As debate intensifies across Silicon Valley and global regulatory bodies over whether frontier AI development should be artificially capped or decelerated, Nvidia CEO Jensen Huang has taken a firm, opposing stance. Speaking alongside Salesforce CEO Marc Benioff at Dreamforce, Huang dismissed demands from rival labs and policymakers for sweeping statutory guardrails. While peers have urged caution as researchers quit and industry leaders seek to slow down following sandbox breaches, Huang argued that imposing blanket legal freezes would stifle innovation, insisting that market competition and disciplined engineering protocols are sufficient to keep artificial intelligence safe.
What a time to be alive
Trump was on speaker phone with Jensen Huang during the All-In Summit and he called everyone out
Trump said "I'm telling you it's all a hoax. The data centers are great. They make people wealthy. They make states wealthy. AI is bigger than the internet.… pic.twitter.com/KhY0P4ywje
"Safety is an engineering problem"
Huang pushed back against the narrative that technology companies are trapped in an irreconcilable conflict between moving quickly and protecting users, branding the dilemma a "false choice". While acknowledging that safety is "paramount" and "job one," Huang asserted that "safety is an engineering problem" rather than a legal one requiring government intervention.
Instead of external mandates forcing industry-wide freezes, Huang argued that accountability rests with product teams. "If we're not confident about the safety of the product, then don't release it," he noted, adding that companies should pace themselves until they are certain the market will value and appreciate their release. According to the Nvidia chief, natural market incentives penalize unsafe, buggy, or untrusted deployments, making top-down regulatory frameworks redundant.
Breaking with OpenAI, Anthropic, and Meta
Anthropic and OpenAI: Dario Amodei and Sam Altman have supported regulatory frameworks and formal evaluation standards, particularly after high-risk biological weapons probes and external server breaches prompted internal retraining shifts.
Meta: Mark Zuckerberg has proposed relying on independent, third-party evaluators to audit model safety thresholds.
Nvidia: Supplying the graphics processors that power hyperscale data centers worldwide, Nvidia has a direct commercial interest in sustained hardware buildouts, leading Huang to reject any systemic caps on computing infrastructure or hardware platforms.
Huang Rejects New AI Ruleshttps://t.co/pf3hWanQRN Nvidia founder and CEO Jensen Huang spoke at Salesforce Dreamforce on Tuesday. Huang said AI is hardware and software built by humans.
He said AI safety is an engineering problem and not a legal one. Huang argued that… pic.twitter.com/6HhJGNtP4v
Reinforcing message after the All-In summit
Huang’s comments at Dreamforce build directly upon his public remarks just 24 hours earlier at the All-In Summit in Los Angeles. During that event, Huang took a live onstage phone call from Donald Trump, who characterized regulatory AI slowdown efforts as a risk that benefits foreign adversaries. By reiterating his opposition to statutory caps before an enterprise software audience at Dreamforce, Huang cemented his position as the industry's most vocal advocate for unrestricted, engineering-led computing expansion.
Jensen Huang’s message draws a clear battle line between chipmakers and frontier model developers. By defining safety as a software engineering benchmark rather than a matter for statutory regulators, Nvidia makes it clear that it expects compute clusters to run at full throttle, leaving individual developers responsible for deciding when their products are ready for market.
(Feature image credits to Kiyoshi Ota/Bloomberg via Getty Images)
Read More: OpenAI reassigns 25% of engineers to defense after experimental model breaches containment