[rzerowan]

As far as Europe is concerned they have recently signed up to the 'Pax Silica'[1] and willingly givrn the LLM space over to the US incumbents with buildtin legislation banning Chinese models and coperation with them. So EU will be a renter of the LLMs that the US allows them to use. In the long run OpenSource will dominate as it did in the DB(MySQL/Postgres)/ServerOS(Linux/BSDs) versus Proprietery rent seeking alts like Oracle and Microsoft et al. Would be interesting to see what the global startups using Qwen/DS/Kimi etc within the EU-US space navigate the cutting edge OpenSource LLMs vs seeking/getting a permission slip from the US gov.

[1]https://archive.is/aiJiq

[tadasv]

I hope that open models will dominate. The difficult part to reconcile for me is the amount of compute that's required to create and run such models. Small models are fine, I run local llms 27b param on a gpu, but it's not even close to frontier in capability. Who wants to drop $40k+ on hardware to run these things. Companies, maybe/perhapts. On the other hand, to run a DB I can get a server for $3k and handle tons of traffic on it and other things too.

[Avicebron]

Has anyone tried to run a data center as a Co-Op?

[fc417fc802]

A data center or a cloud? It's not difficult to find a good data center to colo at. The problem is then you have to bring your own hardware, technicians, and sysadmins.

However, if you don't trust cloud providers or inference providers for whatever reason then you probably aren't going to be excited to enter a co-op model where you're still effectively renting access to hardware that you don't directly own. There are already reasonably priced options to rent bare metal from a cloud provider.

The only way I see it working is if it's a bunch of medium to large sized businesses getting together to be able to rent out the spare capacity on hardware that they physically control. So an AWS equivalent where each rack is owned by a different company and retail VMs migrate between them transparently. But I question the overall economics of such an arrangement.

[embedding-shape]

> enter a co-op model where you're still effectively renting access to hardware that you don't directly own

If we're imagining a co-op, then the participants should all be equal owners in an organisation that owns the hardware itself, otherwise it's not much of a co-op really.

[fc417fc802]

But at that point you don't have the same sort of security guarantees about the hardware. This isn't like a farming co-op where a single expensive piece of equipment gets passed around to each participant one week at a time. There needs to be either an economic, security, or other advantage to entering into this arrangement for a medium to large sized player that would otherwise be colocating multiple racks in a privately owned datacenter or else renting bare metal instances in bulk from one of the hyperscalers.

[mattlutze]

Participants as equal owners doesn't preclude security guarantees.

The advantage is having a voice in setting direction, investment and ultimate shared incremental profit recovery over the rent capture from hyperscalers.

It's an interesting idea really. But it also comes with cooperation, which means trust, and a lot of for-profit enterprises are bad at operating cooperatively outside of 2- or 3- way engagements.

[orbifold]

I was thinking that would be great, too. What would be the equivalent for the property developer: one gpu server is 450k.

[rzerowan]

Would actually be a good biz model for the Colo facilities that keep shutting down as everyone moves to the big cloud providers.Now if they can get their hands on enough GPUs and RAM.

[eru]

Why would it be a good business model?

[danielheath]

I think it aligns incentives way better (but is almost impossible to set up).

[eru]

Incentives are pretty easy to align, if you have repeat customers and repetition. McDonald's sells me burgers I expect at a reasonable price, because they want me to come back tomorrow and not go across the street to Burger King.

[orbifold]

can you give examples of colo facilities in SF / New York that are shutting down?

[rasur]

coopcloud.tech ?

[rzerowan]

I believe until the hardware designs catch up to be more commodized ala cryto mining evolution from GPUs to ASICS for specfic algos. Designs (like Google TPUs equivalent) would also need to evolve to be more memory dense to be able to handle them. Untill then it seems will be system time shares for the larger models , probably with a bring your own model and pay as you go.

[Dylan16807]

> ala cryto mining evolution from GPUs to ASICS for specfic algos

I don't see it happening. A current gen GPU with a huge and fast block of memory isn't a perfect fit for these algorithms but it's relatively close. With cryptocurrency, mass small sha256 hashing was a totally different kind of computation.

[fc417fc802]

> isn't a perfect fit for these algorithms but it's relatively close

I don't think that's true. The best fit out of what's presently available perhaps. Inference is almost entirely memory bandwidth bound at present, to the extent that GPUs with HBM have a massive advantage over those with GDDR. TPUs appear to be a much better overall design.

