The build-vs-buy math changed while you weren't looking

For a decade, the answer to 'should we build this or buy it?' was almost always buy. SaaS was cheap, fast, and someone else's problem to maintain. That default is cracking, and if you run ops or IT at a small or mid-sized company, it's worth understanding why before your next renewal cycle.

The incumbents aren't dying. They're just no longer safe.

Fortune ran a good piece recently arguing that AI agents from OpenAI and Anthropic aren't killing SaaS outright, but the incumbents like Salesforce, ServiceNow, and Workday can't sleep easy either. The workflows those platforms owned for years are now workflows an AI-assisted team can rebuild in weeks instead of quarters. That doesn't mean everyone will. It means the calculation is no longer automatic.

The short version: buying used to be obviously cheaper than building. It's not obvious anymore.

Two things are pushing at once

First, AI tooling is making internal builds faster and less risky. LeadDev has been covering how engineering teams are being handed a mandate from the C-suite to replace subscription tools with internal software. Not because engineers want more work, but because leadership is looking at renewal quotes and doing the math differently.

Second, AI spend itself is out of control. CIO Dive reported that enterprises are struggling with mounting AI costs and tool sprawl, and CIOs are actively rethinking their software portfolios as a result. When your AI line items and your SaaS line items are both growing, something gets cut. Usually it's the SaaS tool that's 'good enough' but doesn't quite fit.

What this looks like for a smaller company

We're a small team that builds for small teams, so I'll skip the enterprise strategy talk. Here's what we actually see:

- A company paying for five SaaS tools where two of them are used by one person each, and none of them talk to the others properly. - A renewal quote that jumped 30% because the vendor added AI features nobody asked for. - An ops manager spending an afternoon a week copying data between systems because the integrations broke or never existed.

Five years ago, building custom software to replace that stack was a big project. Today, a lot of it is a small one. Not zero, but small. That's the shift.

The takeaway

Don't rip out your SaaS stack because a Fortune article said things are changing. But when your next renewal comes up, actually do the math. Include integration cost, seat inflation, and the hours your team spends working around the tool. Compare it to what a focused custom build would cost to run for three years.

Sometimes buying still wins. It should. But 'we've always bought' isn't a reason anymore. It's just a habit.