Recipient of Europe's Largest Ever Seed Round Doesn't Even Have a Product (theregister.com) 40
An anonymous reader quotes a report from The Register: The French recipient of Europe's largest ever seed round doesn't have a product and was founded four weeks ago. The few employees it has only started work in the last few days. All the same, Mistral AI hoovered up [$113 million] from lead investor Lightspeed Venture Partners with contributions from Eric Schmidt, French telecoms billionaire Xavier Niel, and state-backed investment bank Bpifrance. The clue as to why the Paris-based startup is now valued at [$259 million] could be as simple as its name. New AI companies in the US inhaled as much as $25 billion from venture capitalists in the first three months of 2023, and with the artificial intelligence bubble centered on the States, Europe has yet to respond in kind.
There is also the small matter of personnel. Mistral AI chief exec Arthur Mensch used to be a researcher with Google DeepMind and founded the new company with Meta alumni Timothee Lacroix and Guillaume Lample. Presumably these credentials, combined with the promise of a homegrown European AI, are all it takes for investors to part with millions these days. Lightspeed partner Antoine Moyroud sure seems to think so. He told the Financial Times "There's a pool of 80 to 100 people globally who have the level of experience they have. Right now, for better or for worse, the capital requirements in compute and top-tier talent make [launching an AI startup] quite a capital-intensive game."
The promise of Mistral as it stands four weeks after it first emerged blinking in the daylight is another "large language model." That's it. This is the same type of technology that underpins OpenAI's ChatGPT -- a neural network of tens of million to billions of parameters trained on large quantities of unlabeled text using self-supervised learning or semi-supervised learning. That is all we know of the untold riches held in the minds of Mistral's power trio. Its website, a minimalist single-page affair, says little other than: "We're assembling a world-class team to develop the best generative AI models." It also provides an email address, asking for candidates with "a strong background as a researcher, software engineer or product developer in AI."
Mensch said: "There is a rising awareness of the fact that this technology is transformative and Europe needs to do something about it, both as a regulator, as a customer and an investor." Tech investment tracker Dealroom.co pegged the seed round as Europe's largest, which is indicative of the hope that AI will positively transform entire industries by taking humans out of the equation -- computer programs don't eat, sleep or unionize. So while venture capitalists are feeling FOMO and paying high prices to get a foot in the door of the latest tech gold rush, Europe is also pretty desperate, with only $4 billion stuffed into AI businesses this year compared to $25 billion in the United States. [...] Now you don't even need a product, or even the promise of something groundbreaking, just a handsome CV and "AI" in the company name. What we on the outside do know, however, is that at least one person on the Mistral team knows some HTML.
There is also the small matter of personnel. Mistral AI chief exec Arthur Mensch used to be a researcher with Google DeepMind and founded the new company with Meta alumni Timothee Lacroix and Guillaume Lample. Presumably these credentials, combined with the promise of a homegrown European AI, are all it takes for investors to part with millions these days. Lightspeed partner Antoine Moyroud sure seems to think so. He told the Financial Times "There's a pool of 80 to 100 people globally who have the level of experience they have. Right now, for better or for worse, the capital requirements in compute and top-tier talent make [launching an AI startup] quite a capital-intensive game."
The promise of Mistral as it stands four weeks after it first emerged blinking in the daylight is another "large language model." That's it. This is the same type of technology that underpins OpenAI's ChatGPT -- a neural network of tens of million to billions of parameters trained on large quantities of unlabeled text using self-supervised learning or semi-supervised learning. That is all we know of the untold riches held in the minds of Mistral's power trio. Its website, a minimalist single-page affair, says little other than: "We're assembling a world-class team to develop the best generative AI models." It also provides an email address, asking for candidates with "a strong background as a researcher, software engineer or product developer in AI."
Mensch said: "There is a rising awareness of the fact that this technology is transformative and Europe needs to do something about it, both as a regulator, as a customer and an investor." Tech investment tracker Dealroom.co pegged the seed round as Europe's largest, which is indicative of the hope that AI will positively transform entire industries by taking humans out of the equation -- computer programs don't eat, sleep or unionize. So while venture capitalists are feeling FOMO and paying high prices to get a foot in the door of the latest tech gold rush, Europe is also pretty desperate, with only $4 billion stuffed into AI businesses this year compared to $25 billion in the United States. [...] Now you don't even need a product, or even the promise of something groundbreaking, just a handsome CV and "AI" in the company name. What we on the outside do know, however, is that at least one person on the Mistral team knows some HTML.
no business plan (Score:4, Insightful)
https://slashdot.org/comments.... [slashdot.org]
I think they don't have a business plan, never mind a product.
They provided an *idea* at merely 130 million euro. It's that time of the year again, apparently.
Re: (Score:2, Funny)
I think they don't have a business plan, never mind a product.
They're consulting ChatGPT about that right now.
And, since they're French, they're disappointed that ChatGPT has nothing to do with cats.
Re: (Score:3)
It's confirmed by Netcraft anyway; the next hype buzzword after "crypto" is AI! We need to keep the economy rolling and nothing better than hype and buzzwords to make this happening!
Oblig [Re:no business plan] (Score:2)
1. Hint at plan for some kind of big EU-flavored AI engine
2. ???
3. ??????
4. Profit!
Re: (Score:2)
But what do we do with all these underpants?
