The AI industry has a dirty little secret it hopes you'll overlook while it's busy building $1.4 trillion worth of data centers: most companies are getting virtually nothing in return. We're watching the largest infrastructure bet in human history, and the productivity stats are somewhere between "meh" and "lol."

AI is the greatest corporate productivity hoax since open-plan offices. We've burned enough electricity to power a small nation, watched Big Tech's stock prices ascend to theological heights, and the average company's output looks... exactly the same as 2023. PwC's own research shows 75% of AI's economic gains landing in the hands of just 20% of companies. The rest? They're paying for Copilot subscriptions and calling it a "digital transformation."

Let's be real, the nerds aren't completely wrong. Developers are shipping code 40% faster. Customer service bots are handling millions of tickets without crying. AI-assisted drug discovery cut what used to be 12-year research cycles down to 3. The productivity gains are real, they're just incredibly narrow and concentrated in places that were already winning. If you're a software company, a biotech, or a Wall Street quant shop, AI is basically a cheat code. The $1.4 trillion data center buildout is the infrastructure layer that makes all future gains possible. You can't complain about train delays while they're still laying the tracks.

Here's what the charts actually show when you zoom out: aggregate productivity growth in the broader economy is still basically flat. AI agents, the thing everyone swore would automate entire job functions by now, are still making rookie mistakes, hallucinating facts, and generating enough liability nightmares that legal teams are quietly putting the brakes on deployment. Meanwhile, the energy bill is catastrophic. AI data centers could draw up to 29.6 gigawatts by 2030, enough to power the entire state of New York. We're strip-mining the electrical grid for productivity gains that look a lot like "the intern writes faster first drafts." The 80% of companies seeing minimal returns aren't failing to adopt AI. They're failing to find uses that justify the cost.

The hype was always going to outrun the reality, that's just how technology cycles work. What's unusual this time is the scale of the gap. When your infrastructure bet costs $1.4 trillion and your productivity story is "well, coders are pretty happy," something is mathematically off. That said, writing the whole thing off is also wrong. The companies capturing those gains aren't hiding a scam, they're just proving that AI rewards depth, not dabbling. The winners are the ones who rebuilt entire workflows around it. The losers are the ones who bolted a chatbot onto their existing process and expected miracles. The real hot take? Most companies aren't losing to AI. They're losing to the other companies who actually figured it out.

Hot takes delivered every Wednesday, whether you asked for them or not. Subscribe here before the robots start writing these too.