The SaaS-pocalypse Is Not What You Think It Is

Over the last year, roughly £300 billion has evaporated from SaaS market capitalisations.

Household names are down 15 to 40 percent. Headlines are full of panic. Commentators are confidently declaring that “AI is replacing SaaS.”

That narrative is comforting because it is simple.

It is also wrong.

What we are seeing is not AI killing SaaS. It is AI exposing what SaaS never really solved in the first place.

SaaS Did Not Win Because It Was Perfect

It Won Because It Was Cheaper Than the Alternative

For over a decade, SaaS dominated because custom software was slow, expensive, and risky.

If you wanted something tailored to your business, you needed engineers, long timelines, maintenance contracts, and a tolerance for pain. SaaS stepped in and said:

• Use our product
• Adapt your workflow
• Pay per seat
• Accept the trade-offs

That deal made sense at the time.

SaaS vendors sold “flexibility” while quietly locking customers into rigid workflows. They promised “customisation” through settings panels and feature flags most users never touched. They charged per user for software that scaled at near-zero marginal cost.

It worked because the alternative was worse.

AI just removed that constraint.

AI Did Not Make Software Better

It Made Custom Software Cheap Again

This shift is not about smarter apps or better UX.

It is about economics.

A founder can now spin up a bespoke CRM in an afternoon.
A solo operator can build a task system that mirrors how they actually think.
A small team can create internal tools faster than they can negotiate an enterprise contract.

The cost curve collapsed.

What used to require a team now requires intent and a prompt. What used to be a six-figure build is now a weekend experiment.

That changes everything.

This Is the Docker Moment for SaaS

There was a time when scaling software meant massive infrastructure, dedicated ops teams, and complex deployments.

Then containers arrived.

Docker did not eliminate cloud providers. It made everyone rethink what was truly necessary. Suddenly, you could do more with less if you designed things properly.

AI is doing the same thing to software development.

It is not replacing platforms.
It is shrinking the surface area where platforms are actually needed.

The Real Shift Is Economic, Not Technical

Ask yourself these questions honestly:

Why pay £50 per user per month when you can build exactly what you need for £50 total?
Why bend your workflow to fit software when software can adapt to you?
Why accept “good enough for everyone” when you can have “perfect for you”?

Once you see it that way, the market reaction makes sense.

Feature-based SaaS was always fragile. It survived because customers had no viable alternative. That monopoly on convenience is gone.

What Still Matters and What Does Not

The SaaS companies that survive will not be the ones with the longest feature lists.

They will be the ones solving problems that are genuinely hard.

Things like:

• Network effects that compound over time
• Compliance frameworks that are painful to recreate
• Deep integrations across fragmented ecosystems
• Real-time collaboration at massive scale

These are not inconveniences. They are structural advantages.

Everything else is becoming a commodity.

If your product is essentially a workflow wrapped in a UI, you are now competing with a motivated user and an AI assistant.

That is not a fair fight.

The New Founder Question

For years, founders asked, “How do I build a SaaS moat?”

That is the wrong question now.

The better one is:

What can I build myself in a day, and what genuinely requires a platform?

The operators winning right now are ruthless about this distinction. They self-build aggressively and only reach for SaaS when the problem crosses a real complexity threshold.

The Window Is Closing

This transition period will not last forever.

Right now, there is still institutional inertia. Procurement cycles still exist. Many teams are still locked into old assumptions.

But that gap is closing fast.

As AI-assisted building becomes normal, the idea of paying recurring fees for simple internal tools will feel absurd.

The future does not belong to whoever ships the most features.

It belongs to whoever can deploy the right solution the fastest.

And increasingly, that solution will be built, not bought.