September 3, 2025 • 37 mins
In this episode, Ed Zitron is joined by Steve Burke of GamersNexus to talk about the black market for AI GPUs in China - and how Bloomberg suspiciously forced YouTube to take his video down.
Premium Newsletter: AI Bubble 2027 - https://www.wheresyoured.at/ai-bubble-2027
GamersNexus:
Our Channel Could Be Deleted (GamersNexus) https://youtu.be/tUnRWh4xOCY?si=AD2hyE6z5-ZiyInW
Mirror of NVIDIA AI GPU Black Market Story - https://www.youtube.com/watch?v=ZyWelvEP_CQ
https://www.youtube.com/gamersnexus
https://gamersnexus.net/
https://bsky.app/profile/gamersnexus.bsky.social
https://x.com/GamersNexus
YOU CAN NOW BUY BETTER OFFLINE MERCH! Go to https://cottonbureau.com/people/better-offline and use code FREE99 for free shipping on orders of $99 or more.
BUY A LIMITED EDITION BETTER OFFLINE CHALLENGE COIN! https://cottonbureau.com/p/XSH74N/challenge-coin/better-offline-challenge-coin#/29269226/gold-metal-1.75in
---
LINKS: https://www.tinyurl.com/betterofflinelinks
Newsletter: https://www.wheresyoured.at/
Reddit: https://www.reddit.com/r/BetterOffline/
Discord: chat.wheresyoured.at
Ed's Socials:
https://www.instagram.com/edzitron
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:02):
Also media, Hello and welcome to Better Offline.
Speaker 2 (00:06):
I'm your host ed ZiT Tron. And of course you
can go into the notes for the episode. You can
take a look at links from a newsletter. You can
go and buy some merch as always, but much more importantly,
(00:28):
today I'm joined by the esteem Steve Burke of Gamers Nexus. Steve,
thank you for joining.
Speaker 3 (00:31):
Us, Thanks for having me.
Speaker 2 (00:33):
So you did an incredible video on the black market
smuggling of in videos AI GPUs into China, and then
there has been a thing that's happened where the video
has been pulled down due to a DMCA complaint from Bloomberg.
What is happening?
Speaker 3 (00:50):
Yeah, it's so the video, it's really cool video where
we went all over China finding these GPUs. As part
of that, we included a recap of all the news
of export controls or the US controls GPUs that are
above a certain performance threshold and the allowance to export
(01:12):
those to China by American companies, and so that was
the story we wanted to look into. Okay, but they're
still getting to China and so how is that? And
as part of that recap, go through all the news
we've got clips of you know, various politicians. There was
a clip of Obama in there, and there was a
clip of Trump in there, a couple of them talking
about AI or GPU export control bands. Of course AI,
(01:35):
I know ed is one of your favorite topics. Yes, yes,
me as well. And yeah, so we received a copyright
strike from small, small publication called the Bloomberg. Yeah. So
(01:56):
that strike took the video offline and now we are
currently challenging it. And so the strike was for specifically,
I believe it was minutes twenty two until minute twenty
three point fifteen in the video.
Speaker 2 (02:10):
Of Donald Trump speaking the President of America. Correct, Yes,
that is ludicrous. And you went to New York specifically
to have it out with them, did they Where do
we stand as we speak, because this is going out
in about a week from now. It's August twenty eighth,
so who knows what twenty sixth even, I don't know
what happened next.
Speaker 3 (02:28):
Yeah, if it goes out in about a week, then
that'll be pretty close to when we should have an
update from YouTube.
Speaker 2 (02:35):
Because the video is off right now.
Speaker 3 (02:37):
Correct, the video's gone. And it's very interesting how copyright
strikes work because if basically the way it works is
a company files a strike. They decide if there's infringement
of some kind, and they file it. YouTube then basically,
especially for comedy, Bloomberg size will automatically approve it. It's then
(03:00):
up to the creator to dispute that the videos offline.
During this entire process, any money from the video is
held at least in escrow, and so then you can
dispute it if you want. At that stage, YouTube then
decides if it's going to accept your dispute. It accepted.
Ours is written by a lawyer, so that makes sense,
and then they step back and they're like, Okay, we're
(03:22):
not involved anymore. It's between you guys. And so the
party that filed the claimant, so Bloomberg here, has ten
business days to file proof of lawsuits. So they if
they submit to YouTube proof that they are suing us,
which thus far as we record this, they have not,
(03:42):
then at that stage YouTube would keep the video offline
and they kind of let the two parties resolve that
amongst themselves. If they don't submit any proof of a
lawsuit or they don't file a lawsuit and then show
YouTube that they've done so, again, they haven't at this point.
