This is why the EU will never be a tech powerhouse, the culture is just not there.
Some EU countries even charge you a copyright tax when you buy a smartphone per gigabyte because they just assume that you'll fill every gigabyte with pirated music.
> Some EU countries even charge you a copyright tax when you buy a smartphone per gigabyte because they just assume that you'll fill every gigabyte with pirated music.
Mind you we have a hungarian private torrent tracker with 400k users and 2.7 petabyte of data running since 2005. Virtually everyone is using it in the country but no one really wants to shut it down because at this point it's also a cultural archive.
> This is why the EU will never be a tech powerhouse, the culture is just not there.
Yes, France is in the EU. However, the actions of France do not equate to the actions of the EU.
Despite what pro-Brexit type people will tell you, there is a lot of sovereignty in local laws within the EU. And the ability to set your own taxes is one of them.
This is quite stupid, but I really wish those who are quick to jump on a ridiculous piece of legislation as supposed proof of Europe's downfall would be as critical discussing consequences of underregulation.
Where exactly do you see under regulation? Everywhere i look i see massive over regulation. If you mean under enforcement of antitruat law so obviously stupid oligopolies are rampant instead of shattered into hundreds of pieces, i guess i could agree with that but thats not regulation, thats destruction of the bad end product of capitalism so that the amazing middle of capitalism can thrive.
I agree with part of your point, but it also depends on what you call a powerhouse. It has its advantages and disadvantages.
It's true that today Europe relies more on the outside for a lot of tech than the opposite, but it's changing and it's not the only measure of quality.
But I especially think your last point about old people is really both false and irrelevant. If I wanted to troll, I could say it's really dumb, just americans being americans.
Btw, the EU is a tech powerhouse, you just don't know that. ASML, Zeiss, TRUMPF, etc. have full monopoly on eUV lithography and majority of the market on any other nodes. Just one example.
Airbus leads with about 55%–60% of the active narrow and wide-body commercial aircraft market. Airbus has much better track record and doesn't collide with FAA
and doesn't "skimp" on development and QA (I'm looking at you MCAS!).
The culture is right there, we are taking our sweet time.
ASML isn't a great example since it's largely American technology invented by the US government department of energy. The EUV light source itself is researched, developed, and manufactured entirely in California.
ASML is an important company but it is mostly just an integrator of technology from many different countries.
Basically, US DoE invented the core tech and formed a consortium called EUV LLC to further develop the technology. ASML came along much later to acquire SVG in order to gain commercial access to the technology and research. The Dutch company "ASML" didn't actually invent most of it. They are just the integrator.
This is how ASML even describes itself:
> Board member Wayne Allan in 2023:
“We're an integrator – we build some of the modules in our machines, but we procure a huge amount of materials and critical components from suppliers, nearly 80% of the components.”
Most of the valuable tech comes from American, German, and Japanese companies.
Yeah, but so are car manufacturers. A car is greater than the sum of its parts. I'd wager an EUV machine is probably a teensy bit more complicated than a car.
You could say the same about lots of software. You didn't write the programming language, same for libraries and so forth...
I wasn't trying to downplay ASML's important role in the commercialization of the machines. I was just responding to the comment "ASML is as Dutch as it gets" which is not remotely true.
> "ASML is as Dutch as it gets" which is not remotely true.
This statement is either ignorant or stupid. You choose which one it is.
ASML has its headquarters in Veldhoven, Netherlands, around 60% of workforce is based around that area. It's a spin-off from Philips, another iconic Dutch electronics company.
Yes, the corporate entity is Dutch. The technology is not Dutch. ASML's own executives report that 80% of the tech does not originate from ASML, or even the Netherlands at all. This is what we're talking about.
In America it comes largely from California, New York, and Connecticut, as well as Germany, Japan, and elsewhere.
I mean, Toyota is as Japanese as it gets even if Toyota did not invent the original technology for the automobile. And a Kei is uniquely Japanese kind of car (even compared to other micros). Same for Sony and audio equipment. So I'm not sure I agree.
If you are trying to say EUV is not Dutch, I don't think anyone is disagreeing with that.
First and foremost, I think you're underestimating the complexity of integrating the most complex technology invented as of date. Just remember High-NA EUV systems require up to 7 freight jets to deliver.
Second of all, EUV isn't everything. There is a lot of "legacy" node demand.
> Most of the valuable tech comes from American, German, and Japanese companies.
"Some" of the American. It's funny how you name it first despite it's being the minority.
> As of 2025, ASML is the only company that produces and sells EUV systems for chip production.
I was responding to your comment about the US Dept of Energy's role in the history of the technology. Also I quoted ASML executives describing themselves exactly as an integrator, not a developer of the technology. But you seem to be upset about this for some reason.
