“Beware, baffled humans. Beware of false prophets who will sell you a fake future, of bad teachers, corrupt leaders and dirty corporations. Beware of cops and robbers, the kind that rob your dreams.”
— Ryan Ray
This is part one of a two-part series. Part two will be released on August 2nd.
Needless to describe what has happened in the AI world in the past month. I assume you follow the news and spare me the effort. Here, I want to unpack this piece. Occasionally, I might refer to things the news have covered by linking them, if memory allows it.
The Europe 2031 piece is decent, but what it lacks in depth, it covers in mimicry. While predictive scenarios tempt with clairvoyant clarity, stating the obvious is often necessary because “there’s nothing more deceptive than an obvious fact.” I want to say more about the bits left out: the benevolent Leviathan that Europeans could thrive under, if only they were willing to accept a social contract paid for culturally.
Europe isn’t a thing
The United States has fifty states; Europe has forty-four countries, twenty-seven of which are EU members. Pick any two American states and you will find a common denominator homogeneous enough to normalize the “cultural distance” — how many layers of everyday meaning have to be renegotiated when moving from one place to another. You will still find one dominant spoken language, one alphabet and one shared written interface for public life, a shared sense of what a dollar means, a national entertainment canon, a shared calendar of sports, holidays, and rituals, and way more. Now, pick Norway and Romania, or Bulgaria and Ireland, and that cultural distance opens along more axes than it did before. Stand in Romania and look outward and you will see a Romance island in a largely Slavic sea. But do not move too far west and you will see a Uralic island in that same sea. The latter is arguably even more isolated linguistically. Honestly, I do not know which is weirder.
This rough geometry of Europe dictates a perpetual sense of mutual unintelligibility (or maybe negotiated intelligibility) that makes Brussels feel simultaneously overweening and invisible — it governs without ever becoming the government in the way Washington is. And that is because Europe isn’t a thing.
However, the sin is to think of Europe as a failed United States. The apparent inconvenience of cultural difference may turn out to be Europe’s greatest strength. Its plurality enables diversity at a scale almost inconceivable to the American mind. Immigration does not unlock the same ethos. The immigrant diversifies the host society from within. The European neighbor diversifies the continent from beside it. An immigrant and a neighbor are qualitatively different figures, and while both embody difference, only the latter stands beside you as the representative of another whole world.
Constitutional AI done the way Anthropic does it will “align” a system to its creators’ distorted sense of ethics. But I lived long enough in a place that is not a thing and from that vantage point you cannot miss the sunrise: nowhere else is diversity so thoroughly institutionalized as it is in Europe. And that right there could trigger what would go beyond any single continent into an effort meant to demonstrate to cultures all over the world how to distill their cultural uniqueness into a pool of diversity that will align the Shoggoth to humanity and not to what any single company believes.
Europe has an obsession with America
If not a fetishism of sorts. I will weave three threads: financial, political, and, for lack of a better word, clownery.
They say to follow the money. I will do just that. The picture that emerges is clear: the euro favors American markets. These capital flows are a drought. We sow our soil with dry tears. Our intellectual heritage and European pride warm no soul when our technical founders are handed checks lighter than a feather. The same technology that is worth billions across the ocean — why not here?
Those who argue, “Not the same standards of living” or “We obviously don’t need as much of it,” miss the point. A rack of B300s is worth the same inflated sum, regardless of geography.
“Dear Nvidia, we’re a promising European seed startup that just raised €700k to build the next generation of LLMs. We’re active in pre-training. Might we secure that $100M in compute credits to truly get somewhere? Pretty please.”
Take Mistral, Europe’s supposed crown jewel, as a cautionary tale of capital (and influence) drift. I admire Mistral. They have got that undeniable cool factor — dropping their weights as Magnet links, a bold move where I am from. But almost every round that built Mistral had an American lead. European VCs still did not lead their own champion’s largest round. It took ASML to break the pattern. I am not fooling myself. Are you?
