Elon Musk Says AI Will Double US Growth to 4%: Can It?

Elon Musk thinks AI will double US economic growth next year. I think he's directionally right and wildly early. That distinction matters more than it sounds, because the whole bull case for AI stocks, AI compute tokens, and the crypto miners pivoting into data centers rests on the pace of the payoff. If Musk is right, everything reprices higher. If he's off by a few years, a lot of that capital sits idle while the bills come due.

So let's separate the thesis from the timeline.

What the numbers actually say

Musk put the move at roughly 2% to 4% growth next year. Federal Reserve policymakers, in projections released on September 16, pegged 2027 growth at 2.4%. That's a real gap. About 1.6 percentage points on a roughly $29 trillion economy works out to something near $460 billion in extra annual output. Not a rounding error. That's the kind of swing that shows up in earnings, in credit markets, and eventually in crypto liquidity.

The proponents aren't making this up from nothing. Hyperscaler capital spending has been running at levels that used to sound like typos, and a growing share of that money is landing in physical infrastructure, chips, power, cooling, land. That spending shows up in GDP as investment right away. So there's a mechanical boost happening now, whether or not the productivity miracle ever arrives.

There's also a crypto subplot. Public miners with power contracts and real estate have been quietly repositioning toward AI hosting. That's a bet on exactly this narrative. The question worth asking: are they early investors or bag holders?

The case against a 4% year

Here's where I get cautious. Technology adoption is slow, and the productivity data is stubborn. Robert Solow's old line about seeing the computer age everywhere except in the productivity statistics has aged depressingly well. We ran that experiment once with the internet and it took the better part of a decade to show up in the aggregate numbers, long after the hype peaked.

Admittedly, AI deploys faster than the PC did. No trucks, no training, no decade of enterprise procurement. But the bottleneck moved to power, permitting, and grid capacity, and none of that turns on a dime. The extra investment is also gross, not net. If AI capex crowds out other spending, the headline GDP number can flatter the reality.

And the skeptics have a point about measurement. If AI lets firms cut headcount while holding output steady, that's a productivity gain that looks like weak hiring. The track record of official forecasts in moments like this isn't great either. History suggests otherwise when someone promises a doubling.

My verdict

I don't think we get 4% next year. I'm not entirely convinced we get it by 2027 either. But I do think AI is a genuine growth accelerant, and I'd bet we drift toward 2.5% to 3% before a full doubling shows up in the headline number. The timeline is the weak part of Musk's claim, not the direction.

Color me skeptical, but the smart money here isn't long the calendar. It's long the infrastructure. Power, chips, and the data center buildout are where the near-term money gets made, and that flows through to the miners and compute markets sitting closest to the physical layer.

Time will tell, though. Watch labor productivity and unit labor costs in the next few quarterly prints. If those two break higher together, Musk's math gets a lot harder to argue with, and the crypto compute trade gets a second wind.