SaaS is dead, long live SaaS

There was a time when buying software meant buying a box, a CD, a serial number you would lose in a drawer. Then came SaaS, the “service” era, where software lived in the cloud and you rented it instead of owning it. For a while, that seemed like progress. Fewer IT headaches, automatic updates, predictable revenue for the vendors, and the illusion of flexibility.

But the illusion is cracking.

Every few weeks now, another company quietly admits it is trying to reduce its SaaS footprint. Too many tools, too many logins, too many subscriptions. The cost of convenience has finally caught up with everyone. What began as freedom from software maintenance has turned into a kind of digital mortgage.

And most companies have realised something uncomfortable. They are paying for 100 percent of a product and using maybe 20 percent of it.

The fatigue of feature bloat

Walk through any enterprise stack today. You will find Salesforce, ServiceNow, Workday, HubSpot, Monday, Notion, Slack, and a dozen others all coexisting like well-meaning but overpaid roommates. Each has overlapping features. Each has its own pricing plan, its own roadmap, and its own cheerful account manager reminding you that next quarter’s update will finally solve your problem.

The truth is, most of those roadmaps are six-month exercises in incrementalism. Companies cannot afford to wait six months.

They have customers today, fires today, inefficiencies today. Somewhere along the way, the old SaaS model stopped fitting how fast businesses actually move.

Nadella’s quiet bombshell

Satya Nadella put it bluntly earlier this year:

“Business applications as we know them will probably collapse in the agent era. They’re essentially CRUD databases with some business logic, and once that logic moves into the AI tier, people will start replacing the back ends.”

It did not make as much noise as it should have, but it was a warning shot.

When the CEO of Microsoft says business apps themselves will collapse, you know something fundamental is shifting. The logic, he said, is moving. And when logic moves up a layer, the power moves with it.

That single thought ends twenty years of SaaS orthodoxy.

The AI build-your-own era

For the first time, companies no longer need to buy enormous platforms to run their operations. They can build their own.

AI tools, low-code frameworks, and modular APIs have made it possible for small internal teams to create what they actually need, not the bloated version, not the roadmap version, but the working version.

You need a ticketing tool? You can build one.
A CRM? You can build that too.
HR, marketing automation, project tracking, all within reach of a handful of good developers or an external micro-team that knows what it is doing.

The point is no longer to buy software that does everything. It is to build software that does just enough.

And it turns out just enough is not only cheaper but faster and more aligned with how companies really operate.

The Klarna example

Klarna is a case study in this quiet revolution. They reportedly shut down more than a thousand SaaS tools, including Salesforce, and built internal replacements powered by their own AI stack.

They did not do it for fun. They did it because it made sense. Too many overlapping services, too many contracts, too much data scattered across vendor silos.

When you own the core of your workflow, you also own your data, your logic, and your speed. You can change things without waiting for a vendor’s next quarterly update.

Other companies are starting to do the same. Not all of them are public about it, but the pattern is emerging.

The old economics don’t hold

Traditional SaaS was built on two ideas.
One, recurring revenue is good for predictability.
Two, feature expansion keeps customers hooked.

Both made sense when software was scarce and development was slow.

But in a world where small teams can build production-grade systems in a few months using open APIs and model-driven agents, those ideas start to fall apart.

Why pay millions a year for Salesforce when a five-person team can spin up a tailored CRM that fits your exact sales workflow and plugs into your internal data sources?

Why depend on ServiceNow for ticketing when you can replicate its core workflow using an internal agent, a chat interface, and a simple database?

The scale advantage of large SaaS vendors is turning into a liability. The complexity they built to serve everyone prevents them from serving anyone particularly well.

The rise of small service builders

This is where things get interesting.

The next wave of software will come not from massive platforms but from small, independent service builders. Teams that can walk into a company, understand its process, and build custom internal tools in 90 days.

These are not SaaS startups in the old sense. They do not build for a global market. They build for one company at a time, using common components and frameworks that can be reused elsewhere.

Think of them as digital carpenters. Fast, efficient, and paid for outcomes, not subscriptions.

For clients, it means lower costs, better fit, and zero lock-in. For the builders, it means constant work, high margins, and no need to raise a Series A to survive.

The economics flip entirely.

What happens to the big players

The large SaaS vendors will not vanish overnight. They will double down on ecosystems, compliance, and integrations, the areas that still create friction for internal builds.

They will talk about being AI-first and platform-extensible, which mostly means adding an LLM plugin to an interface that still takes six clicks to create a record.

Their contracts will become modular. Their pitch decks will start including the word flexibility. Their investors will ask why renewal rates are dropping.

The smart ones will reposition as infrastructure providers rather than applications. The rest will slowly become utilities.

A quiet correction

It is tempting to declare the whole industry dead, but it is more accurate to say it is being corrected.

SaaS made sense when development was slow and expertise was scarce. AI has removed both of those constraints. Now the natural equilibrium is shifting back toward ownership, not of servers, but of logic.

Companies are realising that their competitive advantage does not come from using the same software as everyone else. It comes from shaping software around how they actually work.

So they are taking back control.

The final toast

The end of SaaS is not a funeral. It is a return to sanity.

If you are still locked into a five-year deal for a tool you barely use, the joke is on you. You bought a mansion because the salesman told you it was cheaper than renting a room.

Meanwhile, somewhere in Stockholm or Singapore, a five-person team is quietly building a better version of it in three months, faster, cheaper, and actually useful.

SaaS is dead. Long live SaaS.

The next generation will not be software you rent. It will be software you shape.