SaaS Is Not Dead. But the Easy Version Might Be.
Everyone keeps saying SaaS is dead.
The argument goes like this:
AI can build software now. So why pay for Salesforce when you can ask Claude to build a CRM? Why keep subscribing when a prompt can generate the thing?
It sounds dramatic.
It also misses the point.
AI is changing software, but it is not deleting the need for real products.
Vibe coding works really well for tools used by a small number of people. It works for internal tools. It works for prototypes. It works for narrow workflows where the stakes are low and the users know exactly what they want.
It does not replace the software that entire companies rely on every day.
Workday, Salesforce, Gmail, and the rest of the enterprise stack are not just apps.
They are years of edge cases. Years of permissions. Years of integrations. Years of compliance, uptime, support, reporting, security, and boring reliability.
You do not reproduce that in a weekend with Claude Code.
That is the part the "SaaS is dead" crowd keeps skipping.
Production software is not just code.
It is maintenance. It is trust. It is uptime. It is knowing what happens when something breaks at 2 a.m. It is handling the weird customer who uses the product in a way nobody predicted. It is surviving the thousand tiny failures that never show up in a demo.
Most companies would rather pay for that than own it themselves.
But the doomsday version is not totally wrong
Some SaaS will die.
Not because software becomes useless. Not because every company will suddenly become a software company. But because the business model changes underneath it.
AI lowers the cost of building. That means more products can be made faster, by fewer people, with less capital.
It also means customers will have higher expectations. More alternatives. Less patience for bloated tools that solve a problem only slightly better than a script, a prompt, or a lightweight workflow.
The old SaaS playbook was simple:
- build a product
- sell seats
- add features
- raise prices
- expand into adjacent workflows
That playbook still works for some products.
But it gets weaker when the software is generic, replaceable, or only marginally better than what a smart team can assemble themselves.
We have seen this movie before
A good analogy is television.
Cinema did not die when TV arrived. But it changed.
People no longer needed to leave the house to watch moving pictures. That shifted the whole category.
Theaters had to adapt. Some formats faded. Some became more premium. Some became cultural events. Movies survived, but they reshaped around a world where entertainment could happen at home.
AI may do the same thing to SaaS.
Not all software disappears.
But the center of gravity moves.
Some products will get commoditized. Some will become more niche. Some will need to become deeply embedded in a workflow to survive. Some will need to become meaningfully better, not just a little better.
The companies that adapt will thrive.
The ones that assume customers will keep paying just because they always have will be in trouble.
I have seen enough to take this seriously
My last company, Crossfill, shut down.
That makes this whole conversation feel less abstract.
It is easy to talk about market shifts like they are strategy deck slides. It is different when you have lived through a product, a team, and a business reaching the point where they no longer work.
Shutdowns have a way of clarifying things.
They show you what actually mattered. They show you what was fragile. They show you how fast a market can move on from a product that once felt necessary.
That experience makes me skeptical of any argument that says a category is permanently safe.
It also makes me skeptical of the opposite extreme.
Nothing is guaranteed. Not even the tools everyone assumes are untouchable.
So what survives?
I think the winners will be the software that becomes harder to replace, not easier.
The products that survive will be the ones that are:
- deeply embedded in real workflows
- reliable under pressure
- trusted by teams, not just admired in demos
- expensive to rebuild correctly
- better than a prompt plus a weekend
The future is not no SaaS.
The future is probably less generic SaaS and more software that earns its place.
That means fewer companies winning by being the default. More companies winning by being indispensable.
It also means some products will need to stop thinking like software vendors and start thinking like workflow owners.
Not just "how do we add AI?" But "what are we actually irreplaceable for?"
That is a much harder question. But it is the right one.
AI is not an asteroid
That is the biggest mistake people make.
AI is not an asteroid that wipes out software.
It is a force multiplier.
It lowers the cost of creation. It increases the speed of iteration. It makes it easier to build alternatives. It raises the bar for what customers expect.
That is not the end of SaaS.
It is a pressure test.
Some companies will thrive. Some will get squeezed. Some really are dinosaurs.
The ones that survive will not be the ones with the loudest AI marketing.
They will be the ones that are actually useful, deeply embedded, and hard to replace.
SaaS is not dead.
But the easy version of SaaS might be.