Investors' P-doom scenario in AI: Lack of profits
Add Axios as your preferred source to
see more of our stories on Google.
The AI buildout is entering a show-me-the-money phase: A growing chorus warns that the pace of spending on the technology is far outstripping the money coming in, and that may be true for a while.
Why it matters: With so much of the economy riding on investor bets on AI, it could all end in tears.
The latest: A new analysis from two Stanford economists finds a nearly $1 trillion gap between spending from the hyperscalers — Alphabet, Amazon, Meta, Microsoft, Oracle and SpaceX — and the revenue they have taken in from AI since 2024.
- And time is running out to make those investments pay off, they contend.
Zoom in: That's because a big chunk of hyperscaler spending is on chips that lose value after around five years.
- "If our assessment is correct, investor patience is likely to run out, and, depending on the pace at which they rush for the exit, the bubble will either pop or start to deflate," wrote Jared Bernstein and Ryan Cummings, economists with the Stanford Institute for Economic Policy Research.
- They found that the companies would need to triple or quadruple their AI revenue next year and every year after that for the next decade for this to work out.
- That's not impossible, but extremely difficult to say the least.
Friction point: Their analysis assumes that the cost of capital doesn't increase meaningfully, even though interest rates are now rising.
Where it stands: These economists are not outliers.
- A Goldman Sachs analysis last week found that the hyperscalers' AI revenues remain below what they need just to break even on their capex. (That's short for capital expenditures, or the money they spend on chips, data centers and real things to build AI computing capacity.)
- The bank is more optimistic that companies will ultimately see a return on their investments.
Meanwhile, leaked information about AI giant Anthropic shows that its revenue is exploding — but spending is super high. Last year, revenue was $4.6 billion, with an operating loss of nearly twice that, Reuters reported, saying that it had seen the IPO prospectus.
- This year, Anthropic's revenues look on track to be far higher: Its annualized revenue run rate topped $65 billion in the second quarter.
- And the company has told investors it will have profitable quarters.
By the numbers: The spending required for the AI buildout exceeds anything we've seen before: $10.3 trillion in infrastructure investment through 2032, according to an estimate presented at Brookings last week.
- That amounts to 3.6% of GDP a year, dwarfing the investment booms that built America's railroads, highways, electric grid and telecom networks, Axios' Courtenay Brown reported.
What to watch: The bet is that over time there will be massive payoffs — as there were for the other big infrastructure buildouts: electricity, railroads, the internet. But it may take longer than hoped.
- The Stanford economists note that it took decades before the benefits of electrification filtered into the factories.
- The 1990s dot-com stock bubble burst when investors started to realize that profits were further off than they'd initially believed.
The bottom line: "Can all these people eventually turn this into something profitable? I think they will," Cummings tells Axios.
- "Now I'm not sure who's going to do that, but I do think there will be trillions of dollars of profits that are up for grabs in the future, but just not in this super-accelerated timeline that is required to justify the investments."
