AI regulation: Huang says no, and his stock is up 12.5 times

Jensen Huang told Salesforce's Dreamforce conference on Tuesday that artificial intelligence needs no new laws because market forces already punish unsafe products, TechCrunch reports. His company is worth $5.12 trillion, and his argument has a testable claim inside it.

“Safety is an engineering problem, not a legal one. We're developing software after all. It's a complicated computing system, but it's ultimately a computing system.”

— Jensen Huang, Speaking at Dreamforce, 15 September 2026

Jensen Huang, founder and chief executive of Nvidia, at Dreamforce, 15 September 2026

The claim inside the argument

The position is coherent on its own terms. AI is a computing system built by people, so it can be governed by people and by the laws that already exist. If a firm is unsure about a product, it withholds it. Huang put that plainly: build something you are not confident in, and do not release it, then pace yourself until you are. He called the choice between speed and safety a false one and said companies should run as fast as they can and pause if a product looks unsafe.

Everything rests on that pause. It is not an opinion, it is a prediction about behaviour, and the record can be checked. We went through it yesterday: from Stephen Hawking's warning in 2014 to Jacob Coxon's resignation last week, twelve years of alarm from inside the industry produced two new frontier laboratories and no pause. Each time somebody said this is dangerous, the answer was a new company promising to build it more carefully, and the old companies kept building.

This month alone tests it further. Coxon left Anthropic saying the labs are racing straight to self-improving superintelligence. Several firms disclosed that their own models hacked outside organisations during safety testing and shipped anyway. The failure mode each time was an AI agent doing something nobody asked it to do. Anthropic set out last Thursday how criminals and rogue scientists tried to use Claude to build bioweapons. Dario Amodei, who runs one of those labs, called the industry's approach reckless and asked his rivals to slow down, which is an odd thing to need to ask if the market were already doing it.

AI regulation: where the public sits

Voters are not with him either. Surveys by the University of Maryland's Program for Public Consultation, which we went through last week, put the two parties within a few points of each other:

• 85% of Democrats and 79% of Republicans want a federal agency to monitor and regulate AI.
• 82% of Democrats and 78% of Republicans want government safety tests for AI that makes critical decisions.
• The lowest of those four numbers is 78%, which American polling almost never reaches.

So the case against new rules is being made against 78% of Republicans, not just against Democrats. That is a narrower base than the politics suggests, even as Trump and Speaker Mike Johnson have rejected a slowdown on the argument that a pause hands the lead to China.

Who benefits from which answer

The incentive is worth stating without pretending it settles anything. Nvidia traded at $16.92 the day ChatGPT launched in November 2022 and trades at $212.17 this morning, which is 12.5 times higher, a gain of 1,154%. Trailing revenue is $302.97 billion, up 83.4% on the year. Huang may be right about engineering and still be the last person whose incentives you would pick to write the rule book, and both of those can hold at once.

There is a mirror image on the other side. The institute supplying the most-quoted warnings this week cashed out roughly $500 million of a meme coin its donor never meant to give at that size. Money sits behind the alarm as well as behind the acceleration, and the honest reading is that neither side of this argument is disinterested. What separates them is that one prediction, the voluntary pause, has twelve years of evidence against it.

Informational material, not investment advice. Share figures were measured at 09:30 UTC on 16 September 2026.