Then basically in ten business days the video goes back up.
We get fucked over for the ten days in between,
(04:03):
and then the money, in theory should be released back
to us, the ad revenue that was made from it,
if it follows that course. But the biggest damage that's
done is from the loss of momentum in those you know,
total time period maybe thirteen days or so. But why would.
Speaker 2 (04:21):
Bloomberg do this? Are they really that attached to videos
of Donald Trump? Or is it something else?
Speaker 3 (04:27):
I mean, I yeah, we talked about this in our
video where we flew out to try and speak with them.
I have a number of theories. We don't know their
motive precisely, but you know, there's a few strange things
going on where first of all, I've noticed that they
(04:48):
have not claimed any or struck any that I'm aware
of the re uploads of the video. So that's interesting
because there's hundreds of them. It was just hours that
want top initially. Secondly, we have used that clip before
and that hasn't been stricken and so specifically in this
(05:08):
black market video. And yeah, the theories we came up
with were there's some competing content from Bloomberg where they
attempted to do what we did. It was really a
failed attempt. They tried to find smuggled ai GPUs in
China as far as they kind of presented it in
data centers, they were unable to do so they couldn't
(05:30):
get any access. They filmed sand in the desert. They
go home. There's also that's all they go. Yeah, they
drive around at a desert, they point the camera at
some buildings that are being constructed. They say, you know,
there's ai GPUs in them their hills, and then that's
about it.
Speaker 2 (05:51):
It's kind of funny that you you appear to on
your own and with your with your team obviously able
to get this access to these large legacy media people don't,
which it's meant to be the other way around based
on what everyone says.
Speaker 4 (06:05):
So what is it the makes your approach.
Speaker 2 (06:08):
Different to bloombugs, Like why did you get stuff that
they didn't?
Speaker 3 (06:13):
I think rather than just the desert, right, Yeah, Well,
to be fair, we did not film the desert, so
they did have an exclusive there. But I think it
comes down to the approach where we looked at it
like this is an interesting concept, this idea of black
market GPUs, But from my point of view, we approached
(06:36):
it as well, let's just see what are the people
involved in this chain think and do they think it's
a black market or do they just think it's a market.
And a lot of the people we worked with were just,
I think, amused at the concept of being involved in
the media process and wanted to see how it works.
(06:57):
And so I honestly, I think a lot of that access,
especially to the people who are just sort of like
normal people. They're not these big, you know, executives at
some huge corporation. They're they're middleman and they're people transacting
GPUs on the ground. I think for them, if you
go into it without this free supposition of some kind
(07:19):
of particular i don't know, leaning or ethics of whether
what they're doing is or isn't okay or whatever, or.
Speaker 2 (07:29):
Even expecting what you'll find exactly.
Speaker 3 (07:32):
Yeah, I think if you go in with an open
mind of like we're just gonna hang out with this
guy for a day and just observe his job, people
are you know, they're pretty happy to share that.
Speaker 2 (07:42):
So, yeah, how did you actually start those? So you
you came up with the idea that you wanted to
see how elicit GPUs making it into China. Did you
have contacts in China or Hong Kong and so on
that you talk to in advance and they kind of say, oh,
if you come here, we can show you this. What
was the process?
Speaker 3 (08:02):
Yeah, the process was really cool. We're still kind of
learning how to iterate on this, but each time we're
The big thing we learned this time for this investigation
was really just trying to that first contact matters a lot,
and being able to work with that first person we
(08:22):
find to then find their contacts and kind of follow
the chain down the line. And so the first guy
we found is a professor at the Chinese University of
Hong Kong. He buys and uses these high end GPUs
in servers for educational purposes and research purposes. And as
soon as we made contact with him, we were able
(08:46):
to work with someone who first of all speaks you know,
native level excellent English, so that certainly helped. And then
secondly who has a lot of local knowledge and contacts.
And I think a lot of it is just kind
of knowing which, in our case, which link in the
chain you need next to make the story makes sense.
(09:09):
So in his case, he's a user of these GPUs.
So next we need is some kind of supplier, you know,
and then we need their supplier.
Speaker 2 (09:17):
Yeah, so what it's all quite complex. I did watch
the video and it was it was what three three
hours or so or more even really wonderful. And so
what exactly are the GPUs that are restricted? Was it
just aigpus? Was it like consumer level ones as well?
What is exactly being handed around?
