Even if it's not the best model, it can be really important step in UE sovereignty. Trained in EU, inference in EU. I guess it will matter for some companies. Hope Mistral won't disappear for the next half year.
This isn't 2016 anymore.
you could say, that's perpetually a tomorrow problem, so long as they never go public, but that should illuminate for you: if everyone "knows this," it's inevitable that this so-called EU company, that didn't invent any of the AI, the hardware nor the training data, where their product is essentially more like a VPN provider than a frontier technology company, just lists and capitalizes in the US anyway.
By your measure anyone unable to build EUV lithography machines without ASML help is doomed. EU could easily corner the market and shut them all off. No more NVIDIA.
Let’s grow the pie, shall we?!
honestly the craziest thing to me is that VC people in the EU, they say, let's grow the pie, and when you look at their little decks with their little flags next to the portraits, it's french, german, italian, spanish, all piling into the same 40 or so companies, which will 110% list in the US in order to exit. Meanwhile, EU member Poland produces tons of entrepreneurs, despite no Polish flags in those growth fund decks, and its best people just leave to the US and raise money there and make fabulously successful stuff.
so yeah, "grow the pie" except for Poland. EU relitigating the same feuds it has been for centuries.
That's true for every EU country currently. See Docker's history for instance. When you raise your first rounds in the US, you have to incorporate there.
didn't mistral invent the MoE architecture?
For Europe, it's "capital, capital, capital".
Europe has lots of capital, it's just fragmented across 30 odd countries and with a huge proportion being invested in US assets.
The last 3% of the stack? They’re essentially borrowing Chinese open source distillation/innovation off the US frontier running on US/Taiwanese designed chips and calling it EU made. This is better than nothing of course.
But is sovereignty really an end in itself? So the EU becomes IT independent…cool, then what? I mean the Yugo was sovereign, it didn’t do much good when the society itself failed to produce prosperity and collapsed.
It seems to me the EU is expending enormous effort on the appearance of “sovereignty” over what is ultimately…the last-mile consumer toilet paper purchasing conversation data…of an aging, increasingly irrelevant population on the political/economic stage.
Meanwhile domestically the entire economic model is failing and the “union” is getting shakey as its 2 biggest members turn more nationalist.
Maybe this “oh you cant compete but here’s a trophy for sovereignty” attitude isnt helping. Less clapping along with the EU bureaucracy’s latest make work projects like transitioning to a new Word processor. If we want the EU to succeed, more tough love is needed imo.
hard to stand up for EU (or even national) values if some dude in Washington has the keys for most of your military
having at least some in-house expertise is the first step.
We should be shaming them into contributing something useful to the rest of world in the most important development of our current time (AI) instead of clapping for domestic regulatory moats.
From where I'm standing, as an American, we should be shaming OpenAI and Anthropic for both pushing towards an IPO while neither of them have profitable revenue streams. In pure game-theory terms, you could successfully argue that the US is the biggest loser of the AI race: an economy propped up on artificial Nvidia margins and terminally reliant on neoliberal trade.
Even if global war doesn't tear down the TSMC house of cards, neither the government, the banks, nor the global economy could underwrite an OpenAI IPO. They make no money!
Like, adjusted can cover a whole host of sins. Until we see the GAAP audited numbers, nobody but Anthropic employees and investors know.
“While Anthropic offers more to society than WeWork ever did, at a $2 trillion valuation, its IPO presents far bigger risks and is positioned to be a far bigger rip off of U.S. capital markets,” New Constructs wrote, adding that the IPO’s purpose isn’t to provide wealth for public markets investors, but rather liquidity for the company’s Wall Street backers.The fact you’ve used the trendy populist boogieman word “neoliberal” and complained about a generational startup not being profitable during the growth phase tells me you watch a lot of internet content from people who don’t understand the topics they cover (breaking points perhaps?)
I would recommend re-evaluating who you borrow your opinions from. Global trade is not a boogieman, and OpenAI will have zero trouble building a profitable consumer advertising business, regardless of whether their IPO flops in the short term.
I didn't make a mistake, or reach into the buzzword hat for that; Nvidia's trillion-dollar GPGPU products are wholly dependent on TSMC. Every single Blackwell die ever made came out of a Taiwanese factory, and the US leadership is inviting rapprochement with the CPP on the eve of Taiwan's invasion. TSMC will destroy their machines before China gets them, and then there are no more Blackwell products. US fabs do not ship a comparable fabrication node and likely won't for another ~5 years, optimistically speaking. This is not a "war will make the stocks tumble" argument, we are describing a scenario where American economic growth could slow or stop entirely if AI dominance is jeopardized. The US economy can't have an Achilles heel like that.
It gets worse! There is currently a global DRAM, SRAM, HBM and flash memory shortage that has already encouraged Samsung, SK Hynix and Micron to start fucking over their most loyal customers like Apple and Google, who have absolutely no alternative vendors. If you're packaging SOCs, neoliberalism already fucked you over! South Korea is counting beaucoup bucks for every iPhone sold at eye-watering prices. Taiwan is gouging US OEMs, making Apple and Nvidia engage in humiliating bidding contests to secure access to a single factory line. If global conflict expands, the AI economy will be chronically dependent on government (read: taxpayer) money. US taxpayers don't have trillions of apology dollars for this type of mistake.
China's got the edge on open models. Free trade does their heavy lifting, US labs are scrambling to put up defenses and beg for pathetic regulatory efforts that pull up the ladder. Yes, neoliberalism keeps them up at night, and apparently it keeps you up too.
They should take a cue from the US, and brain drain the rest of the world and coast on the success of brilliant US scientists and entrepreneurs like Linus Torvalds, Yann Le Cun or Terence Tao.
At the same time France is ablaze now by younger generation asking for a better future while the public spending is 57% of the GDP in 2025[0] and stuck in a vicious cycle.
Only Germany is holding the fiscal stability by a thread and that's even questionable given the current German government is more like a zombie government that cannot agree on the important reforms and most probably would collapse before the end of this year.
It's completely valid to question how "Sovereignty" helps EU when the complete backend (chips, energy, technology, etc.) is imported from US or China, Middle East and just slapping a Sovereign sticker at the frontend on a mediocre product that even most of the europeans don't want to use given the alternatives.
[0] https://tradingeconomics.com/france/government-spending-to-g...