There's No Product Because There's No Software

As someone whose livelihood is intertwined with having the right business model, I’m spending a lot of time thinking about what the future holds for software and software products. Sidu Ponnappa’s “There Is No Product” piece is making the rounds and it’s worth reading. His core argument is that AI is converting software from asset to inventory. If anyone can build it in a weekend, you can’t amortize it. There’s nothing to sell.

He’s right. But I think the discourse is stuck on the wrong transition.

So much of the discussion is on build vs buy economics. Can a company build their own HRMS cheaper than licensing one? At what point does the build cost cross below the subscription cost? Sidu says it already has for most software. I agree.

But building is only part of it. There’s still the attention cost.

I’m not going to build my own time tracking tool. Not because I can’t, but because the cognitive overhead of evaluating, building, and maintaining a custom one exceeds the cost of just paying someone else for a $10 seat license and moving on.

That’s real, for now. But I think the “SaaS is dead” discourse is stuck on this near-term economics debate instead of following the logic to its conclusion: what we think of as “software” collapses into something ephemeral. Software will be generated dynamically as a side effect of intent. You won’t “build a time tracking tool” or “buy a time tracking tool.” You’ll say, “I need a way for my company to track time” and the AI materializes the capability. Anyone who has experienced Claude or ChatGPT whip up a quick website to explore an idea has gotten a glimpse of this.

When capability materializes from intent, there’s no product because there’s no distinct artifact called “software.” It’s just a transient side effect of what you’re trying to do.

What survives? Two layers: compute infrastructure and AI itself (which evolves into a new kind of operating system). The application layer where billions of dollars currently flow to tools, platforms, and SaaS products? Most of it evaporates unless it’s intimately linked to one of the big three things AI will need: storage, compute, networking.

If your business has always centered on making software tools for others, the uncomfortable question isn’t build vs. buy. It’s whether there’s going to be anything left to sell at all.