Speaker 3 (09:39):
It's both. So on the high end side, there's some
smuggling and presence of these data center class GPUs, so
that'd be things like EH one hundreds, Hopper one hundreds,
A one hundreds. We actually physically saw several of those
(10:00):
and things like that. On the consumer side, there aren't
as many restricted consumer GPUs, but there are a few.
So the RTX fifty ninety is one of them that
has thirty two gigabytes of video memories. That's very valuable
for these use cases. The ARTAX forty ninety actually is
another one, and that one's kind of amusing because it's
(10:21):
so GPU right, yes, consumer GPU, yeah, and last gen
consumer GPU at that, but it's easily modified to carry
double the video memory as the actual official end video
spec and the skew so they can increase it from
(10:42):
twenty four to forty eight gigabytes of memory, which is
extremely valuable for these training and LLM type tasks where
it may be a last gen So it's technically a
little bit slower, but if you can fit it in memory,
you can still run it, versus if you can't fit
it in memory of the model you made, not be
able to run it at all. And so those are
(11:02):
the consumer cards the forty ninety. That was interesting because
one of the guys we spoke to who is effectively
a buyer and seller of these band GPUs, he was
asking me on camera. He was seeking confirmation of well,
but are you sure the fifty nineties banned? Yeah? And
(11:23):
I said, yeah, I'm sure. Here it is on the list.
And he was so confused because he said, but they're
just all over the market next door. We walk over
there and there's dozens of them.
Speaker 2 (11:37):
So it's so strange as well, because it doesn't seem
like you basically walked off the plane and had a
price sheet from just a guy, Like it wasn't clear
(11:58):
how you got the price sheet exactly. Perhaps you can't say,
but it was just Yeah, these things a band other
than the fact we have just like guys who sell
them literally everywhere. Yeah, getting in.
Speaker 3 (12:10):
So there's a couple of ways. The one of the
ones that we showed in the video is really interesting.
To the very end of the process. We actually this
is one of the reasons we delayed initial publication. We
were able to, with help of a viewer, locate someone
you could actually classify as a smuggler in the US.
So someone from China in the US.
Speaker 5 (12:34):
And.
Speaker 3 (12:37):
He buys GPUs from people in the US. He drives
around all over and buys most through forty nineties, which
are in a high demand in China, and then he'll
strip those downs that could be removing the cooler. I
don't think this particular guy disolders them. I've heard stories
of some people desoldering GPUs and mailing them that would
(12:59):
reduce your shipping costs a lot, increase your margin. But
he buys the forty nineties ships the boards back his
profits about three hundred dollars US per board before taxes.
It wasn't a one hundred percent clear if he does or
doesn't pay taxes. And then at that point, once they
are either shipped into Hong Kong or Macau directly if
(13:23):
they feel gutsy enough, or indirectly through a third country,
a middle country where there is no export control, you know,
then they arrive and they get redistributed from there. There's
sometimes hand carried as well, like by students, international students
who just fly back with a fifty ninety. They bought
it best Buy or something, and they can double their
money roughly if they're lucky.
Speaker 2 (13:44):
That's so, it's wacky that there is such a big
black market, but I guess the export bends just created it,
as did this whole AI, this crazed AI moment. Did
in fact, while you were over over in Hong Kong,
did you actually go to China or can't remember? It
was just in home? So did you? This is an
anecdotal thing. How was the pressure of AI there? How
(14:06):
big was the conversation the advertisement? Did you see it
on the same level? It's fine if you didn't, I'm
just curious.
Speaker 3 (14:12):
I saw it more in Taiwan, which we went to
as well. So yeah, after the China trip, we went
over to Taiwan, and I don't think it made it
into the cut, but we shot a short clip of
just a bus, like a public bus with a gigantic
AI the laptop advertisement on it. So I would say
it's kind of at least there, it's similar to what
(14:34):
you see here where it's you know, if you're anywhere
in the vicinity of a best Buy or a city
that has technology dealers, then you're gonna see AI ads
In China, Yeah, I saw it in the tech markets
for sure, some banners, but it just it. I don't know.
That's I only understand Chinese if I'm actively trying to
(14:58):
understand it, you know. So it's I think a lot
like it's filtered out for me, right, Yeah, I was good.
Speaker 2 (15:03):
Yeah, I assume you speak some You seem to speak
some Chinese. I can't speak anything but English, so I
could not correct your Chinese, doll.
Speaker 3 (15:10):
You do speak more proper English than I do, though,
so I'll.
Speaker 2 (15:13):
Get Yeah, yeah, I speak all kinds of it. These days.
It's not going well. People criticize my accent all the time.