I expect that a hypothetical advance in fabrication enabling processing elements to be placed directly adjacent to dense RAM on the same silicon (not merely in the same package) would be superior in all regards.

[Dylan16807]

> I expect that a hypothetical advance in fabrication enabling processing elements to be placed directly adjacent to dense RAM on the same silicon (not merely in the same package) would be superior in all regards.

Processing scales better than DRAM does. I think an HBM-like stack where the bottom layer has the math units is probably the ultimate form of that.

And it's possible that flash instead of DRAM is actually the better play, as long as you can hook up enough in parallel. RIP Optane.

[wcoenen]

> I don't see it happening.

Isn't it already happening with Cerebras? It's mentioned at the end of OpenAI's GPT 5.6 announcement:

"We're also launching GPT‑5.6 Sol on Cerebras at up to 750 tokens per second in July"

[bluecalm]

There might be a community effort at some point. This happened in chess where the community recreated and then improved on Alpha Zero. You could run small training chunks on your machine. Some people donated thousands of hours of server time.

[tsss]

They never will. The only reason China releases open weights is because they can't compete on frontier models. Whoever has the frontier model has no incentive to give it away for free.

[swiftcoder]

They have signed a non-binding agreement to potentially cooperate on AI supply chains. It's hardly a declaration of fealty, nor does it have any practical impact on the use of Chinese models in the near term. I'd view this more as hedging their bets for the future.

[Matl]

> It's hardly a declaration of fealty

As a European, the way I see it, Europe declared fealty to the US and relinquished its sovereignty a long time ago, sadly.

Also, the way these agreements tend to work is that you agree that you won't source from the 'enemy side' i.e. China. It works this way for NATO and it will work the same way here.

[fauigerzigerk]

>As a European, the way I see it, Europe declared fealty to the US and relinquished its sovereignty a long time ago, sadly.

No such declaration was ever made. Europe has just failed to compete.

[Matl]

> No such declaration was ever made

Of course not, there's no need for a 'declaration' that would be a hard sell to the European public. Actions is all that matters.

[voakbasda]

Indeed. The system’s purpose is what it does.

[gunsle]

Facts, but the Europeans, tankies, and America haters here will never admit it. Europe sold out its industries for social programs. Not our fault you guys chose social safety nets over industry.

[dadoum]

I don't think this narrative stands, industry and social programs coexisted for a very long time. I actually believe the opposite. It sacrificed its industry for intra-EU fairness and integration. A lot of industries were very reliant on gov grants and other state social programs that have got scarcer as EU gained power.

> Facts, but the Europeans, tankies, and America haters here will never admit it.

When disagreeing, please reply to the argument instead of calling names. "That is idiotic; 1 + 1 is 2, not 3" can be shortened to "1 + 1 is 2, not 3."

[gunsle]

I’m not name calling, I’m describing. If you’re offended, you must be a tankie, lol.

[Matl]

I think Americans like yourself greatly underestimate what advantages having a unified language, legal system etc. brings. It may seem to many Americans that the EU is unified but it's much less so than you'd expect.

Social programs have little to do with it. You can have both industry and social programs if you just choose not to start random wars of aggression in the US too, something for which you need $1.5T apparently.

China has both more industry and plenty of social programs as an example.

[gunsle]

It does have everything to do with it. Why would anyone in Europe be motivated to build anything innovative if they know they will have the government taking 50% of every dollar they make to fund social programs they probably won’t even receive the benefits of? Why would anyone work hard in that environment? You’re completely ignoring the social/productivity implications of high taxation. Additionally, China does not have anywhere near the level of social programs Europe does. No one in China takes two months off a year as a checkout clerk.

[joe_mamba]

EU didn't fail because it never even tried.

Nobody in EU ever tried to build a domestic CPU that was even remotely comparable to what Intel and AMD, for them to have had the opportunity to fail.(No, ARM is IP not a builder)

Nobody in EU ever tried to build a domestic GPU that was even remotely comparable to Nvidia, ATI or 3Dfx for them to have failed.

We just let the US companies fight it out, see who could build the best products, and become forever customers to them and then 20 year later wonder how come their economy grew 2x as much.

