On Wednesday, Nvidia, the world leader in artificial intelligence computing, reported its third-quarter numbers, blowing past Wall Street’s revenue and earnings expectations and issuing a stronger-than-expected forecast for the current quarter. CEO Jensen Huang described demand for the company’s newest generation of accelerated computing chips, Blackwell, as “off the charts.”
Demand exceeds supply for Nvidia’s chips by a ratio of 12:1. A few weeks ago, Huang said Nvidia already sold out of accelerated computing chips for the next two years. Nvidia does not see an AI bubble. Wall Street analysts and a few media commentators who insist otherwise are wrong.
TRUMP DEFENDS SAUDI CROWN PRINCE WHILE DOWNPLAYING KHASHOGGI MURDER: ‘THINGS HAPPEN’
Earlier this year, OpenAI threw down the gauntlet to America’s hyperscalers: Amazon, Meta, Microsoft, and Alphabet. OpenAI is racing to secure compute and expects to invest around a trillion dollars over the next few years on data centers and related infrastructure. Hyperscalers, along with Oracle and other companies, such as Elon Musk’s xAI, have no choice but to follow OpenAI’s capital investment strategy. Each is competing ferociously against the others; they cannot afford to be left behind. AI is a profoundly positive technology that America’s finest companies will refuse to be second best in.
AI promises to be as transformative as the railroad, the electrification of the U.S. economy, and the internet. It will lift the United States’s productivity and raise the long-term growth rate. Will all of this arrive this decade? Perhaps not. But it will arrive. Every American will benefit from AI.
Importantly, today’s equity markets also bear little resemblance to the markets of the late 1990s and early 2000s, during the internet bubble. Many of the companies of that era were not profitable.
By contrast, the hyperscalers are extraordinarily profitable. Cisco Systems, a leader in internet buildout, once traded at more than 100 times earnings. Nvidia, the poster child for AI, trades at roughly 25 times forward earnings, astonishingly a 50% discount to its growth rate. Most large U.S. publicly traded companies trade at price-earnings premiums to their earnings growth rates. And unlike the dot-com era, capital spending is being funded largely through internal cash flow, not external debt.
China is also accelerating its AI investment, but the U.S. starts from a much larger economic base and has far greater resources. Next year, China may invest around $150 billion in AI; in 2026, capital investment by U.S. hyperscalers could exceed $600 billion. The U.S. is ahead and is running fast to stay ahead. Europe, meanwhile, knows it cannot afford to fall behind and is struggling to mobilize sufficient capital. Emerging economies, such as India and Nigeria, have their own ambitious plans to build data center capacity. The race for more computing power is a global phenomenon.
TRUMP FETES SAUDI CROWN PRINCE WITH PSEUDO-STATE DINNER ATTENDED BY CRISTIANO RONALDO
The outlook for AI investment remains strong. Companies simply cannot meet demand for AI compute. Microsoft routinely notes it lacks the computing capacity it needs. That shortage has given rise to the “neo-cloud,” new operators that rent out data-center capacity to the highest bidder. Demand remains intense.
From time to time, worries about an AI bubble will resurface. But as long as valuations stay reasonable, as they are today, and companies continue to fund their AI spending out of cash flow rather than debt, fears of a bubble are misplaced.
The author owns shares in Nvidia.
James Rogan is a former U.S. foreign service officer who later worked in finance and law for 30 years. He publishes a daily Substack on financial markets, politics, and society. He can be followed on X and reached at [email protected].
