The Pattern Repeats

In 2016, a Volkswagen VP of software engineering told me something that stayed with me: "VW doesn't write a single line of code. Our suppliers do it all."

He said it with pride. At the time, it seemed rational. Software was mature, complex, and expensive. Why not let specialists handle it? Accenture, IBM, and other system integrators had been doing this for decades. It made sense to focus on what you knew - manufacturing cars, developing the combustion engine - and let the experts handle the rest.

The economics looked compelling too. Offshore development centers in India offered rates at a fraction of local salaries. The per-person cost savings were obvious. What companies missed was the total cost of outsourcing: the coordination overhead, the communication delays, the loss of institutional knowledge, the inability to iterate quickly. They optimized for unit cost and ignored system cost.

Meanwhile, Tesla was building software as a core competency from day one.

We know how that story ended.

Today, I see the same pattern emerging with AI. Companies across industries are making the same strategic choice VW and other automotive OEMs made 20+ years ago: treating a fundamental capability as something suppliers will handle for them. They're outsourcing the thinking before they've learned to think.

But there's one critical difference this time: AI is only three years old. The system integrators positioning themselves as experts have maybe six months more experience than you do. Right now, Accenture has no structural knowledge advantage over you. McKinsey hasn't figured this out yet either.

This is your window of opportunity. Not to hire consultants to "do AI" for you but to build AI literacy in your organization. To develop the capability to leverage agents. To understand how AI reshapes your business model.

Miss this window, and you're back to the Volkswagen position: dependent, behind, and scrambling to rebuild capabilities you should have owned from the start.