So as far as how the government sets these limits,
one thing he did really eloquently towards the beginning of
the video as well, was he kind of went through
how arbitrary the limits seemed to be and how in
Video seems very capable of adapting around them to the
(15:34):
point they almost doesn't make sense.
Speaker 3 (15:37):
Yeah, yeah, they they've modified the regulations a few times.
And it's it's so this particular part I don't really
blame and video for for this, which is just trying
to you know, the government sets some regulations and a
(15:57):
company says, okay, we'll comply with these, and then they
comply with them, and then the government kind of says
no weight not like that.
Speaker 2 (16:04):
Yeah, that's very standard for pretty much or regulation. It
maybe it's good or bad what have you, but it's not.
It happens all the time with companies.
Speaker 3 (16:14):
Yes, And in this case, one of the things that
as the I guess AI world developed and the applications,
one of the things that changed was this understanding that
memory capacity is basically the most important aspect of it.
And the initial versions of the formula didn't consider memory
(16:36):
in any scenario. What I remember, it was based on
the marketed flops of performance on the spec sheet multiplied
against the bit length of the operation or something like that,
and they later added memory bandwidth. Memory capacity though is
still as far as I'm aware right now, is still
(16:58):
not considered in the formula.
Speaker 2 (17:01):
Yeah, it just it feels just arbitrary. But I imagine
also it's quite difficult to find a way to do this,
especially when the people setting the regulation might not have
the best handle on computers.
Speaker 3 (17:13):
Yeah, I think I think the best way to do
this would probably be if they wanted to and I'm
not I don't have a particular one way or the other,
but if they wanted to control it, I think the
best way is probably some kind of benchmark or set
of benchmarks, just like you would do for a review,
where you know, you say, okay, if it exceeds some
performance threshold in this real world use case or multiple
(17:36):
of them in aggregate, then it's control, it's export controlled.
But instead they just kind of calculate based on the
spec sheet and the companies that make these things. Obviously,
you know, full appropriate credit to Nvidia. The company is
extremely confident with making GPUs. They know what they're doing,
and they're going to know a lot more about how
(17:58):
to tweak those dials to comply then a government agency
will so.
Speaker 2 (18:05):
So on the subject of Nvidia knowing things, do you
think that do they seem aware of any of this,
like any of any of this smuggling happening. Do they
seem to Did you get any comment out of them?
Speaker 3 (18:17):
We asked a lot of the people in the chain
what their perception was of Nvidia's awareness, and so the
professor at the Chinese University of Hong Kong, he had
a great answer when I asked him, do you think
Nvidia knows? And he kind of he paused for a second,
and then he said, how could they not? You know,
(18:38):
like these things are very expensive. You've got thirty plus
thousand dollars GPUs and it just seems like they would
all be tracked. So what was the term?
Speaker 2 (18:48):
There was a specific terms like kind of like basically
equivalent of ignoring him.
Speaker 3 (18:55):
Yeah, so the Chinese has done eg and b egn,
which means open one eye, close one eye, or turn
a blind eye? Is the idiot?
Speaker 2 (19:04):
Yeah, And that's awesome.
Speaker 3 (19:08):
It's just very cool. Yeah, it's super I mean the
coolest part about that was multiple people said that exact
idiom and it's it's just it's such a specific choice
of words. It'd be the same as if you asked
like three people a question, you know, they all said, oh,
(19:28):
they turn a blind eye. At some point you're like, Okay,
this seems to be a really consistent belief here, But yeah,
I think from a video standpoint, it's whether or not
they want to actually control what gets in. It's hard
to say. Certainly, from a pure financial standpoint. The more
(19:49):
gps they sell, the more money they make that they
probably don't really care who they sell them to as
long as they sell them to someone and then where
they end up after that. I mean, maybe to the
extent that video doesn't want to piss off the United
States government. They you know, they try to comply as
much as they reasonably can. But how much do they
actually try to control ingress into a country that's export controlled.
(20:13):
It's hard to say, But certainly I think they are
aware of this happening, and I just I have a
really hard time believing that they are unaware their gps
are making it into China when they shouldn't be.
Speaker 2 (20:26):
And I imagine it must be quite hard to stop.
Like I hate to have any sympathy for in video,
but it does. It does feel like something that would
be kind of a moving target. Because one thing that
was consistent across the whole video was the ingenuity of
the people involved. It does make you see like American
entrepreneurs sometimes and they're like, oh, yeah, it's tough to
(20:48):
you know that the hours are very tough, and you know,
I was covering all not and all of these guys
were just like, yeah, I've got a thing in the
back of my car full of wires. Yeah I've got
I have a whole rig.
