NVIDIA CEO Says AI Spending Flywheel Is ‘Really, Really Flying’

NVIDIA Corp. (NASDAQ: NVDA ) CEO Jensen Huang said demand for artificial intelligence computing is accelerating as AI becomes more useful and companies race to build new infrastructure. Speaking with CNBC's Jim Cramer at Dreamforce in San Francisco, Huang said companies have been building NVIDIA-powered "AI factories" rapidly over the past six months. AI Factories Deliver Strong Returns Huang estimated that a 1-gigawatt NVIDIA AI factory costs about $50 billion to $60 billion to build.

However, he said the same infrastructure can generate about $50 billion in annual rental revenue. That implies a return on invested capital of roughly one year under current market conditions, according to Huang. He also said NVIDIA infrastructure can remain useful for more than five or six years. Meanwhile, AI usage continues to climb.

Huang said the rate of token generation has increased 25-fold in less than a year. Open models now account for nearly 70% of generated tokens, up from about 30%. NVIDIA CEO Says AI Safety Is Creating A New Data Center Demand Engine Huang also pushed back against concerns that AI safety efforts could derail industry growth. During the interview, Jim Cramer jokingly asked Huang whether people should be preparing to "die in 2030" because of AI.

"We're not going to die in 2030," Huang replied. Huang said thousands of companies worldwide are working on AI safety, security and guardrails. He argued that the industry can build AI that is "incredibly useful," helpful and safe at the same time. He said companies should test products thoroughly and avoid releasing systems before they are safe.

In fact, Huang expects testing to become a third major category of AI infrastructure alongside training and inference. He said large testing environments will require additional data centers, creating another source of compute demand. Huang also pointed to rising rental prices as evidence of tight demand. He said NVIDIA Grace Blackwell systems can now rent for about $16 per GPU hour, compared with roughly $5 under some contracts signed a year ago.

As older contracts expire, cloud providers are raising prices, boosting revenue forecasts and investing in more infrastructure, Huang said. S. equity futures. 16%.

Investors are also watching whether NVIDIA can break above nearby resistance after recent rebounds stalled. 81. 24. The longer-term trend remains constructive.

79. In addition, the 20-day SMA remains above the 50-day average, while the 50-day remains above the 200-day. Momentum is neutral. 60.

That suggests neither buyers nor sellers have firm control. Resistance sits near $214, close to the 50-day SMA. Meanwhile, $190 represents a key support area. 54.

22. Piper Sandler initiated coverage with an Overweight rating and a $300 price forecast on Sept. 10. Rosenblatt maintained a Buy rating and $390 forecast on Sept.

4. Needham also maintained a Buy rating and $300 forecast that day. Edge Rankings The Edge scorecard shows NVIDIA scoring strongly on growth and quality. 38.

05. 41, reflecting a more demanding valuation. NVIDIA ETF Exposure NVIDIA also carries significant weight in several exchange-traded funds. S.

73% of the Franklin Focused Dynamic Growth ETF (NASDAQ: FFOG ). Large inflows or outflows from these funds can therefore translate into additional buying or selling of NVIDIA shares. 32 during premarket trading on Wednesday, according to Pro data. Image via Shutterstock Read Also: Stock Market Today: S&P 500, Dow Jones Futures Gain Ahead of FOMC's Interest Rate Decision — INTC, SKHY, LEN, TCOM in Focus