Full Stack Is About to Get Physical

I spent today morning with a room full of third-year students. In our program, 5th semester is when they pick a domain — the thing they’ll say they “do” for the next few years, maybe longer. So we went down the list. Data Analytics and Machine Learning. Full Stack Development. Software Development. Cybersecurity. Cloud. XR. Web3.

Standard stuff. I've given some version of this talk before.

And then, somewhere between explaining what "full stack" means and watching about forty heads nod like they already knew, something clicked for me — mid-sentence, honestly.

We are still defining full stack the way we defined it in 2015. Frontend, backend, database, deploy. Some HTML, some API, some cloud. The whole stack lives inside a screen.

But look at what these students are actually going to build.

A wearable that reads your body and syncs to an app. A headset that tracks your hands and renders a world at 90 frames a second. A drone that sees a field and decides where the water goes. A robot arm in a lab. An EV that ships firmware updates while it's parked in your garage. A retail store with no cashier. A hospital device that talks to a model in the cloud and back again in under a second.

None of that is just software. And none of that is just hardware either. It's a stack — it's simply that the bottom of the stack stopped being a database and started being a sensor.

That's the realization I had standing in front of them: full stack is not going to mean web forever. It's going to mean sensor → firmware → edge → network → cloud → model → interface → back to the physical world. Someone has to hold that whole chain in their head. Someone has to be able to debug across it — because the bug is never where you expect it, and it definitely doesn't respect the boundary between the PCB and the API.

We don't really have a name for that person yet. We split them into "embedded engineer" and "app developer" and then put them on different floors and wonder why the product feels stitched together.

I don't think the split survives. Not because anyone will decide to remove it, but because the products won't allow it. The interesting things being built right now — XR, robotics, spatial computing, agentic systems with actuators attached — all live exactly on the seam. Whoever can work across the seam becomes the most valuable person in the room by default.

So here's what I ended up telling them, and I'm still chewing on it:
Don't pick a domain. Pick a spine. Go as deep as you can in one thing, sure — but make sure you can walk one layer down and one layer up from wherever you land. If you do web, learn what a microcontroller actually is. If you do embedded, learn how a real app ships. If you do XR, you already have no choice — you're standing on the seam whether you like it or not.

The stack is getting taller. It’s also getting a body. Might as well learn where the body is.