To see where software is headed, it helps to look at how technology reshaped other industries. History doesn’t repeat, but it often rhymes.
Before the internet, media worked very differently. Producing content was costly—you had to pay writers, editors, photographers, and distributors. Because making content was expensive, it had to generate revenue. Consumers paid through newspapers, magazines, books, cable, and pay-per-view. Warren Buffett famously loved newspapers, and for good reason: predictable subscription revenue and strong local monopolies.
Then the internet arrived. Media companies saw it as a way to lower distribution costs and expand reach. What they didn’t anticipate was that the internet would drive distribution costs to zero—and, over time, the cost of content creation to zero as well. User-generated content exploded. When content is free to make, it doesn’t need to earn money. That shift triggered a Cambrian explosion: a photo of a coffee cup could reach millions—or no one—and the economics still balanced. The flood of content overwhelmed supply, creating the need for platforms that organize attention and surface what matters. Enter the rise of user-generated content platforms.
These platforms blindsided traditional media. Suddenly, they were fighting for the same user attention but with much higher costs. Every salaried journalist or producer increased their disadvantage against a tidal wave of free creators. Media companies lost their structural edge, and the value shifted entirely to the platforms controlling distribution.
Today, software looks a lot like pre-internet media. It’s expensive to build, maintain, and distribute. Because it costs so much, it must earn revenue—through licenses, SaaS subscriptions, seat-based pricing. Software has enjoyed enviable margins: near-zero distribution costs and 90%+ gross margins.
But software is expensive because developers are expensive. They are translators, turning human intent into computer logic. Large language models are proving they can do this translation at scale, faster and cheaper than humans. As LLMs drive the cost of creating software toward zero, the same question emerges: what happens when software no longer has to make money? The answer will echo media’s story—a Cambrian explosion of software.
Vogue wasn’t replaced by another glossy magazine; it was replaced by tens of thousands of influencers. Likewise, Salesforce won’t be replaced by another giant CRM. It will be replaced by a constellation of tools and services that solve the same problems more flexibly. Just as media companies gave way to platforms, software companies will give way to new distribution gatekeepers.
SaaS metrics like ARR and magic numbers are artifacts of the old world, where high development costs acted as a moat. But that moat is draining. LLMs are unleashing the same corrective forces that transformed media. In hindsight, majoring in computer science today may look a lot like majoring in journalism in the late 1990s.