Re: (Score:1)
Same thing you did for the last failed boondoggle.
Re:no business plan (Score:4, Interesting)
They're a tech startup. They don't need a business plan, or a product. All they need is a story to establish what they're "worth", which has nothing to do with how much revenue they are taking in (if they ever take any in).
Re: (Score:3)
I think their business plan is to build the best possible LLM that complies with EU regulations. And I suspect the French government will help tweak the regulations so that they comply, but Open AI, Google, et al, don't.
Re: (Score:1)
wo the government is usung them to create a monopoly, I agree. They will build their own version of something, but what is the business plan, will they then become the only government approved provider (and all others will be outlaws)?
What will the French do when everyone runs their own version of a system similar to that but not certified?
Re: (Score:1)
Might work (Score:3)
Re:Might work (Score:4, Interesting)
The EU is not too happy with Big Tech from the US
Europe relies on American (and Chinese) big tech because they drown their own tech with overregulation. They are doing the same with AI.
EU AI Act [europa.eu]
Re: (Score:2)
Ah, yes. That must be the reason why there is no meaningful tech research in Europe and no tech products of any value are made here.
Re: (Score:3)
Ah, yes. That must be the reason why there is no meaningful tech research in Europe and no tech products of any value are made here.
You are mistaken. There is indeed a lot of tech research in Europe, but few graduates stay. Much of US tech relies on European talent.
Re:Might work (Score:5, Informative)
I am not mistaken. I am sarcastic. And actually, most graduates stay. The US is getting progressively less attractive for engineers and scientists from Europe.
Re: (Score:3)
There's plenty of EU tech, it just tends to be less consumer facing.
The other issue is that investors are more cautious. In Silicon Valley they will throw huge wads of cash at anything, on the off chance it becomes the next Google or Facebook. That's starting to change now, as this story demonstrates.
Re:Might work (Score:5, Insightful)
It's very much a self-induced problem on the EU's part. VCs don't really care where the business is located, just so long as it's in a business friendly legal jurisdiction with a reasonable ability to attract engineering talent, and little/no risk that it will be arbitrarily nationalized by some jealous government politicians. France in particular seems to think it's wise to be openly hostile to all of that, just like they're openly hostile to any changes to their language [ycombinator.com], and then they wonder why they have to spend millions of tax dollars (euros?) on projects like Quaero [wikipedia.org] in order to restore their sense of national pride, only to have it inevitably collapse anyways.
Re: (Score:3)
It's very much a self-induced problem on the EU's part.
Why is it a problem to protect society against aggressive max-for-profit foreign corporations? And all the while the US is wielding protectionism on a massive scale.
so long as it's in a business friendly legal jurisdiction
'Friendly legal jurisdiction' is just an euphemism for paying little to no taxes...
and little/no risk that it will be arbitrarily nationalized by some jealous government politicians
This is shit you pulled right out of your ass. The EU doesn't nationalize corporations. The EU isn't even a nation.
Re: (Score:2)
It's not a bad idea. The only way realistic way to pay for mass access to LLMs is going to be monetizing user data in any way possible, and foreign companies doing that is not going to be compatible with EU privacy laws. LLMs aren't hard to build, so you've got pretty easy product development and a captive market.
Who is the one complaining? (Score:3)
No product yet, but when they build it... (Score:5, Funny)
it's going to be AMAZING! They just have to figure out what "it" is first!
Where do I send my money???
Re: (Score:1)
Seems "AI" kills natural general intelligence... (Score:2)
At least it seems that in the current "AI" hype is getting increasingly mindless, with people throwing money at it expecting "AI" to deliver things it cannot, like "AGI", and without good reasons and against explicite warnings by the experts. (No, I will not discuss it. No currently known "AI" approach can deliver AGI. Deal with it.)
"Fear of missing out" is going through the roof here and makes people doing really stupid things. Well, at least we will get a spectacular collapse of that house of cards eventu
Scams have always existed (Score:1)
Re: (Score:2)
So it's the same old shit as always
Indeed, it is. That nicely sums it up.
Mistral is a wind (Score:5, Interesting)
in French we have an expression... "vendre du vent", which translates to selling bullshit. seems pretty fitting
Things to look for (Score:1)
AI bubble (Score:3, Interesting)
Re: (Score:2)
I think you underestimate the effect of image generating tools.
I used to work in VFX and animation as a 2D artist. I went into early retirement, so I'm glad I don't have keep up or compete with these technologies, but I have friends who work in various related fields, in games, in film, and they're feeling threatened.
One friend works at a games startup, and they're using a lot of AI generated images for concept art.
I have illustrator friends who are getting less work because for a lot of people AI generated
Re: (Score:2)
A demonstrably false statement. You might say it doesn't have *enough* democratic process, or that its democratic process doesn't wield enough power, but they have a parliament that is elected by the general population every five years.
Re: (Score:2)
So your position has changexd from "they have no democratic process whatsoever" to "their democratic process doesn't have enough power".
They missed quantum, crypto and blockchain (Score:3)
"An undertaking of great advantage" (Score:2)
"But nobody to know what it is."
Typical EU Behavior (Score:2)
Oh no, America (or some other non European country) has invented some new technology for pennies in a garage! We must throw billions at this problem and come up with our substandard competitor!
me too (Score:2)
Everybody wants a seat at the local craps table.