Nvidia's Huang Says Build AI "As Fast As We Can"

Nvidia chief executive Jensen Huang told CBS News that technology companies should keep pushing artificial intelligence development forward "as fast as we can," rejecting recent calls from other industry leaders to slow down, as Nvidia shares closed at 222.27, up 1.34% on Friday, 18 September 2026.

Jensen Huang has put himself on one side of the sharpest argument in technology. The chief executive of Nvidia (NVDA), the world's largest chipmaker, told CBS News that technology companies should keep developing artificial intelligence "as fast as we can" — a direct answer to industry figures who have suggested in recent weeks that the field may need to pull back.

The remark is short, but it lands on a live fault line. A growing set of executives and researchers has argued that the pace of frontier model development has outrun the ability to test, govern and understand what is being built. Huang's position is the opposite: the risk is in going slower, not faster. His CBS MoneyWatch interview makes that stance explicit rather than implied.

Why the chipmaker's view is not a neutral one

Every argument in this debate comes attached to a balance sheet, and Nvidia's is the largest one in the room. The company sells the accelerators that train and run the models at the centre of the dispute. If frontier labs slowed voluntarily — fewer training runs, smaller clusters, longer evaluation cycles before the next scale-up — the demand curve for high-end compute would flatten with them.

That does not make Huang wrong. It does mean his answer to "should we slow down?" is the answer his company needs. Readers weighing the comment should treat it as a statement of strategic intent from the largest supplier in the chain, not as a disinterested safety assessment. Investors who hold the stock are, whether they framed it that way or not, taking the same side of the bet.

The commercial logic behind the accelerationist case is straightforward enough. Capital spending on AI infrastructure is committed years ahead of the revenue it is meant to produce. Data centre builds, power contracts and chip orders all sit on multi-year schedules. A general slowdown would strand a great deal of that commitment mid-cycle. Suppliers have every reason to argue that the right response to uncertainty is to keep building through it.

What the pull-back camp is actually asking for

The warnings Huang is responding to are not a call to stop. They are, broadly, a call to decouple capability gains from deployment speed — to put more distance between what a model can do in a lab and what it is allowed to do in the world. That includes longer evaluation periods, tighter controls on autonomous agents, and clearer lines of accountability when systems act on their own.

Huang's framing collapses that distinction. "As fast as we can" treats development speed as the thing to optimise, and implicitly puts the burden on the safety side to justify each delay. The counter-argument is that the cost of a delay is measurable and the cost of a failure is not.

There is also a competitive reading of the safety warnings themselves — that a lab arguing for caution is, in effect, arguing for a speed limit that binds its rivals more than it binds the lab issuing the warning. That cynicism cuts both ways, and it is one reason the debate has not resolved into anything like an industry consensus.

How the stock closed into the argument

Nvidia last traded at 222.27, up 1.34% on the day, against a previous close of 219.34, with a session range of 218.03 to 222.73, as of the close on Friday, 18 September 2026. The market is shut; those are the most recent prints, not live quotes.

The backdrop was mixed rather than uniformly strong. The Nasdaq 100 tracker (QQQ) closed at $721.45, up 0.63%, while the S&P 500 tracker (SPY) finished at $761.69, down 0.12%, and the Dow 30 tracker (DIA) closed at $515.88, down 0.48%. In other words, large-cap technology carried the session while the broader industrial tape leaked lower — a pattern that has defined much of this cycle. Nvidia's move ran ahead of the Nasdaq 100's gain on the day.

In other words, large-cap technology carried the session while the broader industrial tape leaked lower — a pattern that has defined much of this cycle.

None of that is a verdict on the interview, which was published after the close. But it frames the environment Huang is speaking into: a market that is still willing to pay up for AI exposure while showing less enthusiasm for almost everything else.

