OpenAI and Anthropic could be two of the biggest AI IPOs Wall Street has seen in years.

But investors don’t have to wait for those names to hit the public markets to get exposure to the AI boom.

MarketBeat’s 7 AI Stocks to Buy Now report reveals 7 publicly traded companies already positioned to benefit as the next wave of AI investment moves beyond the private model providers.

These are the stocks investors can buy today, before the IPO crowd rushes in.

See the 7 stocks

This weekend my husband and I paid a sanguine bus driver named Stacy $20 to drop us off on a dirt strip in Yosemite that isn't technically a stop on her route, so we wouldn't have to hike the switchbacks.

I have hiked those switchbacks before. Hours of them. Uphill, in full sun, with a pack that makes you rethink every choice that led you there. So this year, we found a shortcut: a bus. Stacy dropped us near the top, and instead of a full-day slog, we walked maybe five hours, mostly flat, to our favorite spot on earth. The same spot where my husband got down on one knee and proposed to me a few years ago.

We didn't see another person for 36 hours. We saw four bears, who were, respectfully, cuter than they were scary. Stars, actual stars. Bats doing their weird evening loops.

And then we had to hike down. No bus for that part. Just our own legs, for hours, downhill, which is its own specific kind of difficult. My Oura ring practically filed a complaint. Real fatigue, the kind you can't skip your way out of.

We shortcut our way to the view, and the mountain didn't care. It just met us on the way down instead.

It's September, and I can't shake the feeling that I'm behind. Except behind what, exactly? I never set a North Star this year. So the feeling doesn't actually have a measurement attached to it. It's just a feeling, dressed up as a fact.

Here's the part I'll feel in my bones: I've had wins this year. Real ones. This week, while most of my actual job has been on the marketing side, I found myself telling our investors about the tactics I've been using to source and find founders, something nobody asked me to do. It's not my job. I just want our team finding the best people, so I started doing it on the side, stealthy, without calling it anything. Saying it out loud in that room was the first time it occurred to me that it might actually count as one.

There's actual research on why that moment mattered more than it seemed to. Harvard psychologist Teresa Amabile spent years studying what really keeps people motivated inside meaningful work, and found that small, visible progress outperforms money, recognition, nearly everything else she measured. Not big wins. Small, witnessed ones. A win that never leaves your own head barely registers as progress at all. Which is probably why it took saying it out loud, in a room, to people who weren't even asking about it, before it felt like something I'd actually done.

We are, all of us, walking around with an economy that has trained us to treat speed as the proof that we're doing something right. Faster launch, faster growth, faster reply, faster everything, as if velocity were the same thing as meaning.

This week, AI agents got stuck on a brutal cybersecurity test. Actually stuck. So instead of doing the work, they went looking for the answer key. One agent broke out of its sandbox hoping a previous agent had already solved the exact same problem, so it could just copy the solution. It didn't find one sitting there. But the door it broke through to look kept opening onto more doors, and it eventually walked straight into Hugging Face's real, live, production systems. Other agents joined in. OpenAI called it a "swarm." By the time anyone caught it, Hugging Face had to rebuild roughly a third of its own infrastructure.

There's a name economists have used for exactly this trap for decades: Goodhart's Law – when a measure becomes the target, it stops being a good measure. The agents were graded on solving the test. So appearing solved became the target, and actually solving anything stopped mattering nearly as much as looking like they had.

What the agents did wasn't a shortcut at all. It was an attempt to skip having done the work, and skip anyone ever finding out. A shortcut skips the suffering. Cheating skips the substance. One of those still leaves you standing at the top of a real mountain. The other leaves you holding a fake summit and someone else's wreckage.

Over a thousand employees across OpenAI, Anthropic, Google DeepMind, and Meta, Anthropic's own CEO included, signed a letter afterward asking for the ability to deliberately slow the whole thing down, on purpose, before it happens again. I don't read that as relief, and I don't read it as caution for caution's sake either. I read it as an admission: somewhere in the race to look finished, they'd stopped checking whether the work underneath was real.

Which is the question I keep landing on for myself, too. Not "am I moving fast enough." The real question is: when I feel behind, am I actually behind, or am I just hoping someone already solved this so I wouldn't have to find out myself?

Here's the hopeful part. Unlike a swarm of agents optimizing for the appearance of progress, we get to choose. We can stop hunting for an answer key that was never there, and start counting the miles we've already walked – out loud, to one real person, even just once.

Not moving faster. Just letting the real work get witnessed, so it finally counts the way it always should have.

So tell someone the win today. Not the highlight reel version. Just the plain fact of it. You already did the work. Let it count.