The SaaSpocalypse That Wasn't

CNBC journalists rebuilt Monday.com's core functionality in a single weekend using AI coding tools. Monday.com's stock dropped. Boardrooms panicked. A 'SaaSpocalypse' was declared — the end of SaaS as we know it.

But that narrative misses the point. Software is cheap to build. It's expensive to run.

The Real Cost Isn't Development

The source article makes a crucial distinction: AI has changed how fast software can be produced. It has not changed the burden of keeping it operational. Security updates arrive continuously. Integrations break without warning. Regulations shift. Users change their minds. None of this is edge-case behavior. It is the system.

Most of a software's lifecycle cost sits after deployment. Maintenance, not writing, consumes the majority of effort over time. When you replace SaaS with an internal build, you don't eliminate that work — you absorb it internally.

The Hidden Dependency Trap

SaaS vendors distribute maintenance across thousands of customers. Internal builds concentrate it inside your organization. The result: you trade vendor lock-in for individual dependency. The people who built your AI-generated tool become irreplaceable. When they leave, you're left with undocumented, fragile code that's harder to interpret and modify.

This is not a hypothetical. The article explicitly warns: "When the people who built them leave, what remains is software that is harder to interpret, harder to modify, and more fragile than expected."

A Concrete Example: The Monday.com Rebuild

Consider the CNBC experiment. They used AI coding tools to recreate a project management platform. Let's say they generated a Node.js backend with Express and a React frontend. A sample route might look like:

// AI-generated task creation endpoint
app.post('/api/tasks', async (req, res) => {
  const { title, assignee, dueDate } = req.body;
  const task = await Task.create({ title, assignee, dueDate });
  res.status(201).json(task);
});

This works for a demo. But in production, you need authentication, rate limiting, error handling, database migrations, CI/CD pipelines, monitoring, and alerting. AI won't generate those for you — or maintain them when a library deprecates a function.

The Operations Gap

SaaS vendors handle all of that. They have teams for security patches, compliance updates, and feature evolution. When you build internally, you assume that entire stack. The article notes: "The burden is absorbed into the product. When companies replace SaaS with internal systems, those responsibilities do not disappear."

Moreover, you're now paying for enterprise AI developer tools themselves. The savings from canceling a SaaS subscription get offset by new costs: cloud infrastructure, AI tool licenses, and developer time for ongoing maintenance.

What Executives Miss

The build-versus-buy decision is not a cost exercise. It's a question of where operational responsibility should sit. Before canceling any subscription, ask:

  • Who will own the system long term?
  • How will knowledge be distributed to avoid critical dependencies on individuals?
  • Is the organization prepared to absorb the operational burden currently carried by the SaaS vendor?

The Bottom Line

AI has changed the economics enough to make internal builds viable in cases that previously made little sense. But it has only changed one side of the equation. It has made software cheaper to build. It has not made it cheaper to run.

That gap is where most enterprise decisions will fail. Before you cancel your next SaaS subscription, ask not whether you can rebuild it, but whether you're prepared to own everything that follows.