Nvidia chief calls AI 'the greatest equalizer' but warns Europe risks falling behind

PARIS -- Will artificial quality prevention humanity — oregon destruct it? Lift up the world’s poorest — oregon tighten the grip of a tech elite?

Jensen Huang, the planetary spot tycoon, offered his sentiment connected Wednesday: neither dystopia nor domination. AI, helium said, is simply a instrumentality for liberation.

Wearing his signature biker overgarment and mobbed by fans for selfies, the Nvidia CEO chopped the fig of a tech rockstar arsenic helium took the signifier astatine VivaTech successful Paris.

“AI is the top equalizer of radical the satellite has ever created,” Huang said, kicking disconnected 1 of Europe’s biggest exertion manufacture fairs.

But beyond the sheeny optics, Nvidia utilized the Paris acme to unveil a question of infrastructure announcements crossed Europe, signaling a melodramatic enlargement of the AI chipmaker’s carnal and strategical footprint connected the continent.

In France, the institution is deploying 18,000 of its caller Blackwell chips with startup Mistral AI. In Germany, it’s gathering an concern AI unreality to enactment manufacturers. Similar rollouts are underway successful Italy, Spain, Finland and the U.K., including a caller AI laboratory successful Britain.

Other announcements see a concern with AI startup Perplexity to bring sovereign AI models to European publishers and telecoms, a caller unreality level with Mistral AI, and enactment with BMW and Mercedes-Benz to bid AI-powered robots for usage successful car plants.

The announcements bespeak however cardinal AI infrastructure has go to planetary strategy, and however Nvidia — the world’s astir invaluable chipmaker — is positioning itself arsenic the motor down it.

At the halfway of the statement is Huang’s conception of the AI factory: not a works that makes goods, but a immense information halfway that creates intelligence. These facilities bid connection models, simulate caller drugs, observe crab successful scans, and more.

Asked if specified systems hazard creating a “technological priesthood” — hoarding computing powerfulness and stymying the bottom-up innovation that fueled the tech manufacture for the past 50 years — Huang pushed back.

“Through the velocity of our innovation, we democratize,” helium told The Associated Press. “We little the outgo of entree to technology.”

As Huang enactment it, these factories “reason,” “plan,” and “spend a batch of clip talking to” themselves, powering everything from ChatGPT to autonomous vehicles and diagnostics.

But immoderate critics pass that without guardrails, specified all-seeing, self-reinforcing systems could spell the mode of Skynet successful “ The Terminator ” movie — immense quality engines that outpace quality control.

“Just arsenic energy powered the past concern revolution, AI volition powerfulness the adjacent one,” helium said. “Every state present needs a nationalist quality infrastructure.”

He added: “AI factories are present portion of a country’s infrastructure. That’s wherefore you spot maine moving astir the satellite talking to heads of authorities — they each privation AI to beryllium portion of their infrastructure. They privation AI to beryllium a maturation manufacturing manufacture for them.”

Europe, agelong praised for its enactment connected integer rights, present finds itself astatine a crossroads. As Brussels pushes guardant with world-first AI regulations, immoderate pass that over-caution could outgo the bloc its spot successful the planetary race. With the U.S. and China surging up and astir large AI firms based elsewhere, the hazard isn’t conscionable falling down — it’s becoming irrelevant.

Huang has a antithetic vision: sovereign AI. Not isolation, but autonomy — gathering nationalist AI systems aligned with section values, autarkic of overseas tech giants.

“The information belongs to you,” Huang said. “It belongs to your people, your country... your culture, your history, your communal sense.”

But fears implicit AI misuse stay potent — from surveillance and deepfake propaganda to occupation losses and algorithmic discrimination. Huang doesn’t contradict the risks. But helium insists the exertion tin beryllium kept successful cheque — by itself.

“In the future, the AI that is doing the task is going to beryllium surrounded by 70 oregon 80 different AIs that are supervising it, observing it, guarding it, ensuring that it doesn’t spell disconnected the rails.”

The VivaTech lawsuit was portion of Huang’s broader European tour. He had already appeared astatine London Tech Week and is scheduled to sojourn Germany. In Paris, helium joined French President Emmanuel Macron and Mistral AI CEO Arthur Mensch to reenforce his connection that AI is present a nationalist priority.

—

Chan reported from London.