Every technology leader is being asked the same question right now, from every direction: should we build this ourselves or buy it? With AI the stakes are higher than usual, because the wrong call can mean either reinventing a commodity or locking your core differentiation behind someone else’s roadmap.
Buy the commodity, build the edge
The clearest rule we know: buy what is becoming a commodity, build what is genuinely yours. Foundation models, infrastructure and generic tooling are someone else’s economy of scale — renting them is usually the right move. The workflow, data and judgment that make your business specific are the things worth owning.
The questions that actually decide it
Before committing either way, we push leaders to answer a short, uncomfortable list:
- Is this a differentiator or a utility? Utilities get bought.
- What is the cost of lock-in? Including the exit you hope you never need.
- Do we have the talent to maintain it? Building is the cheap part; owning it forever is not.
- How fast is this layer moving? Fast-moving layers are dangerous to freeze into your own code.
Buy what is becoming a commodity. Build what makes you, you.
Usually the answer is both
In practice the right architecture is a thin, owned layer of your real differentiation sitting on top of bought infrastructure — with the seams designed so you are never trapped. Getting those seams right is most of the decision, and it is exactly the kind of call a clear-eyed advisor earns their keep on.