Image: Adobe.
The folks selling the "picks and shovels" of this AI gold rush are the real winners.
So, OpenAI is out there pitching that it’ll pull in $10 billion of revenue next year and a massive $125 billion by 2029.
Sounds great, right? The problem is, it won’t actually make money until it’s already a giant global corporation.
That’s a flashing red light to me, and the only people really cashing in here are its suppliers like Nvidia, CoreWeave, and even folks like Jony Ive who’ve found ways to hitch a ride.
Some leaked numbers are doing the rounds to back up OpenAI’s latest fundraising push, $40 billion at a jaw-dropping $300 billion valuation.
The plan is to double revenue in 2025, then keep compounding at almost 90% a year until 2029. On paper, maybe that adds up in the current hype cycle.
But in the real world, if there’s even a small correction in AI enthusiasm, and I’m betting there will be, OpenAI is going to miss those targets by a mile.
Why? Because large language models like ChatGPT don’t actually understand cause and effect. They can’t really “reason.” That means AI isn’t about to transform the world overnight, it’ll just make us more productive at certain tasks.
In the meantime, there’s massive overbuilding of AI capacity happening right now, which will drive prices down. Selling AI services won’t be nearly as profitable as people expect.
When those rosy forecasts crumble, valuations will reset, and we’ll likely see a big shake-up in who actually survives. That’s when I think Apple will swoop in and buy one of the big AI players currently burning through cash.
Apple needs its own in-house solution, and one of these struggling names, OpenAI, Anthropic, Mistral, will fit the bill.
But even if I’m wrong about Apple, OpenAI’s numbers still look crazy to me. Most start-ups hit breakeven when they’re still small or mid-sized.
OpenAI says it won’t make a dime until it’s already a global behemoth. That’s nuts. Any company that needs to be huge before it’s profitable is one bad market shock away from blowing up.
OpenAI itself expects to haul in about $256 billion in revenue between now and 2029, with expenses roughly matching that. Translation: the company won’t be the one pocketing the profits, its suppliers will.
The folks selling the "picks and shovels" of this AI gold rush are the real winners.
And that’s why I keep coming back to Nvidia. It’s already making serious money, it’s got the balance sheet to handle a downturn, and it’s been smart about heading off competition. Sure, other suppliers, from data center components to nuclear power, will do well too. But as far as direct AI bets go, Nvidia is still the only one I’d touch.
By 2029, I don’t expect OpenAI, Anthropic, or Mistral to still be standing alone. They’ll get bought out when the cash burn catches up with them.
Meanwhile, Nvidia will still be selling the tools everyone else depends on. That’s why it’s still my only direct AI investment, even as I also hold some exposure to adjacencies like inference at the edge and nuclear power.