Every few years, there’s a shift that feels less like a trend and more like gravity. Lately, it’s become hard to ignore how many of the most compelling problems—and the most quietly ambitious builders—are sitting at the edge where software meets the physical world.
I’m not talking about another wave of apps built on the latest foundation model, or agents that book calendar invites slightly better than last week. I’m talking about systems that do something—sense, move, manipulate, extract, assemble. Things that run on code, but are grounded in atoms.
This isn’t nostalgia for hardware. It’s not about romanticizing robotics or trying to find a physical counterpart to SaaS. It’s about a genuine shift in what’s possible. The ingredients are lining up: cheap sensors, abundant compute, increasingly capable edge devices, smarter control software, and lower costs across the board. Combine that with AI models that can operate in messier, less structured environments, and suddenly the idea of intelligent machines in the wild doesn’t feel sci-fi. It feels near-term.
What’s emerging is a convergence—one where software isn’t just an interface, but a brain. And hardware isn’t just a container, but a body.
This matters for a few reasons. First: software is becoming less defensible. Everyone feels it. The ability to replicate, remix, and launch is faster than ever. Moats that once held—data, UI, even distribution—are shrinking. If your product lives only on a screen, it’s probably on borrowed time. But combine it with a physical system that’s hard to copy, hard to ship, hard to scale? You’ve got something much stickier. The bar to entry rises. The moat deepens.
Second: many of the most important frontiers—energy, health, manufacturing, climate, defense—don’t yield to software alone. You can’t debug a mining drill with a slide deck. You can’t decarbonize an industrial process with an LLM. These are problems that live in the real world. They demand a different kind of thinking. A willingness to go low-level. A willingness to get your hands dirty.
That’s where things get exciting. Because we now have the tools to do it. AI can be embedded, not just accessed via API. Robotics platforms are modular and open. Edge compute is cheap enough to deploy everywhere. And talented engineers—some from the consumer web world, some from deep R&D—are quietly building systems that feel like they belong to the next era, not the last one.
This isn’t a call to action. It’s just what I’m seeing, over and over again: the most interesting builders are converging on problems that require both code and circuits. Both intelligence and embodiment. Both software and hardware. And they’re not doing it for novelty. They’re doing it because it’s the only way to actually move the needle.
There’s something important happening here. You can feel it.
George Payne 