AI development is the new online advertising
I was at an event the other day and got talking to someone building a carbon certification platform. It's a compliance-heavy space with a lot of tracking and reporting, and genuinely useful if you can digitize it properly.
The thing is, they had paying clients and real demand. Companies asking for features across the full lifecycle of carbon certification, not just the one-off audits they were currently doing. The bottleneck wasn't the market. It was that they didn't have a technical co-founder, so every feature was a consultancy engagement. Six-month timelines. Tens of thousands of euros. For what amounts to a relatively straightforward data platform.
I kept thinking about that conversation afterwards. Not because it's unusual (it's not, I hear versions of this story constantly), but because it reminded me of something I lived through on the commercial side.
the ads parallel
Before Meta and Google Ads existed, if you were a Belgian startup and you wanted to reach potential customers in, say, India, you had basically two options: trade shows or agencies. Both expensive and slow. You'd spend months planning, fly people out, set up a booth, hope the right person walks by. Or you'd hire an agency to "handle international expansion" for a retainer that made your eyes water. Either way you were looking at tens of thousands of euros and a timeline measured in quarters, not weeks.
Then Meta (well, Facebook at the time) blew that open. And look, you can say a lot of negative things about Meta. I get it. But this specific thing was genuinely transformative for startups. Suddenly a few hundred euros and a weekend of experimentation got you in front of a global audience. The cost of reach didn't hit zero, but it dropped far enough that it stopped being the thing that killed you.
That unlocked a whole generation of companies that simply couldn't have existed before. Small teams with good products who would've burned through their runway on trade shows before ever finding product-market fit. The constraint wasn't their idea or their execution. It was that reach was gated behind money and connections they didn't have. Meta's ad platform (and Google's, to be fair) removed that gate. Over 10 million active advertisers are on Meta's platforms now. That's not a coincidence. It's what happens when you take something expensive and make it accessible.
the same thing is happening to software
Software development has been gated by the same dynamics for decades. You needed experienced engineers. They were scarce and expensive. Building anything meaningful took months, sometimes years, and required either raising capital to hire a team or paying consultancies with timelines that made the carbon certification story feel familiar.
AI is compressing that. Not to zero, let's not get carried away. And most non-technical founders aren't suddenly building it all themselves either. They're still pulling in someone with technical skills, a freelancer, a small team, a technical partner. But that engagement that used to cost tens of thousands and take six months? Maybe it's a two-week sprint now. Maybe it costs a fraction of what it used to. The technical person is still in the loop, they're just moving a lot faster.
I keep hearing people frame this as "AI will replace developers." I think that misses the point entirely. The more interesting story isn't about the people who already have developers. It's about the people who never could afford them.
what this actually unlocks
The ones I find most interesting in all of this aren't developers. They're the domain experts who've been sitting on real problems with real customers and no way to build. The logistics operator who knows exactly what software their warehouse needs. The healthcare consultant who's been sketching app ideas on napkins for years. And the compliance specialist who could automate half their industry if they could just ship something.
These people have something most technical founders don't: deep understanding of the problem and existing relationships with the people who have it. What they've never had is a way to turn that into software without spending a fortune.
That's the gap AI is closing. Not by making everyone a developer, but by making technical collaboration affordable for people who have the idea and the customers but not the engineering budget. A domain expert who used to need a full dev team can now work with a single technical person who's 5x more productive with AI tooling. The expertise still matters. The cost of accessing it just dropped through the floor.
I guess that's the parallel I'm trying to draw here. Online advertising didn't replace marketing departments, it unlocked a generation of founders who couldn't afford the old way of reaching customers. AI development isn't replacing engineering teams, it's making technical talent accessible to people who were previously priced out entirely.
And if the ads story is any guide, the pie might get bigger too. US ad spending was about $247 billion in 2000. By 2024, it was over $550 billion. Even corrected for inflation, the market grew by roughly 20%. The jobs in it changed (print shrank, digital exploded), but the industry as a whole only got bigger. Digital ads didn't carve up a fixed pie. They baked a larger one by letting millions of new businesses advertise for the first time.
I wouldn't be surprised if we see the same thing happen with software development.