The old buy vs. build software question is back.
AI is changing the economics of software development, and companies need to reflect that.
When I started my career in 2007 in enterprise software development, the common mantra was clear:
Buy, don’t build.
At the strategic level, it made sense. Why spend years developing custom software if you could buy an off-the-shelf system with maintenance and support?
But reality was always more complicated.
Off-the-shelf software was not cheap. It was not very flexible either. Customization was expensive and slow to deliver. And in many cases, the result was fragile and only partially aligned with how the business actually worked. So despite the strategy, custom development kept gaining attention and popularity as an accessible and affordable remedy for operational problems.
The strategy was not necessarily wrong. It was incomplete.
Since then, many waves have tried to solve the same problem: outsourcing, then bringing development back in-house, agile, automation, cloud, SaaS, low-code, no-code and DevOps. Different labels, same underlying ambition: faster and cheaper IT, better aligned with business goals. All of them made software development faster, cheaper, more scalable or easier to move from CapEx to OpEx.
And yet, companies still buy software like ERPs, MRPs, WMSs, CRMs, HR systems and finance systems. At the same time, they still build what differentiates them in the market.
Now AI is changing the equation again. It makes building software faster, cheaper and more accessible. The cost of software development is dropping dramatically. “Everyone can develop” is becoming the new mantra. Some organizations are starting to treat token burn as if it automatically translated into business outcomes.
It does not.
AI-assisted coding works best where software engineering discipline already exists: version control, code reviews, testing, documentation, deployment pipelines, ownership and maintainability. Without that discipline, AI does not create a strategy.
It creates faster chaos.
The internet is already full of stories from people who built their first app in two days and then discovered that the codebase became unmanageable almost immediately.
That is not enterprise software.
This is why SaaS and packaged systems are not going away. In many domains, the hard part is not only developing the software. The hard part is maintaining the domain knowledge behind it. Tax, payroll, HR, finance, customs, regulatory reporting and industry-specific compliance. These areas change constantly. Laws change. Rules change. Reporting requirements change. Best practices change. If you build software in these domains, you are not only committing to maintain the code.
You are committing to maintain the expertise.
For many companies, this is exactly where buying a proven, maintained and compliant solution still makes more sense than building their own. But companies also need to learn how to work with a new capability:
Cheap software development.
This will not only allow companies to build more on their own. It will also put pressure on SaaS and off-the-shelf software providers. If custom development becomes cheaper and faster, packaged software will need to become more flexible, more focused, and probably cheaper as well. So what is the strategy now?
Build, don’t buy?
I do not think so.
You do not want thousands of vibe-coded apps becoming the backbone of your company, just as you never wanted to run your business on thousands of shared Excel files and macros. Self-coded AI apps and workflows have the potential to recreate that exact hell—where critical processes suddenly depend on a fragile file running on someone’s local computer or a forgotten shared drive. Not to mention that a self-learning agent constantly rewriting its own memory files can experience drift, failing in ways that are incredibly hard to detect and recover from.
You still need architecture, data strategy, integration strategy, security, ownership, governance and maintainability. The real question is not whether to buy or build. The real question is where the new sweet spot is, and how to maintain flexibility. The companies that win will not be the ones that build everything. And not the ones that buy everything.
They will be the ones that know exactly what should be standardized, what should differentiate them, what they are capable of maintaining, and how to keep enough flexibility to revise their decisions quickly.
The buy vs. build question is not going away. AI just made IT strategy more important than ever, as IT has become a massive lever that works in both directions. The right strategy, supporting your business's growth and flexibility, can skyrocket your company. But the wrong one can bury it forever much faster than a decade ago.
If you are trying to define the right strategy for your company, let’s talk.