EU/Europe lead in airspace (Airbus), industrial automation/electronics (Siemens, Schneider, ABB), 5G network infra (Erickson, Nokia), earth observation (Copernicus), wind energy technology, medical image technology, agricultural/greenhouse etc.
People here are in a bubble who think tech is only FAANG and node_modules with JavaScript frameworks.
Airbus and Boeing are neck and neck - but both rely on American jet turbines, either directly or through joint ventures.
The US has orders of magnitude more satellites thanks to SpaceX, including American direct-to-satellite networking.
Medical imaging is another example of where the US wins on scale alone. MRI machines are extremely common here, almost twice as many per capita as the OECD average.
Europe has lots of talent but a really horrible regulatory environment. That's why it's not a "powerhouse" and is instead almost always stuck in second place.
Which says nothing about whether they ought to be, or whether our healthcare is better for having them. Needless imaging is a massive cost drain in the industry.
Let's please stop using healthcare examples when we don't understand healthcare.
No, it really is in the sense that a lot of technology is developed here. Where things go awry is when someone tries to use that technology to create something and is met by a sheer wall of regulations, taxes, levies, restrictions and demands. This often results in those people jumping over the pond to continue their efforts in the USA. A good example is the development of the first practical electrically driven regional airliner which started off in Göteborg, Sweden: [1] In May 2024, Heart Aerospace announced the establishment of a new research and development hub in Los Angeles focusing on its hybrid-electric propulsion system, and the Forslunds moved there to oversee operations. The company subsequently closed its Swedish operations in 2025 and moved its headquarters to Los Angeles. In interviews with the Forslunds they quote better access to financing, ease of attracting those with the needed knowledge and experience to design and build these aeroplanes, easier access to the required materials and the general attitude towards startup companies. Mind, they're in California, the place where many American companies are fleeing away from due to over-regulation and excessive taxation. Coming from the EU it still seems to win and that is unfortunate.
I think we can do a lot better in this respect but to get there we'll need to scale down the regulatory mammoth which is the EU.
> they just assume that you'll fill every gigabyte with pirated music.
Yes, but... isn't their assumption mostly correct? Spotify and Co have changed the equation in the sense that one doesn't have files anymore but when those laws came out they targeted MP3 players and DVDs who were, indeed, full of pirated media.
I don't like the result for other reasons, but the assumption seemed always relatively accurate.
That is not the reason for this tax, which is called a private copying levy[1], by the way.
The motivation is even more stupid than that, it is to compensate authors for "lost" income due to buyers making private copies of their media, which is thankfully legal. For example, if I accidentally scratch a CD and it doesn't play anymore, I won't buy a new one if I had made a backup.
Very interesting how eager people are to accept headlines at face value when they support biases, before reading the article or trying to understand what's actually happening.
5 and a half cent per Gb?! Thats pretty wild. I would pay at least 10 euros a month in tax. My friends in EU pay like 50 euros for shitty fiber, will they now need to pay 60? That pretty wild. I am all for light apps and low traffic and perma computing but wow.
Counterpoint: I am behind the gfw.. and seriously dont understand how anyone can get anything done… but it still seems to work… i pay like 2 euros for 500MBit fiber
Is everyone you know able to classify a business they run from their home connection as being a "content provider"? That's crazy, I know one person/entity who'd be considered as such.
Judging by the comments here, every single French person runs a CDN at home? What am I missing, or are people completely misunderstanding what this tax is targeting?
it isn‘t a tax in case of Germany but money which is paid to artists (Gema). Before streaming storing music in media was very common, it made sense to compensate for private sharing.
It has nothing to do with old people. You shouldn’t talk about things you don’t know anything about.
> This is why the EU will never be a tech powerhouse, the culture is just not there.
There are people who consider this a good thing. I don’t have a strong opinion on it, but it’s certainly not obvious to me that this isn’t a good thing.
>but it’s certainly not obvious to me that this isn’t a good thing.
Because its not a problem for Big-corp, but Opensource-Projects/Tor-Nodes, for example i would have to pay ~10$ a day just for my Tor-Relay. It's deeply Anti-Democratic and the opposite what the Internet should be.
It's not even clear yet if this would include self-hosted stuff like if you host a website for your FOSS project or a Tor relay. What news website you got that from, as the linked session notes doesn't clarify that at all.
The objective seems to be to target fournisseurs de contenu (content providers), and the mentioned collection mechanism identifies a entreprise émettrice (sending company). Also, they talk about "turnover" mattering as well, so if you don't charger people, another reason this might not affect you at all. If neither of those apply, like you hosting a website for a FOSS project, it's unlikely you'd have to pay this tax.