[fauigerzigerk]

>EU didn't fail because it never even tried.

Of course the EU tries to compete with the US and in some industries it has done so successfully (e.g pharma or aviation). EU companies have also tried to compete in information technology - with some limited or temporary success.

ARM was founded in the EU. DeepMind was founded in the EU. Olivetti and Siemens had CPU designs and made computers. Nokia once dominated mobile. Ericsson was/is a leading telecom equipment maker. Skype used to be one of the most successful messaging apps. Spotify is a success. ASML is a success.

A certain Finnish university student started the operating system that now dominates the cloud, but for some reason I really do not understand, EU cloud providers like OVHCloud have failed to compete with the likes of AWS, Azure and GCP. And now the AI wave has completely washed over Europe.

The cloud situation is a mystery to me. Nothing stops EU companies like OVHCloud from competing there. Any anticompetitive behaviour is a very weak excuse. Europe can't even compete with something as mediocre as Palantir. Now everyone is calling for protectionism. Ridiculous.

AI is easier to understand. It requires huge capital investments. The US has far superior capital markets and far healthier attitudes towards risk taking. Europe's failure here was easy to predict but the consequences could be dire.

[Matl]

> ARM was founded in the EU

Yes, but it was span off from Acorn Computers, which had its own European OS, CPU, computer etc, but ARM was all that was left of what could have been a fully European Apple had it been given more support.

[joe_mamba]

>Of course the EU tries to compete with the US and in some industries it has done so successfully

I was talking only about SW and HW in the context of the current AI race. Not about legacy industries like aero and pharma. Everyone knows EU is only good at legacy industries, well except cars, they're getting ass whipped there too.

> Olivetti and Siemens had CPU designs and made computers.

None of those domestic designs ever competed commercially with the X86 or amd_64, which is why I made that specific reamark. Please read my comment again and don't try to argue in bad faith by moving the goalposts.

>ARM was founded in the EU. DeepMind was founded in the EU. Nokia once dominated mobile. Ericsson was/is a leading telecom equipment maker. Skype used to be one of the most successful messaging apps. Spotify is a success. ASML is a success.

Most of those have massive US shareholders/financial backing or have outright been brought up by US companies or PE firms or US shareholders. And Spotify being called a succes is laughable given they bootstrapped thanks to initially distributing music they stole off the torrents, and then shafting their smaller artist once they became a multi billion dollar music streaming monopoly.

>Europe can't even compete with something as mediocre as Palantir.

France is copying Palantir to make their domestic version. Chaps Intelligence.

>A certain Finnish university student started the operating system that now dominates the cloud

What does it matter if he's Finnish or not? Linus is an American citizen now living and working from the US, mostly consulting on linux kernel topics for US big-tech. Why did he leave lovely Finland for the "third world" US? That's the question EU gospelers can never answer. He didn't like having walkable cities and free healthcare?

[capisce]

> Nobody in EU ever tried to build a domestic GPU that was even remotely comparable to Nvidia, ATI or 3Dfx for them to have failed.

Bitboys tried, but pivoted to embedded graphics and were bought by ATI: https://en.wikipedia.org/wiki/Glaze3D

Falanx Microsystems tried (not EU but close enough), also pivoted to mobile/embedded graphics and made the #1 shipping family of GPUs in the world: https://developer.arm.com/community/arm-community-blogs/b/mo...

[joe_mamba]

>Bitboys tried

Except Bitboys never shipped any actual GPUs, their products were mostly vaporware, all they had was some IP of unknown performance that ...

> were bought by ATI

My point exactly.

[hsuduebc2]

This is beyond ridiculous.

At the same historical turning point when Europeans are finally waking up to the need to become less dependent on their so called US ally for weapons production and security, they are immediately choosing to become dependent on the next layer of critical infrastructure.

Instead of learning the obvious lesson, Europe seems ready to purchase the future from whoever Washington allows it to purchase from. It used to be the guns, now it's the AI.

It is so idiotic and short sighted that you can barely even blame anyone who keeps exploiting this over and over again. It is always the same story.

[WarmWash]

Europe has been burning furniture to keep the house warm for so long that an entire generation now thinks that chopping wood is for suckers.

[diordiderot]

Chopping wood is for suckers. Half your woodpile will get redistributed to strangers who just moved in down the road.

