The build-vs-buy math just changed, and your CFO can feel it

On April 23, software stocks took a beating after ServiceNow and IBM reported earnings. The story wasn't bad numbers. It was investor fear that AI agents are about to eat the margins that built modern enterprise software. CNBC covered the selloff here.

If you run a 30-person company, you probably don't care what ServiceNow stock does on a Wednesday. But the reason it dropped matters to you, because it's the same reason your next SaaS renewal conversation is going to feel different.

What actually changed

For twenty years, the build-vs-buy decision had a clear answer for most small and mid-sized businesses: buy. Custom software was expensive, slow, and risky. SaaS was cheap on day one, fast to roll out, and someone else handled the servers.

The SaaS pitch worked because building was hard. AI didn't kill SaaS, but it did move the cost of building down a lot. A two-person team with the right tools can now ship things that used to require six engineers and a quarter.

Bain's recent analysis frames it well: agentic AI is changing the math on both sides of the build-vs-buy line, and the per-seat pricing model in particular looks shaky when an agent can do the work of ten seats (Bain, 2025).

Fortune's piece on the so-called SaaSpocalypse makes a related point: the moats incumbents relied on, switching costs, integrations, data lock-in, are weaker when AI tools make it easier to extract, migrate, and rebuild (Fortune).

What this means for you

A few practical observations from what we're seeing with clients:

Your renewal is a real negotiation now. When you tell a vendor you're evaluating a custom replacement, that's not bluster anymore. They know it. Use it.

Stack consolidation is the easy win. Most companies we talk to are paying for 4 to 8 SaaS tools that each solve 30% of their actual workflow. The gaps get filled with spreadsheets and Slack messages. A focused custom build often replaces three of those tools, not all of them, and that's fine.

The boring stuff is where the math works. Custom builds make sense for the workflows that are specific to how you do business. They rarely make sense for email, accounting, or payroll. Don't rebuild what's already a commodity.

The honest caveat

The SaaSpocalypse narrative is overstated. ServiceNow isn't going away. Neither is Salesforce. Incumbents are bolting AI onto their products, and for plenty of use cases that will be enough.

But the default answer of "just buy the SaaS" deserves more scrutiny than it got a year ago. The cost curve moved. The renewal letter that landed on your desk hasn't caught up yet.

If you want a second opinion on what's worth keeping and what's worth replacing, that's the kind of thing a discovery call is for. No pitch. Just a look at your stack.