Ask HN: Could Europe build their own AI infrastructure?
I wonder how Europe is set up for the coming AI race around the world.
What would happen if Europe were to be cut off from American supply of GPUs?
As far as I know, Europe has control over one company that is involved in the supply chain to build AI hardware: ASML. They build the lithography equipment which is needed to produce the wafers which are needed to produce the processors.
Are there companies around the world who can/could build this type of lithography equipment?
If Europe loses this piece of control over the AI supply chain, would Europe then be at the mercy of the USA for AI hardware supply? How long would it take for Europe to produce their own GPUs?
22 comments
[ 4.4 ms ] story [ 58.9 ms ] threadWhat would happen if America would be cut off of EU supply of ASML machines?
How hard would it for the rest of the world to build this type of lithography equipment?
As for the chips - TSMC started to also make these in the USA last year.
So the USA have brought this part of the manufacturing process inhouse already.
US companies will want to have consumers and Europe has them, so they would like to offer their services in the EU.
AI hardware is also produced in China. Americans don't want to talk about it, but Huawei has good inference hardware and software.
Look at the German automotive industry. Once they produced something the world wanted. Will they still do that in the age of electric autonomy?
Except for ASML's technology, which competitive advantage does the European industry have to stay relevant in international trade?
On top of that European governments got very good at banning and regulating things, not so much at innovating or creating a good climate for innovators and entrepreneurs.
However, given the current environment, it's impossible for this to happen, because the EU (which is largely equivalent with Europe in this context) is obsessed with regulating everything to death instead of creating opportunities for future prosperity.
They're literally doing the exact opposite of what is needed.
Could any company in the EU make the wafers needed to build GPUs for example?
If you mean Euro balances - that is worth nothing in the context of what Europe can produce. If state currency could be converted to production capabilities, countries could just print money to increase production.
They just decided to use those resources for the wrong purposes. Again, it's as simple as that.
I find it quite baffling that there's even such a thing as an entire conference on a specific law.
> Regulation is intended to be strongest in a quite narrow set of use cases explicitly defined as high-risk.
Intention vs. implementation always seems to be a major issue with EU legislation.
They always claim that these regulations are supposed to target "large companies" or a small number of cases, yet in reality it's usually the small companies and entrepreneurs that suffer while the supposed target audience of large players scoffs at these laws.
Hence, I consider these claims to be outright lies intended to calm people down while the EU continues to collude with incumbents to further entrench their dominant position and smother any kind of change or innovation.
> GDPR will have been more 'disruptive' in practice than the AI Act for the vast majority of developers
Relevant EU representatives at the time claimed there not just wasn't any disruption due to GDPR, but that this regulation would even even spark innovation.
Well, that turned out nicely ...
The EU commissioner responsible literally claimed she could implement all GDPR requirements for a public website herself in a matter of just a few minutes. These people are as arrogant as they are clueless.
The AI Act is a big set of legislation which is interesting and impactful. Why would we not set dedicated time aside to talk about it? I think it would be more baffling if we didn't.
As a developer in the EU having had to consider GDPR in our systems, I support GDPR and believe sometimes disruption is needed to safeguard individuals' interests. Similarly, as an AI developer in the EU I personally support the idea of regulating defined high-risk use cases of AI.
Well, in other parts of the world you have people talking about progress and how to advance the human species while in the EU we have people talking about how to regulate every minute detail of human life. You can't get any more telling than that.
> I support GDPR and believe sometimes disruption is needed to safeguard individuals' interests.
GDPR has completely failed at that. Individuals' interests and privacy are worse off than they were before GDPR.
Small companies are burdened with byzantine rules that achieve nothing to improve users' privacy while the worst offenders - large conglomerates and authorities - simply don't care. The latter are largely exempt from GDPR anyway while the former have huge legal departments to deal with GDPR as they see fit.
This is a kind of strange take. Do you think conference topics are mutually exclusive? I assure you no other conferences on a multitude of interesting subjects were harmed in the making of a one-day AI Act conference.
You're not really providing any concrete examples of GDPR failing, so there's not much there to respond to. I've used the right to erasure and right of access successfully and automatically opt out of trackers when browsing, which for me is a benefit.
To some extent they are, yes.
First, it's about what people spend their time and focus on. If you're focussing on regulating, there's less time you can spend on innovating.
Then, some of that regulation actually precludes innovation in certain areas. So, by establishing the premise that regulation is justified by organising a conference on it, you're delegitimising discussion on the kind of innovation that regulation attempts to prevent.
> You're not really providing any concrete examples of GDPR failing, so there's not much there to respond to.
GDPR was - allegedly - targeted at the likes of Facebook or TikTok, which to my knowledge do very little to implement GDPR in a way that's conducive to user privacy.
At the same time, everyone running a tiny website has to put in enormous effort to implement that regulation, because otherwise authorities might come for them through ruinous fines.
Yes, this does happen, particularly because in some EU countries (Germany being the worst offender) it's so easy to impose fines on small businesses at scale, this literally has become a business model for authorities and law firms.
I am very glad the rights GDPR gives me are applicable to more than just Facebook and TikTok.
It seems like our views on institutional actors' responsibilities to protect individuals' rights and interests are just inherently misaligned - as are our views on subjects worthy of organized discussion/exploration. Thanks for the interesting conversation!