SAN FRANCISCO, 19 November 2025 — Nvidia (NASDAQ:NVDA) CEO Jensen Huang insists the world is not in an AI bubble but standing at a technological “tipping point,” even as market skeptics warn that the chipmaker’s explosive growth rests on increasingly fragile foundations.
Buoyed by surging demand for its AI processors, Nvidia’s shares jumped nearly 5% in premarket trading, after the company delivered earnings and guidance that again exceeded Wall Street expectations. But analysts caution that the forces driving Nvidia’s ascent may also expose it to vulnerabilities the company cannot fully control.
Nvidia, now valued at over US$4.5 trillion, disclosed in a regulatory filing that 61% of its US$57 billion in Q3 revenue came from just four unnamed customers, up from 56% in the previous quarter. Past disclosures suggest they may include Microsoft (NASDAQ:MSFT), Meta (NASDAQ:META) and Oracle (NYSE:ORCL).
The chipmaker is also spending more to rent its own chips back from cloud customers, doubling commitments to US$26 billion, compared to US$12.6 billion in Q2. Several long-term rental contracts stretch through 2031, reflecting Nvidia’s race to secure supply amid unprecedented demand.
Nvidia has also pledged US$100 billion to OpenAI and US$10 billion to Anthropic, deepening ties with AI developers that remain unprofitable.
These interlocking financial relationships have fueled concerns.
“A lot of this growth is coming from loss-making startups or loss-making projects,” said Chaim Siegel, analyst at Elazar Advisors.
“Most likely the cycle ends badly unless all these companies agree to stop spending together… which is a near impossibility.”
Huang: Nvidia Is Powering Three Global Shifts
Huang pushed back strongly against bubble fears during the earnings call.
“We see something very different from the market talk of an AI bubble,” he said.
He outlined three long-term transitions Nvidia believes will drive infrastructure demand for years:
- Shift of non-AI software to accelerated computing: Engineering simulations, data science and other workloads move away from CPUs to Nvidia’s GPUs.
- Creation of entirely new software categories: Including AI coding assistants, agents, and generative tools.
- AI moving from digital to physical environments: Powering robots, autonomous vehicles, and real-world automation.
“Nvidia is chosen because our singular architecture enables all three transitions,” Huang said.
Obstacles: Power, Land and Rising Competitors
Even Nvidia bulls warn that physical infrastructure, not chips, may become the bottleneck.
Ivana Delevska, CIO of Spear Invest, said land and power demands for hyperscale data centers remain a major concern.
Huang acknowledged the challenge:
“We’ve now established partnerships with so many players in land and power… none of these things are easy, but they’re all solvable.”
But competitive pressure is rising.
Alphabet (NASDAQ:GOOGL) and Amazon (NASDAQ:AMZN) are designing their own AI chips, aiming to serve the same customers who currently rely almost exclusively on Nvidia.
Jay Goldberg, senior analyst at Seaport Research Partners, remains skeptical:
“They’ve said they are sold out for the year and probably next. The list of things that could go wrong is longer than the list of things that could go right.”
