The miracle of modern Artificial Intelligence emerged from the chaotic and generous culture of free software. However, this open ecosystem, which democratized knowledge and allowed thousands of programmers and researchers to build the world of tomorrow, now faces a geopolitical and corporate pincer movement that threatens to put a definitive lock on shared code, transforming a public, human infrastructure into the private preserve of a handful of superpowers.
We have been sold the idea that Artificial Intelligence is humanity's latest miracle, an almost mystical technology owned by a few secretive multinationals and Silicon Valley geniuses. However, if we dig a little into the foundations of this ecosystem advancing at breakneck speed, corporate bureaucracy and contracts armored by lawyers disappear completely.
The collective effort of thousands of programmers
Instead, what remains is the collective effort of thousands of passionate programmers who, often at dawn and from their own homes, decided to share their achievements on public forums and screens full of code accessible to anyone with an internet connection. The great technological revolution of our century did not begin with a secret patent or at a shareholders' meeting; it began in the cradle, always chaotic and collaborative, of free software, and all of that is now in danger, because a set of companies and governments seek to control everything with an iron fist.
According to Visa, AI now buys on its own: more than 109 million machine-to-machine transactions
OpenCV, TensorFlow and PyTorch, the free foundations of AI
For any revolution to take root, it first needs tools that ordinary people can use. In the early 2000s, when Artificial Intelligence sounded more like science fiction than everyday reality, computer vision took its first firm steps thanks to OpenCV (Open Source Computer Vision Library).
Originally born in Intel's laboratories and released to the public in 2000, OpenCV functioned as a kind of universal instruction manual. Its more than 360,000 lines of code allowed academic laboratories with modest budgets and independent developers to teach computers to "see." Engineers no longer had to decipher the complex mathematics behind every pixel for a camera to recognize a face or detect a moving object; it was enough to import an open, free library.
AI accessible to all
But the real turning point came a decade later with Deep Learning. That was when free software stopped being a nice alternative and became a monumental necessity. Large corporations realized an uncomfortable truth: the pace of closed innovation was too slow, too rigid for the real world. They had to share their toys if they wanted to see them evolve. And they did so in a big way.
- TensorFlow: In November 2015, Google released this engine under the Apache 2.0 license. The impact was immediate. What they did was, essentially, democratize the blueprints for building digital brains, taking neural network design off university theoretical blackboards and putting it directly onto the industry's production lines.
- PyTorch: Barely a year later, in 2016, Meta struck back with a tool developed under a completely different philosophy. PyTorch did not seek Google's industrial rigidity, but the scientist's freedom. It was flexible, intuitive, "pythonic" and spoke the same language as researchers. The result of that empathy with the developer was a historic shift in science: PyTorch came to dominate cutting-edge laboratories absolutely, being the engine behind approximately 85% of all academic papers and research papers on Artificial Intelligence in the world.
The era of Free Foundation Models?
None of these projects was a secondary player in the story. They became the scaffolding, the cement and the bricks on which modern AI rests. Without this open, shared ecosystem, the entry price for researching AI would have been an insurmountable economic and intellectual barrier for 99% of the planet, turning tomorrow's technology into the private preserve of a handful of corporations.
Today, the story has taken a huge leap. We no longer limit ourselves to sharing the tools or the blueprints to build the engine; now we share the engines themselves. We have fully entered the era of open-weights models.
On the contemporary stage, these free foundation models have become the vital tissue that allows Artificial Intelligence to breathe and evolve outside the control of private laboratories. They are the shield that allows universities, local governments and small startups to audit the code from the inside, ensure that private data does not end up in someone else's hands and shape the technology to their own cultural and linguistic reality, without having to ask permission or pay a toll to a foreign corporate API.
The West: The fragile balance of openness
This democratization of knowledge has drawn a fascinating technological world map, divided into two great hemispheres competing head to head.
LineShine returns the supercomputing crown to China and challenges the West
On this side of the planet, tech giants and new insurgents play a constant balancing act between commercial profit and the open philosophy. Google leads this front with its family Gemma 4, distributed under a commercially permissive Apache 2.0 license.
What is brilliant about these models is their unified native architecture: they are massive but optimized structures, capable of running locally on a student's laptop without losing an ounce of power.
At the same time, from Europe, France's Mistral AI has established itself as the guardian of the old continent's technological sovereignty. It has shown the world that you don't need a superpower's budget to look the market's most gigantic closed systems in the eye and compete with them on pure efficiency.
The East: The miracle at China's scale
On the other side of the ocean, the adoption of free software is not just a development option; it is a state strategy on a colossal scale. Models like Qwen (the titan born under Alibaba's wing) and DeepSeek's disruptive innovations have burst into the ecosystem with seismic force.
The Qwen series, for example, has surpassed hundreds of millions of downloads worldwide, transforming into the fertile ground on which the Asian ecosystem builds its future: from advanced robotics to complex autonomous agents. What has raised alarms in Western offices is not only that these models match or surpass closed alternatives in critical disciplines such as mathematics or programming, but that they do so at a fraction of their operating cost.
The message in any case is clear: cutting-edge AI is no longer a luxury; in the East, it is public infrastructure.
The storm on the horizon: Geopolitics and the fracture of the open paradigm
However, this idyll of shared knowledge now faces its greatest threat in history. Artificial Intelligence has ceased to be a purely technical discipline or a debate among engineers; it has become the main board of global geopolitics.
The so-called "AI War" is pushing superpowers to treat software and the inner workings of these models with the same suspicion with which they monitor dual-use military technology or critical national security assets. The fear of massive disinformation campaigns impossible to trace, automated cyberattacks and a devastating loss of economic competitiveness is causing suffocating regulatory pressure in the offices of Washington, Beijing and Brussels.
The risk of the great lock: If geopolitical tensions continue to escalate, we face a radical and painful paradigm shift. Governments could impose strict restrictions on the export of code, forcing giants like Meta or Google to shut down future generations of their open models under the eternal justification of "national security."
Burgos will host a forum on AI, biomedicine and digital health in October
If open access ends up being closed off, the balance of power in the tech sector will undergo a total metamorphosis, drawing a much darker and more restrictive scenario:
- The victory of the oligopoly: The fate of technology will be concentrated exclusively in a handful of hyper-wealthy corporations. They will be the only ones capable of affording the training supercomputers and cloud infrastructures needed to sell AI in the form of closed, opaque APIs.
- An asymmetric innovation gap: Small companies, independent researchers and developing countries will be completely left behind in the race. They will stop being creators and become mere customers of technological "black boxes" controlled by foreign powers.
- The rise of digital clandestinity: The persecution of open source will not manage to destroy it, but it will push it into the shadows. We will see the birth of a digital resistance: distributed communities that will share and modify models in a decentralized way, outside official laws and in the deepest corners of the web.
After all, free software built the infrastructure that made the miracle of modern Artificial Intelligence possible. Protecting its openness is not a simple romantic or philosophical stance of developers; it is the only real guarantee that the future of artificial intelligence will remain democratic, transparent and, above all, a heritage that belongs to all of us.