Build vs. Buy in the AI Age – Why Most Companies Are Asking the Wrong Question

Whenever I talk to companies about their technology strategy, the same question keeps coming up: Should we build it ourselves or buy it?

It is a reasonable question. But most of the time, it is asked at the wrong moment, namely when the budget has already been approved, the team has already been assigned, and expectations have already been set. And it is often answered incorrectly because the framework needed to make the question meaningful is missing.

‍

The Real Question Is: What Truly Sets You Apart?

Build vs. Buy is not a technical problem. It is a strategic decision – and it does not begin with the question of what a tool costs or how long implementation takes. It begins with an honest answer to this question: What does your company do better than anyone else?

Where the answer becomes specific, you build. Everything else, you buy.

That sounds simple. In practice, it is not. Most companies underestimate how much they want to build themselves and overestimate how much they are actually capable of building.

‍

What Happens When You Build Too Much Yourself

Over the past few years, I have seen many companies decide to develop their data and content infrastructure internally. Sometimes, that was the right decision. Often, it was not.

The typical pattern is this: An internal project starts with a clearly defined scope. After twelve months, the scope has grown, the team has grown, the costs have grown, and the system still does not deliver what is actually needed. Meanwhile, the market has moved on. New requirements emerge. And suddenly, the company is no longer building what it originally set out to create, but whatever problem is most urgent at that moment.

This is not a failure of the team. It is the structural problem with infrastructure projects that fall outside a company’s core expertise.

‍

Infrastructure Is Not a Competitive Advantage – It Is a Prerequisite

A DAM system that structures and connects master data and makes it accessible to AI models is infrastructure. It is necessary. But it does not differentiate you.

What differentiates you is what you do with this infrastructure. How you prepare your product data. How you use AI to support customer decisions. How you design your content value chain.

That is where custom development makes sense. Not below that level.

Imagine building a power plant before you start manufacturing products. No one would do that. But that is exactly what happens when companies start building their own database structure for Product Information Management, developing their own asset management system, or creating their own AI layer for multimodal image scoring.

It is possible. But it rarely makes sense.

‍

What “Buying” Means in Practice

Buying does not mean giving up control. It means delegating complexity that does not belong to your core expertise.

What this means in practice: Tesa went live in six weeks with concrete AI use cases. Not because the project was so simple, but because the foundation did not have to be rebuilt. Weidmüller, with around one billion euros in revenue, needed five months for a complete implementation. Siemens completed a global rollout in nine months, covering more than one million products and product relationships worldwide.

These figures are not a marketing promise. They are the result of building on a system that has been continuously developed for exactly this purpose for years.

When a company wants to build the same thing internally, it does not take five months. It takes years, and in the end, the company still has something that cannot do what a specialized system can do after ten years of development.

‍

The AI Dimension Changes Everything and Nothing

AI has not resolved the Build vs. Buy question. It has made it more urgent.

Any company that wants to use AI productively today needs a data foundation capable of supporting AI models: structured master data, clear relationships, and multimodal accessibility. Building this foundation takes time. And every week a company spends developing this infrastructure itself is a week in which others are already scaling.

The companies I see using AI productively, not just as a pilot project but at scale, took this step earlier. They made a clear decision: We buy the foundation. We build the competitive advantage on top of it.

‍

What I Recommend

Ask yourselves two questions before making a decision:

First: Have we ever done this in-house as well as someone who has been doing nothing else for years? If the honest answer is no, buy.

Second: Is what we want to build the reason customers choose us, or is it a prerequisite for being able to deliver at all? If it is a prerequisite, buy.

This does not eliminate the investment. But it prevents the wrong investment.

‍

Listen to the Conversation

I also discussed these questions with Jonas Rashedi on the podcast MY DATA IS BETTER THAN YOURS, along with the OTTO case, where DAM is headed as a category, and what master data poverty really costs in practice.

‍

👉  Listen to the podcast in German now

Apple Podcasts

Spotify

YouTube

‍

‍