But, this is based on reading the session notes themselves, not news article. I'm happy to be corrected if I misunderstood anything.
If you are right, you can bet that YouTube-France will have Ad's every 2 seconds and Netflix gets really expensive, maybe the French Government really wants Piracy ;)
He means the part he's quoted with the ">" symbol.
It's not obvious that housing "tech powerhouses" is a good thing. I mean look at the US, and how much problems all these companies bring to society. I'm also not sure we'd want anything like that in Europe or EU, if we also need to accept the same consequences as the US did.
IMO Europe is facing most of the consequences and getting almost none of the benefits.
American billionaires are self-made (70% became billionaires during their life) where in Europe 70% inherited it. The powerhouse companies of the European economy are stagnant and the families which held them 100 years ago have a great chance of still holding them.
Are tech billionaires obnoxious? Yes, undoubtedly. I'd take them in a heartbeat over having a more permanent aristocracy.
It completely depends what you want from system and how efficient it will be. Want social protections dialed to 11 (like even small government bureaucrats getting golden parachutes worth 1-2 years of their salary while being otherwise practically un-fireable, regardless if they are actually needed... in fact most perm folks are extremely hard to let go). Or you want strong economic prosperity. Cannot have both, the best mix is Switzerland but French will die sooner to the last person before they would admit somebody did state and laws things better than them and that they should maybe get inspired a bit.
What french public conveniently fails to grasp and loses 100% of my support is the naive expectation that in struggling economy they should be rich, while having social system actually dialed to 12 since that 11 is never enough for them. And have 0 doubt, all the protesters now and also before are at the end after money. Money will magically appear in government budget and in all salaries, prices of petrol/diesel will be magically ultra low regardless of yet another failed war being waged by US causing massive rise in prices globally. If one doesn't travel around the world and just lives there it may feel like a normal default system, but given reality of the world out there its anything but.
Lets talk like adults what is achievable or not and manage expectations accordingly. Not very french way, is it. Any attempt to reform even a bit is met with extreme violence, blocking of economy and other self-harm behavior that ruins the economy further and solves 0 of underlying issues, so most cave in and problems fester. I predict it will get much worse in France before there is chance to get things better again, and how much depends on their ultra comfy, some would even say spoiled population.
Yeah, the tech powerhouse seems to refer to the US. There is no doubt that US dominates a lot of tech at the moments, China dominates a lot of manufacturing.
The US is a failing democracy ruled by ruthless liars and capitalists. China's surveillance and lack of freedom is beyond discussion. So EU not being a tech powerhouse still does not look too bad.
Wow it's almost like they're not taxing pirated music at all and are taxing gigabytes. I suppose France is more like a farm of users and they're not entitled to their natural resources because users are inherently slaves to globalized capitalism.
I think it's pretty evident that Artificial Analysis loves to change the benchmark to place US models on the top, it's no the first time either.
The current placement of many models simply doesn't makes sense, GLM 5.3 is so much better than Astra at backend tasks, Astra does the most weird nonsensical bugs.
This is what happens when the benchmark team is funded by VC capital.
The thing that makes the least sense is that Haiku 5.5 is apparently better than every single open weight model in existence except Kimi K3.1. Which is just... not true. I feel like I don't even have to explain it to anyone who has touched both sides of that equation for more than three minutes.
Even and especially for frontend UI, GLM 5.3 is a whole world better than Astra.
I can see (and believe) Astra beating GLM 5.3 on verifiable hard domains like maths, etc; but we’re at this stage where intelligence scores and even ‘frontier real world unsaturated benchmarks’ really doesn’t reflect real world use anymore.
Makes sense, GLM 5.1/5.2/5.3 is a lot better than DS v4.1, I found the same results in other domains.
Given how fast and cheap DS is, it's just an ideal model with enough "IQ" to let it loose. Another thing they left out of the article, DS becomes really good if you provide custom tools for the task, on it's own it's mediocre.
Worth saying that GLM-5.3 isn't GLM-5.3-Flash's "big brother" the way one might think. GLM-5.3-Flash is not GLM-5.3 scaled down. While GLM-5.3 is based on GLM-5.2, and "every gain comes from post-training" (https://z.ai/blog/glm-5.3), GLM-5.3-Flash uses a newly trained multimodal base model (https://z.ai/blog/glm-5.3-flash).
Could be. With a big conference going on and the updates mentioning resource exhaustion it could be a bunch of people doing demos, AI driven or not. Basically slashdotting themselves.
It's not possible to keep shrinking down parameters and keep "frontier" performance, it's like saying it's possible to take a 3 hour movie and compress it down to 3 megabytes, there are information theoretic limits on the amount of bits of information that can be compressed.
