Six Thousand Liars
The woman at the bar was explaining to her friend why she’d put twenty thousand dollars into a time-share in Cancún. “It’s an investment,” she said, and her friend nodded the way you nod when someone tells you their kid is gifted.
I sat there nursing my glass and thinking about six thousand CEOs.
The National Bureau of Economic Research — the real one, not some blog with a mission statement — went and surveyed six thousand C-suite executives from companies with actual revenue. They asked the question nobody in Silicon Valley wants asked out loud: is any of this AI spending actually paying off?
Almost ninety percent said no. No meaningful impact on employment. No meaningful impact on productivity. Three years of pouring money into a hole and nothing to show for it except a subscription fee and a chatbot that makes up court cases.
But here’s where it gets good. When they asked these same executives what they planned to do about it, the majority said they needed to spend more.
I’ve known gamblers like this. Everybody has. The guy at the track who’s down three hundred bucks by the fifth race and tells you the sixth is the one. His logic is airtight: he’s due. The universe owes him. He can feel it in his teeth.
The executives have a fancier version of the same feeling. They’ve built themselves a decision matrix — the article actually uses those words, “decision matrix” — that goes something like this: if we invest and AI pays off, we win big. If we invest and it doesn’t, at least our competitors are equally screwed. If we don’t invest and it pays off, we’re dead. The only scenario where sitting it out works is the one where the whole thing was a mirage.
Two definites versus two maybes, they figure. So they invest.
Dostoevsky wrote about this. The Gambler. A man who knows the math is against him but keeps playing because the alternative — walking away, admitting the game was rigged from the start — is worse than losing. At least at the table you’re still somebody. You’re still in the action. Walk away and you’re just a guy in a coat with nothing to do.
The survey found something else I loved. These executives — the ones making the spending decisions, the ones staking their companies on the AI revolution — report using the technology about an hour and a half per week. Ninety minutes. That’s less time than I spend reading the newspaper. These are people betting the farm on a tool they barely touch, like a man who buys a boat because his neighbor has one and then lets it rot in the driveway.
Harvard Business Review ran a headline that reads like a punchline at a funeral: “AI Companies Don’t Have a Profitable Business Model. Does That Matter?”
Does it matter. Let that sit. Does it matter if the companies selling the shovels can’t afford their own rent. Does it matter if OpenAI is hemorrhaging money faster than a busted pipe in January. Does it matter if the entire investment thesis rests on a business model that doesn’t actually work.
Apparently not. The fear of missing out is stronger than the fear of going broke. Always has been. The tulip guys knew this. The dot-com guys knew this. The crypto guys are still learning it, slowly, from apartments that used to be nicer.
I think about the people who actually work at these companies. Not the C-suite — they’ll land somewhere soft, they always do. The golden parachute was invented by people who jump out of buildings for a living. I mean the people in the middle. The ones who got the email about the new AI initiative and the mandatory training modules. The ones who spent a weekend learning prompt engineering because their manager said it was the future. They’ll be the first to go when the belt-tightening comes, because the belt always tightens on the people who can’t afford a lawyer.
The woman at the bar was still talking about Cancún. She’d never been. She was describing it from the brochure — white sand, turquoise water, a swim-up bar. Her friend was on her third drink and had stopped nodding.
Six thousand executives looked at three years of nothing and saw a future full of something. That takes a kind of faith I’ve never been able to manage. The faith of people who’ve never had to choose between the electric bill and groceries. The faith of people who fail upward so reliably that risk is just a word in a PowerPoint slide.
The horse is still running. The race isn’t over. But I’ve been at enough tracks to know what a losing bet looks like while it’s still being placed.
The ice in my glass has melted. Somewhere a CEO is approving another seven-figure AI contract because his competitor did it last quarter and he can’t be the one who didn’t.
Screw it, right?
Source: Corporate AI Isn’t Actually Making (or Saving) Very Much Money