[hsuduebc2]

Well this is not really a problem as big as it is presented on the social media. Stats simply do not support these claims.

The bigger issue is the growing number of retired woodworkers collecting pensions for much longer than the system was ever designed for. Life expectancy rose dramatically, yet the retirement age not. These retired woodworkers then vote for politicians who keep increasing their benefits, while costs are cut elsewhere. Add a declining birthrate, a trend started by previous generations, and you get a system that is politically almost impossible to reform. It will probably take a very hard societal crash before people seriously reconsider this old, inefficient model. Like in Greece, where pensions were cut on third of it's original size. Sadly, woodworkers never learn other way than this.

With ever present threat of war on the old continent because some woodworking tribes just can't supress their need to chop someone else, this is the biggest long term issue.

[bluecalm]

EU is very lucky USA still has very strict immigration policy. If USA made it easy for highly skilled professionals to emigrate there then there would be almost no one capable left here.

[emptysongglass]

EU countries are in many cases far stricter.

[hsuduebc2]

I honestly think this is not true.

[Szpadel]

big misconception

open source != open weights

open weights model is like... Winamp for example. it's free, you can download it and use it however you like, you could also do some binary patching or dll injections to alter it functionality but it's not enough to develop next version.

the same is with ai models, weights are the binary final artifacts. for development and improvements you need to have training data, pipelines, RL harnesses, etc.

also of you believe Chinese companies will be releasing weights indefinitely, you are not understanding motivations.

Chinese companies spend significant amounts of money to train a model so why they are releasing it for free? they basically provide researchers starting point for developing tooling and optimizations for serving the model in return. and also get some PR. They also do not have to pay for inference of those models that much, as they probably serve them with loss anyways to gain market. they are gov sponsored so money are not issue there, so they try to speedrun their way to what US companies have. And guess what happens when they reach it. they will stop releasing weights and increase pricing or will use them for gov purposes.

[tarsinge]

It doesn’t look like "the EU" has signed anything, just that a couple EU countries independently accepted to join the summit.

[jdkee]

europe2031.ai

[CalRobert]

Was looking for this. It’s a great short story (it starts with real events and then predicts the future) leading to European fealty to US . Seems increasingly plausible.

[pydry]

European fealty to the US has existed for decades.

[CalRobert]

True, but it was, until 2008, a somewhat less lopsided relationship.

Europe is becoming a museum and a retirement home. Young people leave to make their fortune and maybe return to have a family or avail of healthcare. And while I have nothing against old people my own street in a _very_ family-oriented suburb in the Netherlands is over 50% retirees.

Though I see the flow of migrants between the US and Europe has shifted, so there's a net gain of Americans in Europe. I wonder if it's where young people with dreams want to go, or if it's just where people like my in-laws (nice, culturally aware, but no longer very economically productive) who have a few million dollars in their 40's and want a pretty place to retire young, go.

[specproc]

Number one issue of our times. I'm eternally frustrated discussion of this basic fact of our political lives is near zero in public discourse.

[embedding-shape]

Where you live? In both Spain and Sweden it's a fairly common topic in public discourse already, about how to move away from a bunch of US dependencies and so on.

[jurgenburgen]

Also in Finland and even in EU circles. Trump threatening to annex Greenland really woke up a lot of people. Tariffs didn’t help.

Even my parents who speak no English and aren’t spending time in US-based social media have taken a break from hating Russia for the Ukraine invasion to talking about how we need to get rid of the dependency on the US.

This boat is turning slowly but it won’t turn back anymore.

[specproc]

In the 51st state.

[port11]

I think you’ll find that it was mutually beneficial deal that traded soft-power for preferential agreements. As it started to crumble, we’ve begun diversifying and directed our schmoozing to better partners.

[coffeebeqn]

Since the US entered WW2

[aniviacat]

> [Washington] piles pressure on The Hague to halt ASML’s remaining exports and servicing of its DUVi machines.

> The Commission backs the Dutch [...]. The European position fragments before it has properly formed.

That the EU would, after recognizing AI's value, freely give up control of its few advantages to the US, despite of this being a conventional trade issue which the EU has experience with, seems like a very pessimistic assumption.

(I stopped reading after that part.)

[postepowanieadm]

Germany and Netherlands signed alone, not as EU block.