On the political front, the contrast is subtler but no less insidious. America has two parties and a handful of lonely independents. I live in Europe and I have no idea how many parties there are. Even ChatGPT says “it’s tricky,” but nonetheless (confidently, but wrongly) claims a total of 886 active core parties across 43 countries.
Here is a thought experiment. You are a tech billionaire. In Washington, you face a binary choice, but in Europe, you are met with a buffet of small, angry parties you can walk into with a single endorsement. Do you think Elon Musk needed to buy the AfD? Certainly not. He only had to livestream with Alice Weidel because American elite attention is political capital. In Europe, every promise factory gets its own version of “Make {Country} Great Again.” A union of nations that is not a thing can be skillfully thwarted by gridlock. And we have already seen that Europe is not a thing. Vance traveled to Munich to declare there was “no room for firewalls,” whatever that truly means. But if you notice, what actually “gets traveled” is attention. In other words, never underestimate the legitimacy of an American nod.
And through it all, we mock. We are excellent mockers. America is the continent’s favorite punching bag. The memes are savage. But jesters were employed at someone else’s court, and the jester who mocks the king in the king’s language, on the king’s platforms, about a culture the king owns, is not a dissident but an underpaid, expendable jester — always one joke away from the gallows.
A view from inside the cockpit
I am not inside the room where the future is being built. I am like someone in the next room over, pressed against the wall, straining to listen and decode what is actually being said on the other side — but everything comes back muffled.
I run an AI deeptech company in Central Europe. It is not a frontier lab or a hyperscaler. Nor is it one of the names that appear in the sort of breathless Wired profile where the founder is photographed staring meaningfully at a server rack to connote visionary intensity. I want to be careful here because I do not want to inflate or performatively deflate our own significance. That might be tempting and almost irresistible. I am merely saying that we are not those people.
But that is not to say that we were not a visionary company or that we could not anticipate the future. I have to say that I have never felt the zeitgeist the way I have felt it in the past two years. Most of my ideas instantly manifested in reality after I discussed them in closed circles in very elaborate and specific detail. It was as if somebody would tune in and listen to my own thoughts. You do not have to believe me. People of far greater impact in my life had seen me at both ups and downs and would back what I write here, as they were there seeing me performing.
We were first in Europe in many ways.
We built software that leveraged large language models (text-davinci-002 and text-davinci-003) before ChatGPT came out, and released it at roughly the same time as LangChain; in fact, we went public with the repository two weeks later.
We were seemingly less lucky, or we did not manage to take off the way a San Francisco-based company would, and that is fine.
We were cited by different labs that managed to raise hundreds of millions of dollars, while we did not manage to raise even one million.
We were also the first to build technology in Europe that was very early at that point and is now behind some ambitious projects.
We did all of this and did not earn much recognition because our geography does not give us an advantage.
Our technical prowess was not enough to navigate the muddled waters of European politics and capital flows.
We were a handful of people with a vision, but we were not the right people in the right place at the right time.
We simply wanted to build technology for Europe and not waste time.
Nobody understood the urgency of the moment, and we were not able to convince them.
Nonetheless, I am grateful for these experiences. I had a surprisingly clear view of the entire room, much as a beggar sees more of the city than a king.
What I saw was not a shortage of intelligence. Europe is capable of leading projects of equal, if not greater, challenge. But Europe is teaching ambition the reluctance of a closed door. It is Europe’s way of catching tigers only to tame them, cutting their fangs and claws and confining them to a space designed to fit an already distorted worldview.
I still remember to this day a conversation that we had with one investment fund from Germany that went quite far. We were discussing for at least six months. Then, part of our team met in person with the general partner, who, two weeks after that meeting, told us that he does not believe we can do it. To this day, I still keep that email as a reminder and as fuel for a fire that will one day burn so brightly its warmth will be felt across the continent.
Ursula von der Leyen declared at the Paris AI Action Summit that the AI race is far from over and that global leadership is still up for grabs. However, she was not describing the world as it is, but as it would need to be for the European self-esteem to remain intact. In other words, she was performing. Europe is not going to win the AI race.