Blogs and YouTube had democratized writing and videos over the last two decades, and AI could enable software to follow a similar path.
That, in essence, is the argument AngelList co-founder and prolific tech investor Naval Ravikant has been making — and it is one that is becoming harder to dismiss with each passing month. “Software will proliferate just as videos, music, writing did,” Ravikant posted on X. “The market structure will shift from a “fat middle” to mega-aggregators and a long tail. It’ll be a slower process due to network effects, but many traditional vendor lock-ins will get eaten by AI,” he added.
The Historical Pattern: From Gatekeeping to Mass Creation
The analogy Ravikant draws is not incidental. Each major creative medium in history has followed a recognisable arc: it begins as the domain of specialists, then democratizes rapidly once the tools of production become cheap and accessible, and ultimately fractures into a handful of dominant platforms at the top and a sprawling long tail of niche creators at the bottom.
Writing did this with blogging. Video did it with YouTube. Music did it with SoundCloud, Spotify, and digital audio workstations that turned bedroom producers into recording artists. In each case, the transition was initially resisted, then grudgingly accepted, and finally celebrated as a creative explosion. The “fat middle” — the mid-tier publishers, mid-size studios, regional software vendors — took the worst of it. What survived were the giants and the niches.
Software, Ravikant argues, is next.
Vibe Coding and the New Creative Class
The clearest early evidence of this shift is what the industry has taken to calling vibe coding — a mode of software creation where intent matters more than syntax. Legendary music producer Rick Rubin, drawing from his own experience with punk rock, has described it as lowering the barriers of deep technical expertise so that anyone with something to say can say it in code.
The parallel to music is unusually precise. Just as punk stripped away the need for conservatory training, vibe coding strips away the need for years of engineering education. What you need is an idea and the ability to express it. AI handles the rest — or at least, enough of the rest to get something working.
NVIDIA’s Jensen Huang has gone further, saying that nothing would give him more joy than if his software engineers never had to write code at all — freeing them instead to spend their time discovering and solving new problems. That framing redefines what engineering actually is. It is not the act of writing code. It is the act of understanding what needs to be built and why.
This is not a distant vision. GitHub data already shows an explosion in coding productivity driven by AI tools, with code pushes growing roughly 35% year-on-year in the US in 2026 — a sharp uptick that tracks closely with the widespread adoption of AI coding assistants. Tools like Cursor, Claude Code, and Codex have become staple parts of developer workflows, and the downstream effects are beginning to show up in the data.
The Market Structure Shift: Fat Middle, Long Tail, Mega-Aggregators
Ravikant’s prediction about market structure is where the analysis gets genuinely consequential for the business world. The history of media democratization did not produce a world of equals — it produced Spotify and YouTube, on one side, and millions of individual creators on the other. The labels and studios in between were hollowed out.
The same structural logic applies to software. As the cost of building functional software approaches zero — a thesis Klarna’s CEO Sebastian Siemiatkowski has argued forcefully — the moat that once protected mid-market SaaS vendors begins to erode. Why pay a five-figure annual subscription for workflow software when a comparable tool can be generated on demand, tailored to your exact specifications?
Siemiatkowski has gone as far as predicting that software valuations could collapse from the 20–30x price-to-sales multiples that the industry has historically commanded, down to utility-like multiples of 1–2x. That is not a marginal repricing. That is a structural reclassification of what software companies are.
What survives, in this framing, is what always survives: the platforms with network effects strong enough to be self-reinforcing, and the highly specialised niche products that serve specific use cases better than any general-purpose AI can replicate. The fat middle — the generalist SaaS players without a defensible data moat or dominant network — faces the most exposure.
Microsoft’s CTO Kevin Scott has predicted that 95% of code will be written by AI within five years. Anthropic’s Dario Amodei has offered an even shorter timeline. These are not fringe views. They are increasingly the consensus of the people building the tools.
Vendor Lock-In: Durable, But Not Invincible
Ravikant is careful to add a qualifier. The transition, he notes, will be slower than what happened to writing or music, precisely because of network effects. Software is not consumed in isolation the way a blog post or a song is — it is embedded in workflows, integrated into data systems, and surrounded by years of organisational muscle memory. These switching costs are real, and they have historically been the best defence a software company could have.
But AI is beginning to eat into even that advantage. Klarna’s data architecture decisions, for instance, have been shaped by the belief that data portability will matter more as code becomes cheap. Once AI can migrate data schemas and port records between platforms with minimal human involvement, the second layer of the SaaS moat — data stickiness — comes under pressure too.
The EU’s Data Act, which came into force in 2024, is a regulatory signal in the same direction, requiring data portability in certain contexts. Regulatory and technological pressure are, for once, moving in the same direction. That is rarely a good sign for incumbents counting on lock-in.
What It Means for the Industry
Ravikant’s framework is not a prediction of doom for software companies. It is a prediction of stratification — the same stratification that hit every creative industry before it. Some players will consolidate into near-monopolies with platform power that makes them near-untouchable. A large number of new entrants will flourish in the long tail, building highly specific tools for highly specific problems, at costs that would have been unimaginable five years ago.
The companies that should be worried are those that sit in between: large enough to have significant overhead, but not large enough to have the network effects or proprietary data that justify premium pricing in a world where software itself is becoming a commodity.
Salesforce’s decision to stop hiring software engineers, and its subsequent cuts to customer support roles, are a preview of what operational restructuring looks like in this environment. The question is not whether software will proliferate. By most available measures, it already is. The question is which business models, built on the assumption that software was scarce and hard to build, will survive the transition.
Naval Ravikant’s bet is that the answer rhymes closely with what happened to music, writing, and video before it. The tools got cheap. The creators multiplied. The middle got hollowed out. And the platforms that understood the new dynamics early enough became the new infrastructure layer of the economy. Software is on the same road. The only variable is timing.