AI's $1 trillion risk keeps growing
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Allie Carl/Axios
The biggest tech companies are set to spend $1 trillion on AI by next year, according to multiple banks, a bill so big that it's propping up both the stock market and economy.
Why it matters: Our financial system is now load-bearing on AI spending that may never pay off, and most investors can't even see what the full tab is.
State of play: The biggest tech firms are on track to spend $700 billion on their AI ambitions this year, double their 2025 spending, according to Goldman Sachs. That could swell to over $1 trillion next year, according to multiple estimates.
- AI costs went up, not down, for four of the Big Tech companies that reported earnings last week, according to Bank of America.
- As their cash flows erode, these companies argue they have to keep spending to stay ahead in the AI race.
What they're saying: "The question I want answered is not whether any of these four companies beat the Street by 2 cents. I genuinely do not care about that," Mark Malek, chief investment officer at Siebert Financial, wrote in a note to clients.
- "What I want to know is whether the monetization engine is actually firing."
- That's a very difficult question to answer because there's no end in sight for AI spending: Computing processing power, or "compute," is constrained, and demand is at an all-time high, meaning the cost of servicing AI keeps rising.
Zoom out: While the spending spree ramps up, it's increasingly propping up the stock market and economy.
- AI spending from businesses is a larger contributor to economic growth than consumer spending.
- AI's big spenders, meanwhile, make up nearly half of the stock market, the outperformance of which has minted more millionaires than ever, who in turn are spending more while lower-income consumers struggle under the weight of inflation.
What we're watching: With both the market and the economy relying so much on continued AI spending, the danger is if it doesn't start to pay off and the Big Tech companies start to turn the spigot off.
- "The moment one of those hyperscalers doesn't succeed ... you break a link in the chain," PitchBook's Harrison Rolfes tells Axios.
- That could cause a market correction with "ripple effects to everyone else," he added, saying that 2028 through 2030 will be revealing as more compute capacity is expected to come online then.
Threat level: And there's more spending we don't even know about.
- The biggest tech companies are spending over half a trillion additional dollars on data center leases that aren't on their balance sheets, according to a report from Moody's.
- According to Michael Burry, made famous by "The Big Short" for betting against the housing market heading into 2008, Big Tech's earnings are overstated by as much as 42% due to these unknown risks.
The bottom line: The true cost of the AI buildout is larger than what's showing up on balance sheets, and investors are pricing risk they can't fully see, just as the broader market and economy grow more dependent on the spending spree continuing.
