Will AI Kill SaaS? A Professor Asked. Here's What I Actually Think.
A conversation about cloud computing in 2010 led to a question about AI and SaaS in 2026. The answer is more nuanced than the headlines suggest.
In 2010, I stood in front of a university class and made a claim that most people in the room weren't ready to hear: cloud computing was going to fundamentally reshape how businesses operate. It wasn't a popular opinion at the time. Most enterprise leaders were still debating whether they'd ever trust their data outside their own data centers.
A few weeks ago, the professor who invited me to speak that day reached out on LinkedIn. Professor Segun Oyedele reminded me of that conversation, said the cloud call had been spot on, and then asked me something far more interesting: Is AI going to kill the SaaS subscription model?
He wanted to know whether companies would start vibe coding their own software internally instead of paying enterprise subscriptions. Whether the rise of AI-powered development tools would make companies like Salesforce obsolete. And whether investors were right to be dumping SaaS stocks in response to that fear.
These are the right questions. The honest answer is more nuanced than most people want it to be.
Yes, Parts of SaaS Are at Risk. No, It's Not Universal.
The short answer is yes. If a company has the internal capability to build and operate software well, the math on subscription software has changed. Materially.
This isn't theoretical for me. I started having this conversation with our CEO about four months ago, and we're already acting on it. My team expects to build two applications in the next couple of months that will directly replace paid subscription tools we currently use. With modern AI tooling, the cost of building purpose-fit software has dropped enough that it now makes financial sense for certain use cases. Build it yourself, own it outright, shape it exactly to your workflow. That option used to be expensive and slow. It's becoming neither.
But here's where the blanket narrative breaks down.
Where SaaS Isn't Going Anywhere
In highly regulated or operationally complex industries, the subscription model isn't just about software access. Medical. Pharmaceutical. Manufacturing. Financial services. In those environments, SaaS is compliance. It's liability transfer. It's uptime guarantees, audit trails, and institutional knowledge bundled into a contract.
The cost of getting security posture wrong, failing an audit, or losing operational continuity far outweighs whatever you'd save by building in-house. These companies aren't paying for features. They're paying for risk management. They're paying so that when something goes wrong, the accountability sits with a vendor who has contractual obligations, not with an internal team that was experimenting with AI-generated code.
No amount of vibe coding changes that calculus. Not yet, and probably not for a long time.
The real question isn't whether AI can replace SaaS.
It's which SaaS companies are evolving their value beyond software access into trust, integration, and operational leverage.
The Bigger Disruption Nobody's Talking About
I told the professor that the SaaS question, while valid, is actually the smaller story. The bigger disruption is knowledge work compression.
If companies like Anthropic, OpenAI, or tooling platforms like Cursor move vertically into specific domains, a meaningful amount of traditional knowledge work gets displaced or radically reshaped. Analysis that took a team a week gets done in hours. Code that required a senior engineer gets drafted by a junior with the right tools. Documentation, research, planning, reporting. All compressed. The output stays the same or improves. The labor input shrinks dramatically.
That compression doesn't eliminate people. But it fundamentally changes what people need to be good at.
The Rise of the Agentic Engineer
I believe we're seeing a new role emerge, something I'd call the agentic engineer. This isn't a traditional software developer. It's someone who designs systems, workflows, and guardrails while AI handles much of the execution. Think architecture plus orchestration. The agentic engineer understands what needs to be built, defines the constraints, and directs AI agents to do the building. Their value is in judgment, context, and quality assurance, not in writing every line themselves.
The optimistic view, and I believe it's the correct one, is that demand for software itself will increase. We'll see far more custom, purpose-built systems than ever before. The bottleneck shifts from "can we build it?" to "should we build it, and how do we design it well?" That's a meaningful difference. It means more software gets created, not less. It just gets created differently.
That shift creates enormous opportunity for people willing to adapt.
The Adoption Gap Is Already Visible
Here's something I observe every day within my own team. Roughly a third have fully embraced AI and are compounding their productivity week over week. They're faster, more creative, and taking on work they couldn't have handled six months ago. They treat AI as a multiplier, and the results show.
Others are hesitant. Some are struggling to adapt. A few haven't meaningfully started.
That gap will matter. Not eventually. Soon. The people and organizations that learn to work alongside AI effectively are going to separate from those that don't. This isn't about replacing anyone today. It's about who compounds and who stalls. Over twelve months, the difference between those two groups becomes very hard to close.
What This Means for Investors
Professor Oyedele also asked about the investment angle, specifically whether the market is right to sell software stocks on AI disruption fears. My view: be cautious about blanket narratives.
Some SaaS businesses will absolutely be disrupted. The ones selling commoditized features at subscription prices are vulnerable. If your product is a thin layer of functionality that a motivated team can replicate in a few weeks with AI tooling, your moat just got a lot shallower.
But others will entrench themselves further. Companies that have built deep integration into customer workflows, that carry regulatory certifications, that offer genuine operational leverage beyond the software itself. Those businesses aren't easily replaced by a weekend of prompt engineering.
The market is pattern-matching "AI disrupts everything" without doing the harder work of distinguishing who gets disrupted and who gets stronger. That creates real opportunity for investors willing to think more carefully about where durable value actually lives.
Still Early Innings
I appreciated the professor reaching back out after all these years. In 2010, the cloud question felt like a binary debate: is it real or isn't it? The answer was obvious to some of us, but it took years for the market to fully absorb it.
AI feels similar, except the pace is faster and the surface area is vastly larger. The SaaS model won't disappear. But it will evolve, and some of it will be replaced by companies that decide they can do it better themselves. The organizations that thrive will be the ones that understand which category each piece of their software stack falls into, and act accordingly.
For more on how AI is reshaping software and teams, browse the blog archive or explore the implementation resources.
If a third of your team has already embraced AI and the rest haven't started, what does your organization look like in two years?