NVIDIA CEO Jensen Huang speaks during the Live Keynote Pregame during the Nvidia GTC (GPU Technology Conference) successful Washington, DC, connected Oct. 28, 2025.
Jim Watson | AFP | Getty Images
Nvidia CEO Jensen Huang said connected Friday that artificial quality had reached a "virtuous cycle," tipping the manufacture for continuous growth.
Speaking astatine the APEC CEO Summit successful South Korea, Huang said the immense improvements successful AI models were starring to much concern successful the technology, which was, successful turn, improving the AI models adjacent further.
"We person present achieved what is called the virtual cycle," helium said connected signifier astatine the event, wearing a suit alternatively than his accustomed achromatic leather jacket.
"The AIs get better. More radical usage it. More radical usage it, it makes much profit, creates much factories, which allows america to make adjacent amended AIs, which allows much radical to usage it. The virtual rhythm of AI has been designed, and this is ... the crushed wherefore you're seeing the world's capex going truthful fast."
His comments travel arsenic Big Tech is spending billions to physique retired AI-related infrastructure and service its extremity users.
This twelvemonth was expected to beryllium a large one for AI walk with Meta, Amazon, Alphabet and Microsoft announcing plans to walk implicit $300 cardinal combined connected AI technologies and datacenter buildouts. This looks acceptable to proceed into 2026 arsenic the tech giants program to boost spending again, per their respective earnings, reported this week.
Dan Ives, Wedbush Securities planetary caput of exertion research, described Nvidia arsenic "the instauration of the AI Revolution" successful comments to CNBC aft Huang's comments connected stage.
He described the AI virtuous rhythm as: "The much demand, the much gathering of AI gathering blocks. And request creates much request and capex."

Huang stressed that profitability was astatine the bosom of the existent roar successful AI superior investment.
"When thing becomes profitable, you privation to manufacture much of it, conscionable similar erstwhile you're manufacturing chips and wafers and DRAM, if the manufacturing of those chips were profitable, you privation to physique much factories to make much chips," helium added.
A caller epoch of computing
It is the opening of a caller epoch of computing, as, with AI, "every azygous furniture of the computing stack is being fundamentally changed," Huang said connected stage.
We are astatine the opening of a 10-year build-out of this caller era, helium added.
"AI runs connected GPUs [graphics processing unit], whereas hand-coded bundle runs connected CPUs [central processing unit]. This full bundle stack, from the ... the needs of energy, chips, the infrastructure, each of the bundle associated with the systems, the AI models and the applications connected top, each azygous furniture of computing has been fundamentally changed," helium said.
"Just think: the machine manufacture has been mostly the aforesaid for 60 years, and now, with AI and accelerated computing, each azygous furniture of the computing stack is being changed. All of the computers we've created successful the past, a trillion dollars, possibly more, of computers needs to present beryllium transitioned, shifted to the caller computing platform," helium added.
Nvidia, which became the archetypal institution to surpass $5 trillion successful marketplace worth earlier this week, announced a concern with Korean semiconductor elephantine Samsung earlier connected Friday. Samsung plans to buy and deploy a clump of 50,000 Nvidia GPUs to amended its spot manufacturing for mobile devices and robots.
Huang painted a representation of the aboriginal successful which AI is capable to "work," alternatively than conscionable beryllium utilized arsenic a tool. Highlighting the emergence of afloat automated manufacturing factories, the CEO expects AI to reshape $100 trillion worthy of industries astir the world.
— CNBC's Kif Leswing contributed to this report.










 English (US)  ·
                        English (US)  ·