Every big tech rally eventually gets compared to the dot-com bubble. Brad Gerstner would like to retire that comparison.
The founder and CEO of Altimeter Capital argues that the AI boom is a different animal. His core reason is simple: demand for intelligence, in his view, is unlimited.
Earnings, not just enthusiasm
Gerstner’s case starts with a distinction between two ways stocks go up. Prices can rise because companies earn more, or because investors agree to pay more for the same earnings.
The second kind is called multiple expansion. It is the market equivalent of paying more for the same house because the neighborhood feels exciting.
Speaking on the All-In Podcast on September 17, 2026, Gerstner argued that today’s rally belongs to the first category. He said the expansion is being powered by rapid earnings growth rather than investors simply stretching valuations.
His headline example was Nvidia. Gerstner noted that the chipmaker trades at a forward multiple of just 14 times next year’s projected GAAP earnings.
The broader market tells a similar story in his framing. Market earnings rose 26% year-to-date in 2026, driven primarily by AI infrastructure spending.
That growth arrived even as multiples contracted. Put another way, profits climbed faster than prices, so stocks actually got cheaper relative to what they earn.
Unconstrained demand meets a physical ceiling
By mid-October 2026, Gerstner was describing AI demand as “unconstrained.” The bottleneck, he suggested, is not appetite. It is supply.
Specifically, he pointed to physical limits like power and materials.
The scale of the planned buildout hints at why supply is tight. SpaceX plans to add 10 GW of compute in 2027, a figure equal to approximately 25% of total US compute capacity.
Revenue at the AI labs themselves has also moved fast. Anthropic’s revenue went from $2 billion to an $11 billion monthly run-rate between January and March 2026.
A bull who is trimming his position
In his October 2026 interviews, including appearances on CNBC’s Halftime Report and Squawk Box, Gerstner stayed bullish on AI over the long term.
He also acknowledged that his hedge fund’s exposure to AI investments had dropped to its lowest level in three years.
His optimism is pointed at 2027 and 2028. He expects explosive revenue growth at leading AI labs, combined with supply constraints, to drive the next leg of expansion.
The math hyperscalers need to work
The most important number in Gerstner’s framework may be a revenue target. He estimated that leading AI labs need to scale revenues from $100 billion to upwards of $180 billion annually.
Why that figure matters: it is what he believes is required to sustain capital expenditure from hyperscalers like Microsoft and Amazon.
What this means
For investors, Gerstner’s view offers a framework rather than a buy signal. The core question is whether earnings keep doing the heavy lifting, or whether valuations start running ahead of results.
The AI lab revenue target is the metric to watch most closely. Growth from $100 billion toward $180 billion-plus would support the hyperscaler spending case. A stall would raise uncomfortable questions about the entire capex cycle.
There is also a lesson in Gerstner’s own positioning. A bullish long-term thesis paired with three-year-low exposure suggests caution about timing, even from one of the trade’s most vocal supporters.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.