Mistral is right here, their models are in-between the cheap to run Chinese models and top of the line performances of US frontier models.
The safe money is they are going to be an also-ran for the AI revolution. They did manage to force Apple to switch from using lightening connectors to USB though so their wins can't just be laughed off. Maybe they'll surprise us but it'd be a welcome change from their usual routine.
Production of state of the art semiconductors, yes. NXP, STMicro, Infineon are still there and massive in automotive, industrial, card chips, etc.
> The EU fumbled the software revolution, the successes mainly came from the US
Worldwide massive success, mostly yes. Most European countries have their local or regional success stories though.
> The safe money is they are going to be an also-ran for the AI revolution
Not really. Past performances, or lack thereof, are not indicative of future ones.
Mistral are pretty good and selling well in the enterprise space. Some of the best voice models are coming from France (Kyutai).
If you fall out of the state of the art then the claim of EU fumbling semiconductors is correct.
NXP, STMicro, Infineon are still there and massive in automotive, industrial, card chips, etc.
The EU semi companies you listed are absent from the state of the art and only make low margin commodity parts that don't have moats. No, ASML is not enough for claiming EU superiority since the EUV light source is still US IP designed and manufactured.
>Worldwide massive success, mostly yes.
Worldwide success is where the big money is, and you need a lot of money for cutting edge research and experimentation to build the future. Hence the claim of EU fumbling software is correct.
Most European countries have their local or regional success stories though.
EU mom and pop shops aren't gonna make enough money to be able to afford risky ambitious ventures the likes of FAANGs have. Which is why you work for Hashicorp, a large global US company, and not some local EU company.
Who said anything about mom and pop shops? You're arguing in extremely bad faith, as usual with this topic.
Doctolib, Revolut, Adyen, Nokia, Ericsson, Siemens, and tons of others I can't be bothered to list.
> The EU semi companies you listed are absent from the state of the art and only make low margin commodity parts that don't have moats
You think industrial controllers don't have a moat?
> If you fall out of the state of the art then the claim of EU fumbling semiconductors is correct.
Absolutely not. There is more to the world that state of the art.
Care to explain your wild accusations. I never attacked you directly, just the points you made.
>Doctolib, Revolut, Adyen, Nokia, Ericsson, Siemens, and tons of others I can't be bothered to list.
Do those make anything the US or China can't? A doctor appointment scheduling app? Seriously?
>You think industrial controllers don't have a moat?
I never mentioned industrial controllers. Just the chips and microcontrollers those companies make.
>There is more to the world that state of the art.
If you like competing in low margin race to the bottom jobs, sure. Just don't be surprised your tech wages are low then.
You twisted "national successess" to "mon and pop shop". It's a typically American argument "unless it's the global behemoth that has a global monopoly in the domain, it's a failure", which is, frankly, absurd. Would you say Venmo is a failure because they're not used outside of the US (because other countries have better banking infrastructure)? Or that GM are a failure because they barely sell outside the US (because their cars are not adapted to other markets)? Or that United Healthcare Group are a failure because they only operate in the US?
Leboncoin are a massive peer to peer marketplace in France and a few neighbouring countries (IIRC Belgium), like Craigslist or Facebook Marketplace. They do a couple of hundred million in annual revenue. They are, undoutedly, a local success story. Are they a failure because they don't rival Ebay or Facebook Marketplace? No, because that would assume that the goal of each and every business is to become a global behemoth monopoly, which is an impossibility.
Similarly, Doctolib run healthcare appointment and everything related (online appointnments, digital prescriptions, secure storage and sharing of medical data like test results, AI voice note taking assistants for doctos, etc.) in France, and are expanding in a few neighbouring countries. In France they are the standard and pretty much what everyone uses. They are undoubtedly a success.
1. I'm not American, I'm European. And cool it with this finger pointing around nationality as I never brought it up. We can't have a civil discussion if you resort to identity politics as an argument.
2. I said no such thing. I never called those companies failures. You're the one saying that by twisting my arguments.
And those online marketplaces and doctor apps you mentioned that are "local success stories" don't have invented any core tech that can be exported and monetized globally the same like Microsoft, Intel, AMD, Nvidia, etc can. export products abroad, they just used existing FOSS technologies to build some local websites in the EU. Any other country on the planet can build their own versions of those apps, and they have, from India to Argentina. It's nothing special the EU made here. So how you can consider them in the ballpark of the tech companies before is beyond me.
