I started thinking about the old build versus buy debate. At first, it seems obvious. AI changes everything. Software can be written faster than ever. Surely the advantage shifts toward building. Then I thought about the real cost of building. Not the programmers. Not the budget. The risk.

We often say changing a core operating system is like changing the engine while the airplane is still flying. I’m not sure that’s the best analogy anymore. An airplane engine is a component. It can be done.

Just ask the customers on that flight if they’re comfortable with the risk. They may demand a discount before volunteering for the experimental flight.

A core operating system is more like a heart. A heart transplant can be done. Medicine proves it every day. But notice what it takes.

Even then, the operation lasts hours, not months. Recovery takes months. No one expects the patient to run a marathon the following week. Survival comes first.

Now imagine stretching that surgery over two years while asking the patient to continue running marathons every day.

That begins to resemble replacing the operational heart of a financial services company while continuing to onboard new clients, satisfy regulators, process millions of transactions, and reassure customers that nothing has changed.

Then another thought occurred to me. Even during a heart transplant, we don’t build a new heart. We transplant one that has already proven it works.

Maybe that is the hidden advantage of buying mature software. You’re not simply buying code. You’re buying years of production experience. Thousands of defects someone else already discovered. Hundreds of implementation lessons. Countless edge cases that nobody imagined until reality exposed them. Regulatory examinations. Unexpected failures. Operational scars.

We call it “bleeding edge” for a reason.

The source code is only part of the asset. The proof may be the larger asset.

That made me wonder whether AI changes this equation. AI can certainly write software faster than human developers. Perhaps much faster. Perhaps much better. But writing code and proving a system are two different things.

AI may compress the time required to construct the heart. It does not compress the years required to prove the heart can sustain life. Reality still tests one production day at a time. Even evolution builds patiently.

Then I realized this doesn’t mean mature organizations should never build. Quite the opposite. Think about a newborn.

Nature builds a new heart because nothing yet depends upon it. Innovation belongs there. If the design is better, the entire organism may grow stronger than anything that existed before. That is where startups thrive.

Now think about the sedentary fifty-year-old who decides to become a runner. Remarkable change is still possible. The heart becomes stronger. The lungs adapt. Muscles develop. Weight disappears. Performance that once seemed impossible becomes routine. But notice what changed. The organism evolved. It wasn’t rebuilt from scratch.

Perhaps that is the better lesson for mature organizations. Build where building creates advantage. Modernize. Automate. Use AI aggressively. Strengthen what can be strengthened. But first understand which parts of the organism are evolutionary and which are foundational.

One last image came to mind.

Oscar Pistorius’ carbon-fiber blades demonstrated that some components can outperform nature itself. Technology can produce extraordinary advantages. But even those remarkable prosthetics depended upon a functioning cardiovascular system, nervous system, lungs, balance, and decades of human adaptation. The innovation attached to a living organism. It didn’t replace one.

Perhaps that is where AI’s greatest opportunity lies. Not replacing every foundational system simply because it now can generate code. But attaching extraordinary new capabilities to foundations that have already spent decades proving they can carry the weight.

And perhaps one final thought.

A heart surgeon cannot operate on himself.

There is probably a lesson there too.