Think about how much content the internet has produced, and the order it arrived in. First the early web — a few thousand hand-coded pages. Then websites, as every business went online. Then blogs, when publishing dropped to zero cost and anyone could write to the world. Then social media, when everyone published constantly, from a phone, for free. Then SaaS and app stores, which turned software itself into mass-produced content. Each step multiplied the volume of data in the world by orders of magnitude.
And then AI arrived and did what it does: it drove the marginal cost of creating content to roughly zero. Text, images, audio, video — all of it can now be produced faster than anyone could ever consume it. The curve that had been climbing for thirty years went vertical.
That's the story everyone is talking about. But it's the setup, not the punchline.
AI doesn't just write content. It writes software.
For decades, the limiting factor in software was never ideas — it was the cost of building. Software took scarce, expensive engineers, months of work, and a real budget. So organisations only built software when the payoff clearly justified the spend. Everything else stayed manual, lived in a spreadsheet, or simply never got done.
AI collapses that cost the same way blogging collapsed the cost of publishing. When a working application can be described instead of hand-built, far more people build far more software, for far smaller reasons — including software that lasts a day, serves one person, or solves a problem that was never worth a project before. The exact same dynamic that exploded content is now pointed at software itself.
And software is not just another category of content. Software is the thing that now runs everything. So a software explosion isn't "more apps in an app store." It's everything getting more software — more of it, more specific, more disposable, and increasingly made by people who were never developers.
What a software explosion changes
The content explosion didn't just give us more content. It gave us the attention economy, recommendation algorithms, misinformation, and the creator. A software explosion has its own second-order effects, and most of them rhyme with what content already taught us:
Software becomes disposable. When an app costs minutes, you build one for a single task and throw it away — the way we went from books we kept to posts that vanish in a day. Distribution becomes the bottleneck. When anyone can build anything, having a solution stops being special; being found does. Trust gets expensive. A flood of unvetted, machine-written code makes knowing what's safe and correct the genuinely hard work — software's version of fact-checking. Maintenance debt explodes, because code is cheap to write and expensive to own, and every generated app is a future liability to secure and patch.
And the deepest shift: the job moves from building to deciding. When implementation is abundant, the scarce skills become judgment and verification — framing the right problem, deciding what "good" looks like, and checking that the output actually delivers it. Moats stop being about who can build a feature and start being about proprietary data, trust, distribution, and the taste to know what's worth building at all. When publishing got easy, the valuable role shifted from typesetter to editor. Software is about to make the same move.
See it for yourself
We built a free interactive tool — The Software Explosion — that lets you play the curve forward through each era of the internet, watch the volume of content go vertical when creation hits zero cost, and then see the same regime change applied to software. It also breaks down the eight shifts a world of abundant software brings, each with its parallel from the content era.
The organisations that did well out of the content explosion weren't the ones who made the most content — they were the ones who understood what abundance changed and positioned for it early. Software is the next one. It's worth getting ahead of.