The build-vs-buy spectrum is dead. Here are the five postures that replaced it.

"Build vs. Buy" is dead.

The SaaSpocalypse wiped $2T off enterprise software valuations in Q1. Every board is asking the same question: "With AI coding this good, why are we still buying software?"

The honest answer is that the binary is obsolete.

The real decision isn't two options. It's five:

• Buy — vendor product as-shipped. Your process bends to the software.

• Configure — live inside a platform (Salesforce, ServiceNow, Workday). Their framework, your semantics.

• Extend — keep a trusted core, build differentiating capability on top via APIs and agents.

• Compose — orchestrate multiple systems where no single one owns the truth. Value lives in the orchestration layer itself.

• Build — own the full stack. The capability IS the business.

Here's what investors are getting wrong about AI coding:

It didn't push everything toward Build. It pushed the optimal posture one step to the right for a specific band of use cases.

AI coding collapsed the cost of building edges and glue. It did not collapse the cost of building cores.

ERP, HCM, EHR, billing, GL - those are still dominated by domain depth, compliance burden, and multi-decade data models. A vibe-coded system of record is a lawsuit waiting to happen.

The real sweet spot AI unlocked is Extend — bolt-on differentiation on top of trusted cores, without ripping out the system of record.

Three questions decide the posture:

1. Strategic advantage or table stakes?

2. Who absorbs liability when it breaks?

3. Is the data uniquely yours, or is someone else's dataset the moat?

The enterprises that get this will quietly compound advantage while competitors oscillate between "rip it all out" and "nothing changes."