An uncomfortable observation

I've always had a soft spot for open source. It represents something important to me: transparency, independence, and the freedom to build without asking permission. Much of the digital world rests on open foundations, and I suspect we often underestimate how much we owe to them.

Yet when I look back at the tools that have shaped my own career, I notice something that doesn't quite fit that instinct. Apple. Microsoft. Adobe. Google. Almost without exception, the products I have relied on most have been commercial. That isn't a conclusion so much as an observation, but it has been nagging at me for some time.

At first I wondered whether this revealed an inconsistency in my own thinking. How could someone who values open ecosystems so highly end up gravitating towards proprietary products? The more I thought about it, however, the less convinced I became that these are opposite choices.

Different layers

Perhaps I've been comparing different layers of the digital world. Open source and proprietary software are often discussed as competing philosophies, especially now that the same debate is emerging around AI models. But maybe that framing is too simplistic.

Increasingly, I find myself wondering whether open ecosystems and commercial products perform different roles within the same digital landscape.

Many of the technologies that quietly underpin modern computing are open. Linux, BSD, Chromium, Python, PostgreSQL and SQLite are everywhere, even if most people never notice them. The products we interact with every day are often built on top of those foundations, adding integration, support, design, documentation, security and accountability. Those qualities do not emerge simply because software is open or closed. They require sustained effort of a different kind.

Looking back, I no longer think I was choosing proprietary software over open source. More often than not, I was choosing products over infrastructure.

An ecosystem, not a battlefield

That distinction changed the way I look at the debate.

Open source and proprietary software are often presented as rivals, as though one should eventually replace the other. But ecosystems rarely work like that. Different organisms occupy different niches, and their relationships are often more complementary than competitive.

The digital world seems remarkably similar.

Commercial companies build on open foundations. Open source projects benefit from commercial investment, engineering capacity and widespread adoption. Developers move between both worlds. Ideas flow in both directions. Neither side exists in isolation, and neither would look the same without the other.

Perhaps that is why the discussion feels unsatisfying whenever it becomes ideological. It encourages us to choose sides when the more interesting question is how these different parts reinforce one another.

AI through the same lens

I am not suggesting that AI will follow exactly the same path. Every technological shift brings its own dynamics, and history never repeats itself in quite the same way.

Still, I cannot help noticing a family resemblance.

Open-weight models seem exceptionally well suited to spreading knowledge, encouraging experimentation and reducing dependence on a handful of providers. Commercial companies, meanwhile, often excel at turning those capabilities into products that organisations can adopt with confidence. They provide integration, reliability, support and a coherent user experience. Those are different strengths rather than obviously competing ones.

That doesn't mean one model will inevitably dominate the other. If anything, it suggests they may continue to evolve together.

A different question

That leaves me with a different question than the one I started with.

Instead of asking whether AI should be open or proprietary, perhaps we should ask what kind of ecosystem allows both to flourish. Healthy systems rarely depend on a single species, a single organisation or a single way of working. They derive resilience from diversity, not uniformity.

If that's true, then my own career wasn't a contradiction after all.

The products I came to rely on were often standing on open foundations all along.

Related reading, if the proprietary products piece resonated with you:

Is there really one AI economy?
Published the same day, and it uses the exact same move: taking a concept everyone treats as one thing (an "AI company" there, "open versus proprietary" here) and splitting it into layers that serve different purposes. If the layers argument is what stuck with you in this piece, that one applies the same lens to the wider AI market.

The €60,000 LLM Well
This is the practical flip side of the argument. Where the proprietary products piece asks why open ideals keep losing out to commercial tools in practice, this one puts a number on what "AI independence" through self-hosting would actually cost, and lets the reader decide if it's worth digging the well.

Deep Tech AI and Open Source licenses
A more governance-focused companion. It looks at the licensing mechanics behind open source AI, specifically how copyleft terms can force companies to give improvements back to the community, which adds a legal and structural layer under the more philosophical "ecosystem, not a battlefield" argument in the proprietary products piece.