Nobel Laureates, Economists, & the death of SaaS

There’s a shallow version of the “SaaS apocalypse” debate.

It goes like this: AI lets anyone build software. Therefore SaaS dies.

Or its mirror image: Enterprises can’t maintain or secure custom software. Therefore SaaS survives untouched.

Both sides miss the deeper question: 

Why does packaged software exist in the first place? To answer that, you have to go back to Coase, Williamson, Hart – the economists who studied how organizations actually coordinate labor, capital, and tools. Their work is always relevant and it guides how Moative thinks about the evolving economic model of buying software and labor.

SaaS is a governance choice

Ronald Coase’s core insight was simple: Firms exist because markets are not frictionless. Searching, contracting, negotiating, monitoring – all of these cost money. When the cost of using the market is high, firms internalize. When the cost drops, they outsource.

For decades, enterprise software reduced coordination costs enough that buying a packaged system was cheaper than building internally. SaaS was just the latest governance structure: a vendor took on maintenance, updates, uptime, security, and adaptation in exchange for subscription fees.

What AI does is collapse the production cost of software, not the coordination cost of outcomes. This is what most SaaSpocalypse arguments miss. SaaS is not just software. It's not just a distribution model. It is also a governance substrate secured by contracts.

Writing code, wiring APIs, generating UIs – these are now cheap. Ensuring systems behave correctly across time, regulations, edge cases, future states, and adversarial environments remains expensive. That cost is baked into SaaS contracts.

So does that mean that SaaS will persist because someone's neck is on the line?

When risks increase, governance matters

My next favorite work is that of Oliver Williamson. He answers the question well. Williamson's core idea was simple: when something is risky and hard to undo, you don’t just buy it like a commodity – you govern it more tightly. Much commodity SaaS is at risk simply because there are no governance risks. I don't mind building an internal "task list" or even a project management software because neither creates governance risks (in our context).

When customizations increase, future states become unpredictable, and dependency is high, the need for control increases. The need for governance increases. Organizations will build or buy depending on their governance needs. Hybrids evolve. But one thing is clear: What we call "orchestration" systems are really "control" systems. These "control" systems speak the same language across organizations. They will eventually take the burden shouldered by SaaS and along with it will go away the assurance premium. 

How many control layers will an organization need? One. How many SaaS tools do they use? Several dozen. This IS the argument for why SaaS dies. 

Features without governance assurance offer little value, which explains the market's current pricing of SaaS companies. SaaS multiples will compress even further, when control layer gets owned by a central platform.

The market will reward control layer companies. There will be orchestration platforms of choice. Consulting companies will use them and assure governance. Both will survive and thrive. SaaS is structurally a point solution. Governance is anything but a siloed activity. Therein lies the reason why SaaS as an 'assurance service,' matters less.

So is all SaaS dead? No. The more localized governance you can provide, the more valuable the SaaS still is.

Why did we never fire the person who hired IBM?

Hart’s property-rights theory explains why firms integrate: contracts cannot specify every future state. Whoever owns the asset holds residual decision rights when surprises occur. This is why we supposedly never fired anyone for buying IBM or Salesforce. If correctness under future states cannot be fully contracted, someone must hold ongoing control.

This is why core systems of record won’t vanish. Firms will still want specialized vendors holding residual control over payroll, ledgers, identity, compliance infrastructure – areas where failure has high blast radius and adaptation must be continuous.

But outside that core, where neither governance nor assurance matters, all hell breaks loose for SaaS.

The kicker is that much of today's work is done by labor, not software. Labor work is rarely as governed as software work. So custom-built AI can consume a lot of this ungoverned work and bring it into the control plane for the first time. Each firm's 'labor work' is unique (at least that is what they believe), so the advantage goes to custom-builders who can assure governance.

What replaces the old SaaS model

The old model was simple: Vendor builds → customer uses.

The new models threatening SaaS are not custom software. The new models will separate software production from software assurance. Control layers will take away the assurance/residual control premium that SaaS charges. Where the assurance does not matter, custom-build will eat into SaaS. Where SaaS never existed, replacing labor with AI will be custom. In all of these cases, the control layer wins. In some cases where function-level governance is highly needed (like treasury, accounting, compliance, etc.), SaaS vendors may offer localized control layers and thereby keep their premium.

The emerging structure

Three stable layers form:

1. Certified core systems (where SaaS may still survive)
Finance, identity, compliance, payments.
Highly governed. Vendor-operated or tightly controlled.

2. Governed custom layer (where new platform winners may emerge)
High-value workflows built rapidly but running inside policy, audit, and security frameworks.
Largest growth area.

3. Long tail internal tools and labor replacement (where consulting will win)
Cheap, disposable, locally optimized tools. Big, chunky workflow compression that come from from eliminating labor inefficiencies (Service as software).

The profit pool migrates to whoever controls the second layer – the governance and assurance plane. In our opinion, it's the platform and the service provider that uses the platform to deliver assurance, not SaaS.

Enjoy your Sunday!

Top reads