The Nvidia GPU Black Market

[The Rogue Wolf]

​

Gamers Nexus did an absolutely fantastic in-depth report on how smuggling is getting China all the "banned" CPUs and GPUs it wants and more, and how Nvidia is not only turning a blind eye to it, but blatantly gaslighting the public by saying it isn't happening. It's a 3 1/2 hour long video, but it's really worth watching.

 

[Gordon_4]

I don’t know why nVidia would bother denying smuggling of their product occurs. They’re not unique in that respect and pretending they are is fucking stupid.

 

[Phoenixmgs]

The bans are stupid so it's good that they are essentially ignoring them.

 

[Agema]

Gamers Nexus did an absolutely fantastic in-depth report on how smuggling is getting China all the "banned" CPUs and GPUs it wants and more, and how Nvidia is not only turning a blind eye to it, but blatantly gaslighting the public by saying it isn't happening. It's a 3 1/2 hour long video, but it's really worth watching.

I've also found out there's a Pope in the Vatican.

nVidia is there to make money, it will do the bare minimum to comply with any restraint on its business activities. When nVidia says that China isn't getting chips, what's going on is that nVidia has minimum legally required actions to prevent chips getting to China, which it will carry out and monitor. So whilst by the standard of absolute reality China is getting nVidia chips, by the standard of nVidia's legal obligations, China isn't getting chips. These sorts of reasons are why sanctions often are much weaker than we'd like. They're just too easy to circumvent.

The good news for anyone who supports the ban is that bypassing normal trade channels costs money - middlemen, especially if criminal, need paying - so at least China will probably spend more for its chips than normal retail price.

 

[The Rogue Wolf]

Bloomberg issued a (likely spurious, potentially illegal) DMCA strike on the video. Steve isn't taking it lying down.

​

 

[BrawlMan]

Bloomberg issued a (likely spurious, potentially illegal) DMCA strike on the video. Steve isn't taking it lying down.

​

Bloomberg are a bunch of cowardly bitches. What are they even trying to accomplish? Other than silencing people and circumventing to China. Funny how they like to be in bed with people like PayPal and Nividia.

 

[Gergar12]

China has the population and the resources(ex, rare earths) to build its own complete supply chain of raw rare earths to 5nm or below semiconductors to the GPUs because its government is trying to invest in it via subsidies, non-tariff barriers, regular specific tariffs, and forced company partnerships. Whether or not we give them 90 series graphics cards and above is a non-factor. If we don't give it to them, they speed up their research and development via more subsidies. If we don't, they win too, because their AI research will speed up, where they have more papers, and even if 1/10 of those papers are credible, they will still have more.

The only way the US gets out of this is to have some sort of two-tier system for its education. One tier for associates, trades, and non-college-skilled blue-collar work, and the other for college degrees from bachelors to post-docs. I don't know how this will work with regards to the ages, telling an 18-year-old to pick may be too early, but by the time they're 21, when their brain is developed, it may be too late due to the fact that their schooling systems would already have kids in college...

 

[Agema]

If we don't, they win too, because their AI research will speed up, where they have more papers, and even if 1/10 of those papers are credible, they will still have more.

That doesn't necessarily mean anything.

Papers are bits of academic knowledge output, but the majority of research is done in companies and companies don't tend to produce academic papers. After all, once you publish an academic paper, you've just told all your opponents what you know: it's hardly where you'll find much that's commercially sensitive.

Which is to say that China publishing lots of papers might make it a world leader in producing a small proportion of knowledge on AI, and it's the least economically exploitable.

I don't know how this will work with regards to the ages, telling an 18-year-old to pick may be too early, but by the time they're 21, when their brain is developed, it may be too late due to the fact that their schooling systems would already have kids in college...

That's not how it works. 30 years hard manual labour down a coal mine is undoubtedly poor preparation for a new career direction in economics, but in practice lots of people can readily change career. If a government wants to support adult education to help its population retrain in mid-life, it can do so - and it's not that expensive.

 

[Gergar12]

That doesn't necessarily mean anything.

Papers are bits of academic knowledge output, but the majority of research is done in companies and companies don't tend to produce academic papers. After all, once you publish an academic paper, you've just told all your opponents what you know: it's hardly where you'll find much that's commercially sensitive.

Which is to say that China publishing lots of papers might make it a world leader in producing a small proportion of knowledge on AI, and it's the least economically exploitable.



That's not how it works. 30 years hard manual labour down a coal mine is undoubtedly poor preparation for a new career direction in economics, but in practice lots of people can readily change career. If a government wants to support adult education to help its population retrain in mid-life, it can do so - and it's not that expensive.

