A European Vision for AI: Navigating the Financial Reality
In the AI Innovation Center on the High Tech Campus in Eindhoven, Bernardo Kastrup exudes the composure of someone who has lived multiple lives. A computer scientist, philosopher, ASML strategist, and founder, he now reluctantly takes on the role of Europe's champion in reinventing AI computing from scratch.
Euclyd, the company he founded a year ago, has quickly become a sensation. Backed by industry giants like former ASML CEO Peter Wennink, Intel engineer Federico Faggin, Silicon Hive's Atul Sinha, and Elastic founder Steven Schuurman, the startup emerged from stealth mode into the European tech scene earlier this year, instantly dominating the conversation. It wasn't just the ambition; it was the tone: confident, technical, and boldly ambitious, a European company speaking with Silicon Valley panache.
"Yes, you can call it boasting," Kastrup laughs. "It's tongue-in-cheek. But we should be confident. Why should only Americans be allowed to speak with ambition?"
Beneath the humor lies a deeper frustration, the catalyst for Euclyd's creation.
"I looked around and asked: Who allowed this to happen?"
When generative AI exploded into public consciousness in 2023, Kastrup noticed a stark contrast in the European landscape. World-class researchers, cutting-edge chip equipment, and a sophisticated industrial base, yet no serious attempt to build the next-generation silicon to power future data-center AI.
"I remember looking for someone to blame," he says. "Why is nothing happening in Europe? How did we allow that?"
The realization hit hard: no one else was to blame. "Who could do something about it? People like me. People I know. So yes, I felt a responsibility."
This sense of responsibility led to a prolonged period of secrecy. Euclyd wouldn't announce anything until the company had concrete proof of its idea's feasibility, a proof that Kastrup and his friends could trust. "When you take money from a VC and fail, that's the risk of the business," he says. "But losing your friends' money? That's something else."
For months, the small team, many of them former chip venture colleagues, worked on a groundbreaking AI inference architecture. No legacy GPUs, repurposed gaming hardware, or shortcuts. Instead, a system built from the ground up. The early design work even took place in Kastrup's attic, where he set up a simulator and spent countless nights sketching microarchitectures, much like others sketch ideas in a notebook.
When the test chip emerged, and Samsung agreed to produce it, Euclyd stepped into the spotlight.
A European Chip with Global Ambitious
Euclyd's promise is substantial: extremely energy-efficient inference, potentially one hundred times more efficient than Nvidia's current data-center chips, achieved not through magic but through architectural sanity.
"Nvidia built for video games," Kastrup says. "When large language models arrived, they were simply in the right place at the right time. But pretending a neural network is a video game with a global variable space... that's just about the worst thing you can do for efficiency."
Euclyd's architecture does the opposite: no off-the-shelf IP reuse, no reliance on generic bus architectures. Everything is specialized, deeply pipelined, built for neural inference first. The test chip contains 64 processors; the product version will scale to 16,384.
In a world of incremental optimization, Euclyd is attempting the 'forbidden move': starting from scratch.
The Dream: A European Nvidia. The Reality: Global Capital
Despite the technical bravado, one topic makes Kastrup pause, choose his words carefully, and speak not as an engineer but as a founder navigating forces beyond his control.
The money.
Euclyd is raising a major financing round, and with it, the possibility—perhaps even the likelihood—that the company will no longer be 'purely European.'
"The original dream was to do AI in Europe," he says quietly. "To have Europe play in that space. But when you have a business, and you're spending other people's money, you have fiduciary duties. You must follow business sense. And that may put us in a situation where we must act in ways that are not fully aligned with the dream."
It's a reality he acknowledges: "We are raising money. This is a situation we may confront—or have already confronted. That's part of the game."
He pauses, emphasizing his point. "I can dream, right? I can dream whatever I dream. But I cannot, on my own, guarantee that the dream will become or remain true. I do not control the world."
It's a rare admission in the European tech landscape, where idealism often prevails. Kastrup refuses to pretend the tension doesn't exist. He wants Euclyd to be the European Nvidia, built on European engineering and rooted in the Brainport ecosystem that shaped him. But he also wants the company to survive the global race for compute, a race Europe has joined late and underprepared.
If the right investors come from outside Europe, and if refusing them would threaten the company's ability to compete, what should a founder do? "I will do the best I can within the boundaries of what is allowed or required of someone in my position," he says. "But I cannot control everything."
A Dream Worth Fighting For
The future of Euclyd as a European company in ownership structure remains uncertain. The likelihood of its technology becoming global is increasing daily. Whether Europe will finally have its own AI-compute champion may depend not just on Kastrup's design brilliance but on Europe's willingness to move at global speed.
Despite the challenges, the dream persists. "Europe needs to play in this field," he says. "It's my wish too."
With that, Bernardo Kastrup—philosopher, engineer, and founder—returns to his office, where a small test chip no larger than a fingernail awaits. The architecture within it could reshape AI's global energy footprint. The question remains: Will Europe still be the place where that future unfolds?
