Last year, I open-sourced a utility library that solved a real problem I'd been wrestling with in production. The documentation was solid, the code was clean, and I released it under MIT because that's what developers do, right? We share. We build in public. We contribute to the commons.
Within six months, a startup had taken that library, wrapped it in a polished UI, added some cloud infrastructure around it, and launched a paid product. They weren't stealing, legally speaking, they were following the MIT license to the letter. They kept my attribution. They didn't hide what they were doing. But they turned my nights-and-weekends work into their revenue stream, and I got nothing but a footnote in their about page.
That's when I started thinking seriously about the infrastructure layer, and why control over the platform matters more than I'd realized.
The Real Problem Isn't Openness, It's Who Controls the Roads
The original article makes a point that I've been chewing on for weeks: open source code doesn't guarantee an open economy. GitHub is genuinely open. You can publish anything. But Microsoft owns the platform. Hugging Face remains accessible to everyone, but Nvidia now controls the business around it.
This isn't about malice. It's about structure. When you have a billion developers releasing code to a platform you own, you have extraordinary power. You see trends before they exist. You can promote what you want promoted. You can make certain projects discoverable and others invisible just through ranking algorithms. You can introduce pricing tiers. You can make competing services incompatible.
I've experienced this at smaller scale. On GitHub, projects with corporate backing get better visibility. On Stack Overflow, answers from employees of big companies float to the top. The algorithm doesn't have to be rigged, it just has to favor the well-resourced.
The Wealth Concentration Problem Is Real in AI Infrastructure
What makes this especially sharp for AI and superintelligence development is the hardware cost. You can't run a large language model on a laptop. You need cloud credits, and most of those credits flow through a handful of companies: AWS, Google Cloud, Azure. The code may be open, but the compute infrastructure forms a bottleneck.
A researcher from Islamabad, or Lagos, or São Paulo can publish brilliant work on model optimization. But to actually build and test at scale, they need to rent GPUs from companies whose business model depends on them not becoming too independent. The developer has freedom of expression. The platform owner has freedom of profit.
And the consolidation keeps accelerating. Nvidia acquiring Hugging Face isn't a hostile takeover, it's the natural end state of a system where one layer controls the infrastructure everyone else depends on.
What I'm Actually Worried About
Here's what troubles me most: I don't think this is fixable through licensing alone. The article mentions that even permissive licenses like Apache 2.0 allow commercial use without contribution back. That's by design. But it also means the person who builds the tool rarely benefits from the scale the tool achieves.
My open-source library taught me that the value isn't in the code, it's in:
- The validation that the problem matters
- The market evidence that people will use the solution
- The roadmap of what features users actually need
- The architectural decisions that worked in production
All of that gets transferred to whoever has the platform to commercialize it.
I'm not suggesting we stop open-sourcing. But I think we need to be much more deliberate about what we release, where we release it, and under what terms. Some of my recent work I've kept proprietary, not out of selfishness, but out of self-preservation.
The Question for Developers Like Us
If you're building something valuable, whether it's a library, a dataset, or a model, what's your strategy? Are you comfortable with the current model, where openness means visibility but not necessarily ownership of outcomes?
I don't have a perfect answer yet. I just know that my next library won't go on GitHub with zero guardrails.
Source: This post was inspired by "The SI Wealth Transfer Is Becoming the Largest Rug Pull in History." by Dev.to. Read the original article