And I didn't say you're American, just that you're using the traditionally American bad faith argument.
> I never called those companies failures
You just called them "mom and pop shops".
> And those online marketplaces and doctor apps you mentioned that are "local success stories" don't have invented any core tech that can be exported and monetized globally
And that's a different argument altogether. Not everything has to be core tech exportable all over, and one can be very successful without doing that.
If you're looking for core tech developed by European countries exported all around the world, enjoy Airbus, Siemens, Infineon, Alstom, Spotify, DeepMind (ok they were acquired by Google), VLC, ASML, SAP and plenty of others.
> Microsoft
> they just used existing FOSS technologies
Can you explain to me the difference between using FOSS and proprietary software to build a product, and what Microsoft are doing?
The obvious takeaway here is that a country / blok can't regulate their way to innovation... so I'm not exact sure why you included it in your list of paradigm shifts. If anything, when the next paradigm shift around charging drops, the EU will be once again on the back-foot due to these short-sighted USB-C regulations they enacted.
I do share your sentiment that EU will miss the train once again on AI.
Then they need money.
So most of the talent flee or get bought, typical example in machine learning space is huggingface or fchollet.
Then European government plays catch-up and offer subventions, but at the same time makes rules to make sure companies don't threaten US dominance, or Asian manufacturing.
Mistral is typically playing catch the subsidy game.
Europe is constructed so that it can't win, but can "pick" the winner between scylla and charybdis, pest and cholera.
Because they have no spine and no leverage/muscle on the international stage to throw their weight around and make sure they get what's best for them at the expense of everyone else the same way US, China, etc do.
They play the international nice guy at the mercy of US tech, energy and defence, and at the mercy of Chinese manufacturing after dismantling their own manufacturing, so they can't do anything radical that upsets either or risk massive repercussions, meaning they get treated as a doormat, constantly being taken advantage of by everyone.
By design European laws are superior to national laws. Leaving the union is also instant bankruptcy because all countries have very high level of debt which are only guaranteed because they are in the union.
European population is getting old and replaced by a migration coming mainly from previous African colonies.
Future paying for the past.
That seems to violate basic physics and accounting laws. It isn't possible for everyone to be in debt all at once, because when everything nets out then there isn't anyone to make the loans. Someone has to be producing the goods that get consumed.
These interests are currently ~2% for France. Which mean the debt is manageable and the interests can be paid with the citizen's tax and the music can continue to play. But once France get out of the UE, interests rates become 5% then the citizens tax are not enough to pay the debt, and nobody wants to lend money to France anymore because even at 5% interests the risk of default becomes too great and they risk not getting the full amount-owed back so nobody lends, and since their is no money in reserve, and they can't borrow it means they default => bankruptcy. France doesn't have its own currency anymore so it cannot print its own money which compounds the problem. National resources get plundered, citizens get poor.
It is a game of musical chair which is highly non-linear.
Uhm, Europe is not the US. We still have a lot of manufacturing. It varies by country - the UK unfortunately had structural problems, finance supremacy and a Thatcher who hated unions so much that she'd rather destroy unionized industries than have unions. Central Europe still does a pretty large amount of manufacturing.
Then why are we afraid of China and the US and cave in to their demands?
Why is german manufacturing output back to where it was in 2006?[1]
[1] https://x.com/ThorstenPolleit/status/2047436171903394294/pho...
Meanwhile Germany's and the EU's short sighted energy policies have caused a lot of manufacturing to move outside the EU to places like the US or China.
The EU is in a very bad position right now manufacturing wise compared to how it was and how it could be. It basically never failed to snatch defeat from the jaws of victory for the last ~20 years.
Europe is constructed so you can take 60 days vacation, work 32 hours a week, get tons of social benefits, can't really lose your job, and retire when you are 65 with a full pension.
Which is excellent. Unless you need to be economically competitive.
Not a thing.
>work 32 hours a week
Not a common thing.
>get tons of social benefits
That you pay for in high taxes.
>can't really lose your job
Layoffs happen at the same rate as elsewhere.
>retire when you are 65 with a full pension
Unless the government decides to push back your retirement because it's insolvent.[0]
[0] https://en.wikipedia.org/wiki/2023_French_pension_reform_str...
Mistral is good for many tasks where you do not need SOTA or near SOTA performance. They cannot compete if you do.
Of course not. When it comes to SOTA LLMs you have the choice between two bad options. For many, choosing the Chinese option is just choosing the lesser of two evils (and it's much cheaper).