Cognition's $26 Billion Valuation: A Robotics Guy's Take on AI Coding Money

Here's the thing. Every industrial robot I ever worked on runs on software. The motion planning, the sensor integration, the safety systems, all code. And that code is expensive to write and maintain.

I talked to a guy I know at a systems integrator (I won't name him, he didn't want to be quoted) and he said they're already experimenting with AI coding tools for PLC programming. Not Cognition specifically, but similar stuff. His take: "It's not replacing our engineers, but it's making the boring parts faster."

If tools like Devin get good enough, the cost of developing robot applications drops. That's potentially huge for the industry. Or potentially terrible, depending on whether you're a young controls engineer trying to build a career.

Is this valuation insane?

Maybe. Maybe not. I genuinely don't know.

Bloomberg reports the funding round was over a billion dollars. That's a lot of confidence from investors. But I remember the late '90s when money was flowing into anything with ".com" in the name, and we know how that ended.

The difference, I suppose, is that Cognition has real revenue. Substantial revenue. When I compare that to some robotics companies I've seen raise money on promises and prototypes, well, at least there's something there.

But here's what makes me nervous: we don't know the margins. We don't know the customer retention. We don't know how much of that revenue is from a handful of big customers versus a broad base. The company didn't disclose those details, at least not in anything I could find.

What happens to the people?

This is the part that sits with me. I've trained a lot of young engineers over the years. Good kids who went to school, learned their craft, expected to spend careers writing code for industrial systems.

If AI can do 30% of their job, maybe they become more productive. If it can do 70%, maybe there just aren't as many jobs. I don't think we know which it is yet. Anyone who tells you they're certain is selling something.

Cognition's CEO Scott Wu did an interview with Bloomberg Tech, but the substance was thin. Mostly the standard startup founder optimism about how their technology is transformative. I've heard that pitch a hundred times at trade shows. Sometimes it's true. Often it isn't.

So what's the takeaway?

I don't have a neat conclusion here. A two-year-old company is worth $26 billion because investors believe AI will fundamentally change how software gets written. That belief might be right. The revenue suggests it's at least partially right.

For robotics specifically, cheaper software development could accelerate adoption. More applications become economically viable when the programming costs drop. That's good for the industry, sort of, in the aggregate.

But I think about the integrator I mentioned, and his team of controls engineers. They're the ones who actually make robots useful in the real world. If their skills get commoditized by AI tools, that's a genuine loss, even if the spreadsheets look better.

I'm going to keep watching this one. Not my usual beat, but the money is too big to ignore.

James Chen · 8 hours ago · 4 min