Every AI Model You've Ever Used Runs on a Machine Only One Company Knows How to Build.

There is a machine that makes the modern world possible.

It is called an extreme ultraviolet lithography system. It weighs approximately 180 tonnes. It costs around $350 million. It takes a team of specialists months to install and calibrate. It is manufactured in Veldhoven, a small city in the south of the Netherlands that most people could not locate on a map without help.

Only one company in the world knows how to build it.

That company is ASML. And if you have used ChatGPT, or an iPhone, or any piece of consumer electronics made in the last five years, you have used something that would not exist without ASML's machine.

This is worth sitting with for a moment. The entire AI industry — OpenAI, Anthropic, Google DeepMind, xAI, every model being trained on every cluster of GPUs in every data centre in the world — runs on chips. Those chips are designed by companies like Nvidia, Apple, AMD. They are manufactured by companies like TSMC in Taiwan and Samsung in South Korea. But before any of that manufacturing can happen, somebody has to print circuits onto silicon wafers at a scale so small that the wavelength of visible light is too large to be useful. That is what ASML's machines do. They use extreme ultraviolet light — a wavelength of 13.5 nanometres, shorter than any light that occurs naturally on Earth's surface — to etch patterns onto chips with a precision that has no parallel in any other human manufacturing process.

Without ASML, there are no advanced chips. Without advanced chips, there is no AI.

ASML did not set out to become the fulcrum of the global technology supply chain. The company was founded in 1984 as a joint venture between Philips and a Dutch tool manufacturer called ASM International — hence the name, which stands for Advanced Semiconductor Materials Lithography. The EUV technology that makes the current generation of chips possible took decades to develop, cost billions of dollars in research, and required the solution of physics problems that many experts considered unsolvable.

The machine that resulted is not just a product. It is a technological achievement so singular that geopolitics has reorganised around it.

In 2019, the United States government pressured the Dutch government to block ASML from shipping its most advanced EUV machines to China. The argument was straightforward: whoever can manufacture the most advanced chips controls the most powerful AI systems, and the most powerful AI systems will determine military and economic dominance in the coming decades. ASML's export licence was denied. China's most advanced semiconductor manufacturers are frozen at a generation of chip technology that predates EUV.

This is, by any measure, one of the most consequential technology export decisions in history. A single Dutch company's export licence has become a primary instrument of great power competition.

ASML is currently valued at approximately €250 billion, making it the most valuable company in Europe. There are analysts who believe it could become Europe's first trillion-dollar company as AI infrastructure spending continues to accelerate. Every new data centre, every new GPU cluster, every new model that requires more compute than the last — all of it increases demand for the chips that only ASML's machines can produce.

The competitive moat is not a brand or a patent or a market position. It is accumulated knowledge. The physics, the engineering, the manufacturing process, the supply chain — all of it exists inside ASML and nowhere else. A competitor would need decades and tens of billions of dollars to begin to close the gap, and ASML is not standing still.

There is something about this story that I find personally resonant, and I want to explain why.

FreeMalta started with a name that sounded like a disadvantage.

Malta is a small island of approximately 600,000 people in the middle of the Mediterranean. When people encounter the name FreeMalta for the first time, the assumption is sometimes that this is a local platform, for local information, of interest to a local audience. The name, on the surface, appears to limit the ambition.

It does the opposite.

The "Free" in FreeMalta does not mean without cost. It means without constraint — free access to information, free of the paywalls and gatekeepers and consultancy fees that have historically made it difficult for people to understand how to live, work, and build businesses in Malta. And the "Malta" is not a limitation. It is an origin story. It is the specific place from which a platform that now covers company founding stories, global markets, international brand partnerships, and an AI newsroom that publishes 43 digests daily decided to begin.

ASML is from Veldhoven. Nobody knew where Veldhoven was before ASML. Now Veldhoven is on the geopolitical map of the AI era because a company that started there built something that the rest of the world could not build without.

FreeMalta is from Malta. The name is also, once you hear it, not easy to forget. Something about the combination — the specificity of the place, the implied contradiction of a small island producing something global — creates a memory hook that a more generic name would not.

ASML proved that where you are from is irrelevant if what you build is singular.

We are trying to prove the same thing.

In the meantime, the next time someone asks you how AI works — not philosophically, but physically, materially, in terms of what has to happen in the world for the words to appear on your screen — the honest answer starts in Veldhoven.

With a machine that weighs 180 tonnes and costs $350 million and that only one company in the world knows how to build.