Before May 6, 1954, no human had ever run a mile in under four minutes.
Physiologists speculated it might be impossible. that the human body simply couldn’t move that fast for that long. The barrier felt absolute.
Then Roger Bannister did it. And within three years, sixteen other runners had done it too.
The barrier was never physical. It was imagined. Once one person proved it could be done, everyone else’s internal ceiling lifted.
I keep thinking about this story because I think we’re living through something similar right now. Except it’s not about running. It’s about building.
The old model went something like this: have an idea, raise money, hire a team, ship slowly, scale or die. The VC-backed path was the path to anyone wanting to scale quickly or overcome the big barriers. If you wanted to build something real, you needed permission, from investors, from the market, from a system designed to fund a few big bets and let the rest quietly fail.
That model isn’t dead by any means, but it’s already shifted rapidly.
I know people, normal people (not tech moguls) who are now shipping products on weekends that would have required a team of fifteen just three years ago. One founder I caught up with recently had spent £265,000 on building their app just 4 years ago that others have now built in a weekend.
Those who believe that there is an AI bubble, therefore AI has no use have clearly never tried the new Claude Code, Gemini, or Codex CLIs.
Yet for every developer I see on my team or who I meet who dreads this and feels existential at the capabilities of claude opus 4.5, I see a only a fraction who realise this is an incredible opportunity to go after their passion project.
The best opportunities have often come from exploring things not meant to make money. Exploration is inherently costly, but AI has made the speed and cost of that exploration exponentially smaller.
Here’s what I think is actually happening: we’re splitting into two camps of key beneficiaries from the AI wave:
(infographic made with Gemini Nano Banana)
The Owners. These are people with a capital stake in the AI infrastructure itself. Not just tech executives, your neighbour who’s heavily invested in NVIDIA counts. They benefit from AI’s rise whether they ever touch a prompt or not.
The Experimenters. These are people with no particular stake in the existing system, but who’ve realised the tools are now available to build. They’re using AI not to defend what exists but to create what doesn’t. Solving real problems. Shipping things. Learning by doing.
The counterintuitive bet is that the experimenters might have more upside. The owners are playing defence, protecting positions, optimising returns, guessing at what will have a moat. The trick is understanding that the game is changing, Joseph Schumpeter’s model of creative destruction is accelerating. Those who simply believe they can and are willing to put in the work to test it out will be able to do extraordinary things over the next year.
I want to be honest about something, though.
Not everyone has equal access to this moment. Capital helps. Network helps. Even just knowing this is possible helps, and that awareness isn’t evenly distributed. The floor is rising, but some people are starting from the basement. Those who haven’t even heard of AI may be seriously left behind, unfairly.
I don’t have a clean answer for that tension. I just think it’s worth putting out there.
But I keep coming back to one thing:
AI is extraordinary at solving problems that have already been solved somewhere else. It can synthesise, recombine, accelerate. What it can’t do, at least not yet, is feel its way toward something genuinely new that serves the real world well. It can’t sit with ambiguity the way a human can. We like to personify AI to a ridiculous extent but it may not even be capable of having its own desires. Much like Andrej Karpathy makes the point: it is more like a ghost than an animal, a facsimile but not an original.
The wanting is still ours.
And right now, for maybe the first time in history, the tools exist to turn wanting into building without needing anyone’s permission first. You don’t need a team. You don’t need funding. You need a problem you care about and the willingness to figure it out in public.
Produce more than you consume. Not as some productivity hack. As a posture toward the world.
The ceiling has lifted. The question is whether you’re looking up.
Reminds us that experimentation, not permission or capital, is the real lever in this era of AI-enabled building