The Beginning of the End for Big SaaS
Engineering headcount and capital are no longer the constraint.
I started learning to code in fifth grade.
By the time I finished my Master’s in Computer Engineering and joined early-stage startups, software followed a clear economic law: engineering headcount was the constraint. Building complex systems required teams. Teams required capital. Capital justified SaaS.
That constraint shaped the modern software industry.
When I began my career, provisioning infrastructure meant writing custom Azure scripts to spin up virtual machines, execute data workflows, and tear everything down manually. Later, orchestration tools like Apache Airflow reduced operational overhead. Then platforms like Databricks abstracted even more complexity.
Each wave made engineering more efficient.
But the core constraint remained: you still needed engineers.
That constraint is weakening.
Modern AI systems and autonomous coding agents have materially collapsed the marginal cost of producing software. An experienced engineer, amplified by AI, can now perform the functional output that previously required an entire team. Scaffolding, integration, test generation, refactoring - tasks that once took weeks now compress into hours.
The bottleneck is no longer code production.
It is judgment: deciding what should be built, what should not be built, deciding which complexity a business should own, and which it should outsource.
Engineering scarcity shaped SaaS.
AI weakens that constraint.
When constraints shift, business models shift.
Recently, while completing my MBA at Duke, I helped my wife acquire and operate her first dental practice. I was involved in financial diligence, validating valuation against patient base realities, and ensuring we were buying a viable asset rather than a fragile one.
Post-acquisition, I led the rebrand, migrated the practice to a simpler and more cost-effective patient management system, and modernized the underlying IT infrastructure - upgrading aging out of date machines, tightening security, enforcing compliance and reducing vendor complexity.
Working inside a real operating business clarifies something quickly: most practices are not buying software because it is optimal. They are buying it because building alternatives has historically been unrealistic.
Today, for a defined operational scope, a technically fluent operator can stand up a viable internal system in days - not months or years.
Not a global enterprise platform.
Not a replacement for every vendor.
But something economically rational for a specific business.
That distinction matters.
If custom systems become economically viable for small and mid-sized operators, the foundational advantage of generic SaaS weakens. Many vertical SaaS products depend not only on features, but on the fact that building alternatives was prohibitively expensive.
When implementation complexity stops being a moat, pricing power erodes.
This does not mean SaaS disappears.
It means SaaS fragments.
It means generic, off-the-shelf solutions face pressure from AI-native, workflow-specific systems built closer to the business operator.
The competitive advantage shifts from code production to:
Architectural judgment
Domain expertise
Data ownership
Distribution
Execution discipline
In the past, the companies with the largest engineering teams held the advantage.
In the next decade, the advantage may belong to operators who understand their workflows deeply and can leverage AI to build systems aligned precisely to their needs.
Most independent operators are not asking whether they should replace SaaS.
They are asking whether they even can.
The answer is increasingly yes - but only with technical discipline and operational clarity.
The economics of software have shifted.
That shift will not eliminate SaaS.
But it will change who captures value.
I work with independent operators and acquisition-minded owners navigating this shift - assessing where SaaS dependence creates risk, where AI-native systems are economically viable, and how to architect technology around long-term control rather than convenience.
If you’re seriously evaluating those questions, you can reach out to me directly.