Is software dead? No. The thin layers are.

In December 2017, I tweeted this: “Finally the day has come. My father asked me what is #bitcoin? Had fun explaining him.”

I meant it as a milestone. Bitcoin had made it to the family dinner table, which felt like the moment crypto went mainstream. And the conviction around me back then was hard to overstate. Fiat was finished. Banks were finished. The whole financial system would be rebuilt on the blockchain.

That didn’t happen. What did happen is that adoption grew many times over, and crypto found its real place in the world. Smaller than the hype, bigger than the skeptics predicted.

I think we’re watching the same movie with SaaS. I put my thoughts into a short thread today, quoting that old tweet:

That was the moment crypto went mainstream. Bitcoin had made it to the family dinner table.

I think we're watching the same movie with SaaS.

Your grandma vibe coding an app on Claude or ChatGPT doesn't mean software is dead. It means the barrier to building software is collapsing.

Those are very different things.
— Nikunj Thakkar (@nikunjness) October 13, 2026

This post expands on it.

Your grandma vibe coding something on Claude or ChatGPT once in a while doesn’t mean software is dead. It means the barrier to building software is collapsing. And that’s a very different thing.

Full disclosure: I run a software company, so I have skin in this game. Read everything below with that in mind.

The debate, in two tweets

Over the last few days the “software is dead” argument got loud again, and these two posts sum up the poles better than anything I could write:

Software is dead

It's now the era of personal software and everyone just builds what they need whenever they need it I think
— @levelsio October 11, 2026

the fundamental basic economics misunderstanding that ppl have esp on here is that they think normal ppl need to use models to build their own software to be disrupted. they obviously don't & they obviously won’t.

but even if just 1% of the population can build software, that's enough to flood the entire supply side with millions of new products collapsing the cost to near zero.
— signüll (@signulll) October 11, 2026

I agree with the second one more than the first, and I think it’s the most useful way into the whole debate.

“Software is dead” is really three separate claims stuck together, and they don’t all age the same way.

Three claims hiding inside “software is dead”

Claim one: simple software becomes nearly free. True, and it’s already happening. You don’t need most people to start building. You only need enough people who can, because supply doesn’t scale with the number of customers, it scales with the number of builders. Once a few percent of people can clone a basic tool in an afternoon, the price of that tool heads toward zero. If your product can be described in one sentence and rebuilt over a weekend, you’re exposed.

Claim two: everyone will build their own software. I don’t buy this one. I’ve shipped working products in a weekend, and I wrote about how building is cheap now, but I’ve also watched how real people use software. Most of them don’t want to build, maintain and debug anything. They want a thing that works and someone to call when it doesn’t. “Personal software” will be real for a small group of people who enjoy it. For everyone else, the default remains pressing a button to install something that already exists.

Claim three: therefore the whole industry collapses. This is where I get off. Cheap supply of simple software is a problem for simple software. It’s not a verdict on products that own a whole workflow, hold years of a customer’s data, or sit inside compliance and security requirements. Those are not weekend projects, and they don’t get cheaper to own just because the first draft got cheaper to write.

The market’s own reaction this year shows how confused the thinking is. Software stocks fell hard on fears that AI would replace everything, in the same weeks that other investors worried AI spending would never pay off. Those two fears can’t both be right. To me that looked more like a mood than an analysis.

What I think happens next

1. Thin layers of software will collapse. When anyone can build a basic CRM, dashboard, landing page or internal tool in hours, standalone products with little differentiation face enormous pressure. This is the part of “SaaS is dead” that is simply correct, and it’s the part I’d take most seriously if I were a founder.

2. End-to-end workflows will become more valuable. Businesses don’t want fifteen tools. They want outcomes. They want to acquire customers, convert them, serve them, collect payments, and understand what’s working. The platform that owns the whole workflow has a stronger position than the tool that handles one tiny piece of it. This is also where AI agents help rather than hurt: an agent needs somewhere to act, and context to act with.

3. Platforms will eat adjacent layers. AI makes it cheaper to expand into neighboring workflows. A product that once solved one problem can move upstream or downstream and solve several related ones. Expect more consolidation, more vertical software, and fewer reasons to keep disconnected tools around.

4. Software won’t disappear. The definition of software will change. More people will build it. More businesses will use it. More workflows will be automated. But the value shifts away from shipping features and toward owning workflows, distribution, proprietary data, trust, and business outcomes.

I also think the pie gets bigger, not smaller. When software stops assisting with work and starts doing it, it competes for much more than software budgets. The money moves from seats to outcomes, and there’s more of it.

On “the moat is audience”

A related argument has been going around among indie builders: if anyone can build your app, the moat becomes content, audience and personality. Here’s the thread where Pieter Levels (@levelsio) made that case:

True

I think you have the real economic effects of AI companies sucking up entire industries now (a lot where indie hackers operate)

I do think the SaaSpocalypse is at least partly real

So indie hackers are pivoting toward content, audience and distribution to find some moat, essentially becoming influencers

[…]

So the distinguishing element becomes personality kinda!
— @levelsio September 1, 2026

I agree that distribution matters more when building is cheap. Building was never the hard part, and now it’s the easy part. I’m writing on a personal blog partly for that reason.

But I think audience is a way in, not a moat. It gets people to try you. It doesn’t make them stay. What keeps a customer is the unglamorous stuff: their data lives in your product, your product sits inside their daily workflow, and they trust you when something breaks at 2 a.m. Audience gets you the first purchase. Workflow, data and trust get you the renewal.

It also depends on who you sell to. A solo builder selling a $9 tool to other builders is competing with whatever those builders can prompt in an afternoon, and there, personality may well be all that’s left. A company selling into a team that needs permissions, audit logs, integrations and a support line is competing with something else entirely. The same wave can crush one and barely touch the other. I suspect a lot of the “software is dead” mood comes from people who live mostly in the first world.

Switching costs and maintenance are the quiet moats

Two things get underrated in this debate.

Switching costs. I’ve worked on the vendor side of enterprise software. Nobody rips out a CRM or finance system because a demo got cheaper to build. It means retraining people, migrating years of data, and taking on a risk nobody wants to own. AI can copy the features. It can’t copy the trust of a system that has run your business for five years.

Maintenance. Generating code is cheap. Owning it is not. Security patches, compliance, uptime, edge cases, and a new hire trying to understand what a model wrote eight months ago. It’s the same reason companies pay for hosted versions of open source tools they could run for free. They’re paying to not think about the upkeep.

A simple test for your own product

If you run or are building a SaaS product, here’s the gut check I’d use:

Could a smart customer rebuild the core of it in a weekend, and be happy with it?
Does your product get smarter or more valuable as their data accumulates, or does it just display it?
Is it a system of record, or a convenience layer on top of one?
Would an AI agent need your product to do its job, or could it go around you?
Do you already own the whole job, or just one tiny step in it?

If the honest answers are uncomfortable, that’s useful. It’s far better to find out now than from a churn report.

So, is software dead?

No. Crypto didn’t replace finance, but it didn’t disappear either. It found its place, smaller than the hype and bigger than the skeptics predicted. Software is heading the same way.

The thin layers are under pressure. The bigger opportunity is just getting started.

If you think I’m wrong, I’d like to hear where. Reply on LinkedIn or X and tell me which layer of the stack you’re standing on.