Into energy, AI, hard problems, humans' unique quirks, traveling, surprises, games, startups, late-night philosophical questions, sports, art, and new cool stuff I didn't know about till you just told me
Here to do my part in making tomorrow’s world better than today’s!
Should Europe leave frontier AI to the Americans and the Chinese?
The idea has gained ground in Europe. It is a fundamental mistake.
Build or surrender: for now, the choice is still ours.
Full article below.
ARR is a consequence of problem space understanding, execution on the few axis that matter for your case and distribution through the channels relevant in your industry
start by spinning up the learning flywheel through as much direct contact as possible with the problem, get constant feedback on your solution, then individuate what truly matters to make a difference respect to the status quo and execute on it relentlessly, while you ensure to communicate constantly and frequently about what you’re doing to settle the terrain for when you’ll be ready to rip the fruits of your work
Sorry, Seb, usually lots of love for your work, but how is Europe now having n=2 players in the LLM game a generational fumble for Mistral? Can we please just celebrate having more teams who try and build models, rather than immediately using this as a reason to hate on either? Europe‘s market is big enough for several sovereign players (btw, the world is also a pretty nice TAM) and different models (from my understanding Mistral is working on a much bigger model).
Small bird, fast wings, Kolibri is here.
78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.
Now the weights are yours. Run it on your own hardware, under Apache 2.0.
I applied to YC 4x before I got in
first time was 4y ago, with my current co-founder
what got me in this time wasn't one simple thing, it was the culmination of all that I learned and the momentum I gained over time, made effortlessly pattern-matcheable by the YC partners through every single sentence of the application
what you need is not one trick or good idea, it's just to be better at startups
so here is all I learned in 5y of doing startups
- don't speedrun, take time to compound everything: life, knowledge, skill, intuition, (co-founder) relationships, insight
- don't give up: it will necessarily take an absurdly long time or insane luck to put yourself at the right place at the right time for sure
- don't compare to others: if you can't keep compounding while in a censory vacuum, with no positive feedback for monthes, you can't survive the hard moments. the trick is not to think "gosh I wish i'd be at <goal>" it's to think "only way towards <goal> is to keep going today at least!"
- don't try please VCs, just do something very ambitious and grow good at it, the rest will follow
- meta progression is key: while you work on startups, the odds of failing that individual startup is high, so make sure to not play within your level of comfort, such that you can improve your eq, your skills, and make sure to use this time to grow your social media or IRL cred, your technical skills
- information is what makes you compound: keep learning, asking, noticing, reading, doing the hard search
- moats are built with iterations: what is hard to reproduce isn't your given wealth, intelligence or insight, it's all the intermediary steps to get to a position of power, insight, momentum, that you can't skip, that you must pass one after the other
if you grind all that for long enough you'll gain the level of clarity, insight, and momentum that YC expects
An announcement: Under the leadership of Mario Draghi and @patrickc we have set up the Rhine Group: policymakers, economists, entrepreneurs and business people pushing European reforms and the Draghi agenda.
rhinegroup.eu/
if europe wants to have a say in how such a critical technology (AI) will be developed, we gotta get our shit together *fast* and do a truly outstanding leap forward to catch up (or at least come close) with usa and china
this is about survival. we need true action, not ppts and concerned statements from political figures super far from the technology and completely disconnected from the reality and gravity of our current situation
it's not an easy feat, but if we don't even try, we're gonna be (most probably) doomed to just accept whatever other cultures will decide for us (and tbh deserve to be in that position as so far we played it too safe, lazy and comfortable)
at a high level, i believe that if we wanna have a shot at making this leap, we need to do the following:
- completely change the mindset: we need to think of this technological revolution as a civilization-level matter. not as one of many technological revolutions. THE technological revolution. the consequences of having only a few players leading this change, and none of them european, could mean that our culture and the (in my opinion great) values we've fought so hard for thousands of years to achieve will be just a footnote in the history of humanity
- take risks and aim big: bring out the big bucks from both public and private entities. cut significantly the time to deploy capital. invest in european neolabs building sota models, but don't elect just one champion (=mistral). instead push for a ecosystem of skilled, technical, young or experienced teams aiming *truly* high. competition is good and brings out the best in humans. build extensively infrastructure (energy -> compute) to sustain these efforts
- push for team humanity: at the end of the day, this is a technology that will impact all humans, and therefore all humans should have a say on it. so, europe should try its best to have a say from a technological perspective on AI, but also leverage the other tools it has access to (eg critical companies like ASML or Zeiss, still strong - although declining - purchasing power...) to ensure that usa and china will take into consideration europe's and the rest of the world's opinions and interests in developing these "countries of geniuses in data centers"
après moi j'y vois aussi une méthode, depuis 2 ans musk se colle à tous les mouvements populistes de l'occident, Trump aux états unis, l'afd en allemagne, Farage UK, milei en Argentine et maintenant le RN en France
derrière l'habillage idéologique il y a pour moi une logique froide de businessman, vous remarquerez que ces mouvements partagent un point commun, une fois au pouvoir ils cassent la régulation et la régulation reste la seule chose au monde qui bride encore son empire
