The Headcount Fallacy: Why 10 Million Engineers Don’t Guarantee AI Adoption

Forget the hype about AI in design. Ask yourself this: if the market for AI mechanical design tools is so obvious, why are factories still losing decades of know-how every time a veteran retires? Why are workers still relying on instinct to tell apart molten aluminum from molten zinc?

I stumbled on this interview with Maor Farid, CEO of Leo AI, over at CB Insights. He’s building AI tools for mechanical design, and he frames the market size this way: there are 10 million mechanical engineers worldwide, each earning around $100,000 a year. Multiply those, and you get a trillion-dollar payroll—a tempting target for any automation startup.

It’s a seductive math, especially if you’re a vibe coder or indie hacker looking for the next big niche. But having tracked manufacturing problems through PainSignal, I’ve learned that headcount is a lazy proxy for real demand. The manufacturing sector has 169 problems and 143 app ideas actively documented—and many of the top problems aren’t about lack of tools; they’re about deep, dangerous gaps that no one has filled yet.

Take KnowledgeCapture Pro, for instance. That’s a problem we track where companies are bleeding specialized knowledge as older workers leave. It has a severity of 5 out of 5 and an opportunity score of 54 out of 100. That means it’s both urgent and wide open. You can’t solve that with a generic design co-pilot; you need AI that captures the workflows and heuristics that live only in someone’s head after 30 years on the floor.

Or look at MeltSense Pro, another 5/5 severity monster. In die casting, workers have to visually distinguish molten metals—a high-stakes guessing game that leads to defects and safety incidents. Again, a design tool alone isn’t enough; you need something that brings process intelligence into the design phase so that these hazards don’t exist in the first place.

Leo AI might eventually get there, but the interview glosses over these adoption barriers. Farid admits they can’t estimate an existing market because AI tools for mechanical design are essentially new. But that’s exactly why you have to dig deeper than headcount. The question isn’t how many engineers exist; it’s how many of them are screaming for a solution to a problem that’s costing their company money or putting lives at risk every single day.

Our data suggests that demand is real—but it’s lumpy and highly specific. There are problems like PermitPulse (severity 4/5), where missed inspections lead to regulatory fines, or a variety of other pain points that any AI design tool champion will have to address if they want adoption beyond the early adopter tinkerers. The opportunity score ceiling we see is around 59/100, indicating plenty of room but also a lack of perfect solutions yet. That’s a call to builders, not a guarantee of a quick win.

For the seed investors scanning this, the takeaway is that the market’s not as simple as “10 million engineers need AI.” The real TAM is defined by the severity and pervasiveness of these unsolved problems. And for the vibe coders and indie hackers, there’s a playbook: don’t build another generic design tool. Go mine the manufacturing problem lists, and target one of those high-severity, high-opportunity gaps that big players are ignoring.

The headcount fallacy is seductive, but the data tells a more nuanced story. Yes, the mechanical engineering world is massive. But the winners in AI for design won’t be the ones who just count heads—they’ll be the ones who solve the problems that keep factory managers up at night.