethics to geopolitics. He urged the US not to push China toward seizing Taiwan and to act with more restraint. For the Nvidia chief this is not an abstract issue: Taiwan remains a key node of the global semiconductor chain, so any escalation around the island hits not only regional security but the entire AI industry. In his framing, it is better to reduce tension than to create new triggers for a crisis.

At the same time, Huang argued that dependence on Taiwan should gradually decline by expanding advanced chip production in the US, Japan and South Korea. But this did not come across as a call to turn away from the island. On the contrary, he stressed that Taiwan's chipmaking expertise deserves friendship and support. In essence, his position boils down to two points: do not stoke fears about AI, and in parallel build more resilient infrastructure for the whole industry faster.

What this means

Huang's statement shows that the AI debate is moving beyond the familiar "accelerate or brake" standoff. Model safety, military contracts, US-China competition and the vulnerability of chip supply chains now sit in the same conversation. For the market, this is a signal: the tone of the AI discussion is becoming as important a factor as the models and hardware themselves.