The end state of the "end of software"

Ever since the launch of ChatGPT and the collective realization that LLMs are surprisingly good at coding, there have been murmurs about the “end of software.”

Chris Paik more or less started it with his claim that “majoring in computer science today will be like majoring in journalism in the late 90’s,” though I think Paul Kedrosky got there first with his incredibly prescient piece on Baumol’s cost disease and the economics of software development.

But in the past few weeks, these murmurs have become, well, deafening.

To start, you’ve got the sickeningly steep growth curves of app building platforms like Lovable and Bolt, which reached $17M ARR in 3 months and $20M ARR in 2 months, respectively. Who knows how high their churn will be (and both players are likely stretching the definition of “ARR”...), but you don’t reach numbers like that if you haven’t tapped into something new and meaningful.

Benn Stancil experimented with these tools (read about his experience in his excellent newsletter) and came to the conclusion that we might be witnessing the “end of YC.” Not because these tools will replace software engineers entirely, but because YC (and the entirety of modern Silicon Valley) is built on the conceit that people who can build win, and only a select few can build. As Stancil writes:

“Flagship incubators like Y Combinator are built on the thesis that a smart kid with a computer and summer internship at Goldman Sachs can outwit all of American Express. That’s not because the kid understands the needs of payment processors better than people at American Express, or has better ideas than they do; it’s because the kid can build their idea.”

But what if everyone can build their idea?

The most common counterargument I hear is some version of - but Lovable and Bolt and Cursor and v0 aren’t good enough to build a payment processor! And yes, that’s true. But I think that misses the point, because if these previous predictions are even close to right then software is going to look and feel fundamentally different.

As I wrote a few months ago, perhaps the right analogy to the cost of code going to zero is something like high quality cameras becoming ubiquitous. Today, we all have cameras that young Spielberg could only have dreamed of. But we haven’t all become Hollywood directors. And with Lovable, Bolt, Cursor and all the rest, we all won’t become “proper” software engineers either. But spend some time scrolling YouTube, Instagram or TikTok, and you have to agree that smartphone cameras have led to an absolute explosion in the variety of audiovisual content.

And that leads me to the end state of the “end of software,” courtesy of Every’s Dan Shipper. Not only has software creation been democratized, Shipper argues, but “software is now content.”

Shipper posted that three months ago, but now you see this formulation everywhere.

I find this prediction particularly interesting because of how uncomfortable it makes me. It just barely makes sense. You really have to squint to see it. And that makes me think it’s closer to right than not.

Software is, at its most basic, the combination of logic, interface and interactivity. How is that not a compelling medium for self expression?

And AI-written software has a bunch of additional, unique affordances that make it a particularly powerful platform for creativity and creation.

As an example, AI-written software can be single use and disposable. We think of software as requiring iteration and upkeep. Not so in a world where a simple piece of software can be written for (essentially) free (almost) instantly. Dispose of it when you’re done and build it from scratch the next time.

Here’s another one. AI-written software costs so little to make that it can profitably serve much smaller markets. There’s been a lot of talk of the “one person billion dollar startup.” But what about the opposite? I think we’re more likely heading toward something far more interesting: a million micro-startups, each serving hyper-specific needs with precision and care.

Perhaps the true end state of 'the end of software' isn't a world without software engineers or code, but a world where software becomes as malleable and expressive as any other creative medium. To me, “software as content” means that applications won't just be tools we use, but expressions we share - unique creations that reflect individual needs, perspectives, and aesthetics. In this end state, the boundaries between developer and user blur completely, and software becomes another canvas for human creativity - disposable, personalized, niche-focused, and continuously reinvented. We're not just changing how software is made; we're fundamentally reimagining what software can be. And that’s a future worth investing in.