The build versus buy decision is being reset by AI. What used to be a predictable tradeoff around cost, time, and complexity is shifting fast as the cost of building software collapses.

Non-technical operators are now generating real products with AI tools. In days, sometimes even hours.

I met with a commercial banker recently who doesn’t know a lick of coding, wrote 170k lines of code in 8 weeks, building a full ERP suite for a niche vertical. Nights and weekends for business professionals are now obsessive product sprints. According to Claude, this would have taken a team of devs 18 months to build.

This changes the question. It is no longer “Can we build this?” It is “Should we?”

AI’s impact is often overestimated in the short term and underestimated in the long term. AI is expanding who gets to build. Enterprises are building more because they already have engineers and internal complexity worth solving. Custom tools, internal copilots, and workflow automation now justify themselves faster.

At the other end, small teams are building because they finally can. AI collapses the skill barrier, making bespoke internal tools viable without deep technical talent.

Middle-market companies may face the most disruption. They are too complex for scrappy internal builds but often lack the resources to replicate enterprise-grade development. That leaves them dependent on external software while facing pricing pressure and fragmentation.

Not all software is equally exposed.

Horizontal tools face the most pressure. Generic SaaS layers and commoditized workflows are increasingly vulnerable to internal builds or AI-generated alternatives.

Vertical software remains more defensible. Products that compound value through proprietary data, network effects, and embedded workflows are far harder to replace.

This creates a split: commodity features compress, deeply embedded platforms persist.

One Head of AI at a Fortune 500 company framed it simply: “We buy what accelerates us, and build what differentiates us.”

What AI changes most is supply. The real risk is not just internal builds replacing vendors, but a flood of new software creation.

At a recent construction panel I moderated, two multi-billion-dollar GCs noted that they are primarily building tools internally with their IT teams. 

As Jamin Ball recently put it, “the cost of creating software is going to zero.” If that holds, the real disruption is not replacement but saturation.

When Buying Still Wins

Highly regulated domains like payments, tax, and security favor buying. The operational burden extends far beyond code, and the cost of failure is high.

Systems of record and foundational platforms benefit from scale and specialization that are difficult to replicate internally.

AI does not remove implementation friction either. Whether built or bought, adoption, training, and internal change management still exist. The friction doesn’t disappear, it just moves.

Pricing Pressure Ahead

Even when companies continue buying software, AI shifts the leverage.

If customers can credibly threaten to build alternatives, pricing power moves. SaaS margins, especially in horizontal categories, are likely to compress.

As building gets easier, distribution becomes the moat. Founder-market fit, networks, and creative go-to-market strategies become your true advantages. In a world where anyone can build, not everyone can reach users.

For SMBs and middle-market companies: build lightweight, buy heavy infrastructure. Internal tools and AI layers are built quickly, while foundational systems remain external.

For enterprises: buy the foundation, build the intelligence layer. Systems of record remain external, while internal teams build AI overlays that automate workflows and orchestrate data.

The closer software sits to proprietary workflows, the more likely it is to be built. The closer it sits to infrastructure or regulation, the more likely it is to be bought.

Categories built on proprietary data and long research loops remain difficult to replace. Products powered by unique datasets or complex training pipelines retain strong defensibility, often shifting toward partnerships rather than internal builds.

The Takeaway

Build versus buy is becoming dynamic rather than static. Companies will mix strategies, revisit decisions more often, and layer internal AI on top of external foundations.

The direction is clear: as the cost of building falls, leverage shifts toward speed, distribution, and proprietary context. The winners will not pick build or buy. They will develop the judgment to do both.

Software won’t disappear, but it will compress. Margins tighten, and differentiation concentrates in fewer hands.