What I'm saying is, if you're expecting a model that can be run on a 16GB or 32GB machine with the intelligence/knowledge of Mythos or Sol, it will never happen. It cannot happen, just like you cannot watch the Odyssey saved as a 16MB file.
Smaller models can get faster and smarter, but by definition they can never compress all of the knowledge of a frontier model and they will approach a limit by which they cannot get better.
The current models are not close to approaching the limit of compression for intelligence. They aren’t even focused on it like Chinese labs are. The training of Qwen’s 27B parameter model showed that by structuring model training from fundamentals to more difficult topics they were able to drastically reduce the number of parameters needed.
The ‘frontier’ models rely on scale to achieve their results but that’s not the only approach. Eventually we will hit up against the fundamental limits but we are not close with Sol and Mythos.
Yes they are approaching the limits, try asking smaller models niche questions about almost anything, they hallucinate massively because you cannot simply pack in all the raw knowledge from a massive frontier model into something that’s quantified down to 20GB etc.
It breaks fundamental laws of information theory. It’s like saying you can extract 100 joules of energy from 10 joules of energy source. Not possible.
I think the assumption here that might not hold is simply that increases in efficiency and smaller size will be achieved by linearly just training smaller models better.
You are absolutely right that there is a physical limit about these things, but very often I find that the solution is a clever way to work around the problem. Maybe the problem with knowledge of the models will be improved by them looking the information up in a better way - so smaller models will not have to have the knowledge trained in but will default to checking. Maybe Models will, I dunno, focus on training in assembler and start to only ever check the compiled output so they only ever need to learn assembler and will then compile the solution to reason about the assembler code.
Obviously that last part is a ridiculous example because I'm not gonna be able to come up with a solution myself - I'm not nearly smart enough for that. But I h ope you get what I mean. Not going the direct route but instead finding solutions people didn't think of before.
It doesn't really matter though. Hardware performance is still growing. The new Mac Studio could just about run this model locally (rather slowly) - something that sits on your desk, that you as a consumer can buy.
Imagine prosumer desktop hardware 10 years from now. The 2036 DGX Spark. For a few thousand dollars you will be able to buy something with hundreds of GB (maybe TB if manufacturers step up) of unified RAM, memory bandwidth in the 10-20TB/s range. Overall AI "compute" will increase 10-20x, while at the same time AI model capability per byte will increase 5-10x.
The hardware would fit today's models, something like Kimi K3, quite comfortably and give performance of maybe 100 tokens/second. So what needs data center hardware today will run on your desk.
But if we also assume the models become more efficient, a 2036 Fable-class model (in terms of intelligence/capabilities, not size) will easily run on this thing at hundreds of tokens per second.
Unfortunately it'll still slow to a crawl with 5 Chrome tabs open, and every Electron app will need at least 200GB of RAM.
Sounds like you are describing a quantized model which is a naive form of compression, not a model that is trained more efficiently.
Additionally the information theory angle is for information storage, but a model can access resources and tools to gain information and what we are really seeking to train is reasoning not information retrieval. We reduce the needs to the right capabilities and we don’t get upset if it does not know the lyrics to every song ever written.
You heard of JEPA? LLM's have all sorts of garbage they have memorized. Reasoning in latent space instead of in text significantly reduces the number of needed parameters.
You should look at some literature around it. I don't have time to pull it up now but it's been shown that much smaller small million parameters JEPA model outperforms much bigger LLMS in some applications. Keep in mind JEPA is area of active research.
Let's see.... Unemployment, AI Lunatics, forced to use AI, AI Slopping...
Much of joy we got from coding was taken way, I can not even imagine anyone wanting to get back to a coding role to vibe slop all day long.
I even heard from colleagues that presented algorithms better than the AI slop and heard verbatim from their boss: I prefer the AI solution instead of your.
Eh, this is just the latest rendition. Before that we were still complaining about tech bros ruining tech.
With all this said, the demand for ever increasing profits squeezing everything that exists till it bleeds out will eventually ruin anything so even without AI tech would have reached that point.
I don't buy it, this is essentially vibe-consulting.
Deal with one customer can be quite a nightmare, and honestly, sometimes using off the shelf solutions can result in saving time and money.
Even today I'm creating SaaS, not every problem is simple to solve. Take a look at CRMs, there are thousands of CRMs due to enormous egos wanting to build "my way", "the way", "the superior way", "our company is not like others (LOL)", etc. At the end of the day it's just a database with CRUDish UI.
My customers pay me to solve their problem well because they can not solve it themselves, when they do they just don't have the time.
Reality check: I have been coding many products with all frontiers models: Still takes a lot of time and, surprise surprise!, money for tokens. Ironically software development became more expensive for serious software that won't vibe-break/vibe-delete-prod/vibe-delete-database.