Lots - does not mean most. Supporting adult education has a poorer ROI than just getting it right for most people the first time. Are there people who are late bloomers or early bloomers who regret having a plan, sure, but the person who has coded/programmed since age 13 is likely better than me, who has been on and off since age 22-24ish, and I am likely going to be better than most people who attempted it in their 40s-50s.

Plus, in college, you often need to take advanced math classes like linear algebra and Calc-3/Multivariable Calc for some of the harder majors, so this requires you to have a good foundation in math in high school, and possibly middle school for most people. If your school district or education was basically less than high school calculus, your odds are lower, not to 0%, but they are lower.

Not to say the German system is great either, but the American system of doing this is leading to people dropping out, not getting their jobs for their majors, and not attending at all. They don't call it the lucky 20% of people who can attend college, finish, and get a job that matches their major for no reason. Basically, if I wanted to be a Data Analyst for say a large S&P100 company, I would need to, at a bare minimum, beat 80% of people, and for the type of jobs I want, blue collar analyst jobs, it's likely way higher. I did that for a while on and off, but there are people with bachelor's degrees still working at Starbucks, which isn't even sustainable since Starbucks could make way for automated coffee makers that are just as good, which may be a thing soon.

 

[Agema]

Lots - does not mean most. Supporting adult education has a poorer ROI than just getting it right for most people the first time.

Do you have some special insight to know what the job market of 20 years into the future will need? No? Me neither, nor anyone else. Talking about "getting it right... the first time" is pointless. Whole industries can grow or die in less time than a career length.

I'd agree that it's rare that people will change career in later life (e.g. 40+) into certain highly specialised jobs very different to their prior experience. But these sorts of jobs are a small proportion of the economy. In practice, loads of people are moving fields throughout their career.

 

[Gergar12]

Do you have some special insight to know what the job market of 20 years into the future will need? No? Me neither, nor anyone else. Talking about "getting it right... the first time" is pointless. Whole industries can grow or die in less time than a career length.

I'd agree that it's rare that people will change career in later life (e.g. 40+) into certain highly specialised jobs very different to their prior experience. But these sorts of jobs are a small proportion of the economy. In practice, loads of people are moving fields throughout their career.

Yes, I do know of an industry that won't die anytime soon, taking care of old people. Sadly, many people don't want to do it, and I don't blame them, but the Bureau of Labor Statistics does have jobs available that will still have growth in the future, and some industries like Logistics and Healthcare will still do well even if the world is a nuclear apocalypse.

 

[Satinavian]

Not to say the German system is great either

It might have issues, but i generally like it.

Not only for the free university education but also for the properly regulated vocational career path. If you hire someone to fix something in your house, you are basically guaranteed to get someone knowing what they are doing.

That does make it difficult for foreign blue collar workers without the proper credentials though. Foreign white collar workers can just use their academic degrees and be fine. But an electrician ? Back to trade school.

 

[Agema]

Yes, I do know of an industry that won't die anytime soon, taking care of old people. Sadly, many people don't want to do it,

That's what immigration is for.

Just so long as everyone who wants to heavily restrict immigration understands they'll be dedicating hours of their spare time to look after their elderly relatives because there's no-one else to do it. Or, you know, let 'em die. Get to inherit faster that way, too.

Bureau of Labor Statistics does have jobs available that will still have growth in the future, and some industries like Logistics and Healthcare will still do well even if the world is a nuclear apocalypse.

Yes, there will certainly be types of jobs that will survive in substantial quantities to some degree. Healthcare is one. I suspect education will remain relatively robust as a source of employment. Middle administrative / management (generally) will be another. Education and administration may require high skills, but they are not necessarily highly specialised: some of those skills are perhaps general/transferrable that they will have a grasp on from their prior career.

 

[Satinavian]

Not sure.

If white collar work opportunities really take a big hit and the numbers of frustrated jobless academics rise, the result might also be a downscaling of education. I mean, what is education good for in capitalism if it doesn't fit the industry needs ?

 

[Agema]

Not sure.

If white collar work opportunities really take a big hit and the numbers of frustrated jobless academics rise, the result might also be a downscaling of education. I mean, what is education good for in capitalism if it doesn't fit the industry needs ?

I was largely thinking primary and secondary education rather than higher education. Kids will presumably still need basic education, and I suspect that's going to be relatively hard to fob off onto AI.

But you do make a good point: if people don't need educations, education itself will go into decline as society will not see the point in it. (During the industrial revolution, literacy and numeracy in the UK declined, because jobs on factory production lines doing simple, repetitive tasks didn't require them.)

 

[BrawlMan]

​

 

[Gergar12]

​

Fuck this I had to pay 400 dollars for a T7 Shield SSD that's only 2TB. Light the entire GenAI market on fire if you need to.

Edit: SSDs are also impacted.