Where a slowdown would show up first

If the pull-back argument were ever to win — through voluntary restraint, regulation, or a loss of nerve among the buyers of compute — the transmission into Nvidia's numbers would not be instant. Order books at this scale are set well in advance. The earlier signals would appear elsewhere:

Frontier lab behaviour. Announced pauses, extended evaluation windows, or public commitments not to scale past a stated threshold.
Hyperscaler capital expenditure guidance. Any softening in the multi-year data centre build plans that underwrite chip demand.
Power and siting constraints. Physical limits can impose a slowdown that no one chose, independently of the safety argument.
Policy. Binding rules on deployment, testing or autonomy would change the shape of demand even if the total stays intact.
Supply-chain orders. Equipment and component makers upstream of the chipmakers typically feel a change in tempo before the chipmakers report it.

Until one of those moves, Huang's position is the operating assumption of the entire trade. The market is priced for continued acceleration, and the largest supplier in the industry has just said in plain terms that acceleration is what it intends to deliver.

The question the interview leaves open

What Huang did not do — at least not in what has been reported — is engage with the specific mechanisms the cautious camp has proposed. Saying the industry should move fast is a direction of travel, not a governance position. The harder questions are about who tests what, against which standard, and who carries liability when an autonomous system does something nobody specified.

For shareholders, the practical takeaway is narrower. The chief executive of the company at the centre of the AI build-out has signalled, publicly and without hedging, that he sees no case for slowing down. That is consistent with the demand outlook embedded in the share price. It also means that any future shift toward restraint — voluntary or imposed — would arrive as a genuine surprise to a market that has been told, repeatedly and by the best-placed source available, not to expect one.

Key facts

NVDA last close: 222.27, +1.34% (as of 18 Sep 2026, 20:00 GMT)
Prior close / day range: 219.34; session range 218.03–222.73
Huang's stance: Develop AI "as fast as we can," per CBS News interview
Tape context: QQQ $721.45 (+0.63%), SPY $761.69 (-0.12%), DIA $515.88 (-0.48%)

Frequently asked questions

What exactly did Jensen Huang say?

Nvidia's chief executive told CBS News that technology companies should keep developing artificial intelligence "as fast as we can." The comment was a direct response to warnings from other industry leaders who have suggested the field may need to pull back on the pace of AI development. Nvidia is described as the world's largest chipmaker.

Why does Nvidia's view on AI speed matter commercially?

Nvidia sells the accelerators used to train and run large AI models. If frontier labs slowed development — fewer large training runs, longer evaluation cycles — demand for high-end compute would flatten with them. Huang's call for speed aligns with his company's commercial interest, which is why the remark should be read as strategic intent rather than a neutral safety judgment.

How did Nvidia shares close?

Nvidia last traded at 222.27, up 1.34% on the day, against a previous close of 219.34, with a session range of 218.03 to 222.73, as of the close on Friday, 18 September 2026. The market was shut at the time of writing, so those are the most recent prints rather than live quotes.

What are the AI safety warnings Huang is pushing back on?

Industry leaders have argued in recent weeks that there may be a need to pull back on AI development. Broadly, the cautious camp wants more distance between what a model can do in testing and what it is permitted to do in deployment — longer evaluation periods, tighter limits on autonomous systems, and clearer accountability when things go wrong.

How did the broader market trade that session?

Technology outperformed. The Nasdaq 100 tracker QQQ closed at $721.45, up 0.63%. The S&P 500 tracker SPY finished at $761.69, down 0.12%, and the Dow 30 tracker DIA closed at $515.88, down 0.48%. That split — large-cap tech firm, the broader tape softer — has characterised much of the current cycle.

What would signal that the industry is actually slowing?

Early indicators would sit upstream of Nvidia's reported results: frontier labs announcing pauses or extended evaluation windows, hyperscalers softening multi-year data centre capital expenditure guidance, binding policy on deployment and autonomy, physical power and siting constraints, and order slowdowns at equipment and component suppliers.

Sources

Nvidia CEO calls for AI to be developed "as fast as we can" — CBS MoneyWatch