d’ailleurs dle pen et bardella (deux gros débiles manipulables) cochent toutes les cases du profil idéal, souverainistes devant les caméras mais déjà arrimés à trump et à la silicon valley dans les faits, le cheval de troie parfait, tu brandis le drapeau d'une main pendant que tu signes la dépendance de l'autre et ça cnest pas con lol
sachez que la souveraineté d'un pays se mesure aujourd'hui ailleurs que dans les discours, elle se mesure en satellites, en lanceurs, en puces, en modèles d'ia, en réseaux qui font circuler l'information (j’en ai parlé des milliers de fois ici )et sur chacun de ces terrains la france joue déjà sa survie
alors imaginez ce qu'un tel pouvoir offre à un seul homme, starlink qui remplace notre réseau souverain jusque dans l'armée, spacex qui envoie ariane au cimetière et tient l'accès à l'orbite de tout un continent, X libéré des règles européennes et transformé en robinet d'influence branché sur nos urnes, xai qui grossit à mesure qu'on démante nos garde fous, son empire prend un étage à chaque fois que bruxelles baisse la tête
et c’est vieux comme le monde hein les empires ont cessé d’annexer par les armes, ils annexent par la dépendance, la compagnie des indes gouvernait un sous continent entier sans être un pays, aujourd’hui un homme négocie d’égal à égal avec des états, pose ses satellites au dessus de leurs têtes et glisse son réseau dans la poche de leurs citoyens, il lui manque juste des gouvernements dociles etil est en train d’essayer dd se les choisir un par un
voilà le vrai visage du piège, ceux qui jurent de défendre la nation seront les mêmes qui la brancheront sur une prise américaine, un faux souverainiste qui détruit la régulation européenne tend le pays à la silicon valley sur un plateau
retenez que la vraie guerre se joue dans les infrastructures, celui qui tient le satellite, la puce et le modèle gouverne un pays bien + longtemps que celui qui gagne une présidentielle et je pense que le jour où on aura intégré ça on cessera de choisir entre deux façons de se faire coloniser
A few words on the Sovereign AI debate, having built several LLMs in Meta while in the UK and now working as a UK based startup:
1. Lots of people are trying to do the right thing to make the UK a better place to start AI companies. Time lags until the benefit show, but you should judge on the intent now. I support the direction of travel!
2. DeepMind has been enormously beneficial for the UK, but it has muddied the waters for a sovereign LLM company to emerge as (until recently) the Government continued to celebrate it as a British achievement / push it as a national champion.
3. Similarly, people are now celebrating recent US investment in King’s Cross, while also wanting more UK sovereignty. Clearly some income effects here, but I would worry about the substitution effects too. AI is not like other types of foreign investment.
4. The relevant talent nexuses in UK that could develop a competitive foundation model are from GDM and old Meta AI GenAI. Also some folks from smaller groups, ex Conjecture, Stability. The talent is still there, although a lot was snapped up by US FM companies in the past year. I personally think it’s not too difficult to develop new talent either from UK universities, but you probably need an ex GDM or Meta core (Gemini or Llama). Or if not: show evidence first (technical reports) before claiming you can do it.
5. Building an LLM is very different from doing regular AI research - skillset is different. Former is closer to engineering; long hours, often unsexy work. Important to distinguish between these two types of talent in the UK ecosystem; arguably too much focus on the latter / ideas guys.
6. On research - DeepSeek R1 post-train cost $300k . Yes, they also needed an ablation budget and to train a base model, invest in infra and talent - and yes the cost of an R1 moment is increasing year on year - but the idea that you need $1bn plus immediately to show results is complete FUD. You need billions to scale, not to validate new directions.
7. In my experience, every failed LLM effort (from model results perspective) I witnessed in the past came from a combination of poor leadership, politics, unclear vision, and premature scaling. Good efforts usually started from small teams who had worked with each other for a long time, had shared thesis, and scaled progressively in bite-sized pieces. Some recent lessons here for neolabs as well.
8. Things take time. Eg we’ve spent ~12 months mostly on internal infra just to get into the position to be able to make big swings. It’s important to nurture new companies through the initial phase. Expectation management is also crucial. I think expecting new UK companies to have single big bang releases is very dangerous; sort of like overwatering a plant. The correct release pattern is “decent”. “decent”, “decent”, “quite good actually”, “holy shit”.
9. Please don’t allow politicians or journalists to kill recent or upcoming AI investment efforts. We will need way more - at the price of potential inefficiency in places - as AI is existential for the country. Ambitious projects are usually incredibly fragile in the early stages; look after them!
10. Mythos is a good triggering moment, but what’s coming will make it look like a toy, so it’s worth building for what’s coming in 5 years time - not a current generation model.
Very proud to be building in the UK - more to share on that soon - alongside many other great early stage AI companies! 🇬🇧
i'm not convinced banning openai/anthropic in EU would be the right move, as they are still great tools that can be leveraged to get back into the game, but i share the sentiment of "we gotta do some bold moves otherwise we're screwed"
- all of this needs to happen in the next 2 max 3 years to avoid the collision of our Titanic with the iceberg and escape the gravity field of the black hole we are currently heading towards if we'd continue with business-as-usual
there's a cultural shift from "everything is doomed and there's no point in even trying to make a change" to "fuck, actually things CAN change if there are competent people that put themselves out there and fight hard to make things happen"
- the public space gains the spark it needs to get finally rid of lot of overhead / regulations and commit resources on a scale that matters, especially after not having to worry of the public opinion previously onboarded because ->
- these positive stories of "Europe is still able to innovate and make new great businesses" impacts culture and unblocks the political conversations we are having, where it's clear that something needs to change but there's not enough reference points to figure out the right dir