Speaker 3 (21:01):
Yeah it's crazy, and that is I mean again, you know,
from that particular side of things, I don't suspect. I mean, look,
it's a government regulation, and if the government wants to
regulate it, it's kind of up to them. Obviously the
companies need to comply with it, but I wouldn't expect
(21:22):
a company, even Nvidia, to send in any kind of enforcement.
Maybe what they would do is if they wanted to
enforce it, you know, they might identify, Okay, this purchaser,
there's evidence of this purchaser exporting to China anyway, and
so we're just gonna stop selling to them or something.
But that's kind of I think the most you might
(21:44):
expect of them. But yeah, as far as the ingenuity,
it was pretty crazy. Like the guy in the US
who buys the cards, he he just got a Prius
with a massive like battery bank in the back, like
a ups with a lithium ion battery in it. He's
got a test bench hooked up to it, an ATX
test bench. Pops the trunk open. He's got a spare
(22:05):
license plate in the trunk for some reason. I don't
know why. Uh, that's not to ask. Yeah, I don't
need to know. And uh yeah. And if you want
to sell him your card, you're just some American new
found him on Facebook Marketplace. He at the time we
were talking with him, he was paying two thousand dollars flat.
He makes three hundred when he sells it, and you
(22:27):
bring them the card, meet him wherever he tests it,
and in the back of the you know, in the
trunk of the car, he just runs a benchmark, make
sure it's good, and then buys it and ships it
back or carries it back, depending on how secure he feels.
And then also to your point of ingenuity, the shops
(22:47):
that are capable of modifying cards, where they're just they're
repair shops, and they just happen to be really good
at board level BGA or ballgrid array device repair or
or swapping, so they can pull memory modules. They can
add memory modules, swap the GPU between PCBs, and they're
(23:08):
able to keep Silicon in service where you know, they're
just repair shops, so it's kind of like from their perspective, look,
they're not doing anything illegal in China, they're just keeping
boards in service. But for people who need these high
capacity GPUs for AI tasks and they having to be
export controlled, a shop like this would be capable of
taking a card that has died, you know, through through
(23:31):
use or just needs to be upfit to be better,
and they can do that for pretty cheap and basically
in a couple hours.
Speaker 2 (23:40):
And it's not illegal to sell them in China once
they are there, right, They've like that is not a
like China is not prosecuting people for any of this.
Speaker 3 (23:50):
Correct that it is not illegal to own or sell
these GPUs or buy them in China. It's legal if
you are an American company, if you're Nvidia or your
best buyer or a partner or whatever. It would be
illegal to sell the export controlled cards to a company
(24:12):
or an entity or person in China. But once it's there, yeah,
there's no real control over it. The only the only
place that we found that China steps in is if
there's some kind of undeclared ingress through port. So like,
for example, there were CPUs and GPUs at various points
(24:34):
in the last several years that were smuggled in with
creates of live lobsters or prosthetic baby bombs.
Speaker 4 (24:40):
But was that real?
Speaker 2 (24:41):
Was the lobes the thing real?
Speaker 3 (24:43):
The lobster things real? Despite what Nvidia said.
Speaker 4 (24:59):
Hell yeah, I love the So what actually describe the full.
Speaker 2 (25:03):
Loves the situation? I think everyone wants will want to
hear it over lobs.
Speaker 3 (25:08):
Yes, So the loster situation as it were is basically
there was a Hong Kong uh port, so customs intercepted
a white van I think it was. There's like a
customs post they post photos of the things that they
(25:28):
are able to snipe being brought in improperly. So Hong
Kong's a free trade port, but there's still certain declarations
you have to make and sometimes people won't make those.
And so there was a van bringing in crates of
undeclared live lobsters driving on I think it was the
(25:49):
Macau or the jew Hai Bridge or something. But anyway,
they got intercepted and alongside the live lobsters were GPUs
and they were older, so I think they were quadros,
but they're brought it all being brought in improperly. And yeah,
so customs post the thing. They're like, we found this,
you know, and I think that's just kind of like
(26:09):
the here's us doing our job and maybe scared people
off or whatever. And then later AI startup Anthropic released
a statement talking about how there needed to be more
strictly enforced export control in the US, and they mentioned
(26:31):
smuggling of GPUs with lobsters, and Vidia responded and I
don't have their statement in front of me right now,
but it was something along the lines of it would
be better if American you know, AI companies would focus
on innovation rather than the part I do remember was
tell tall tales of prosthetic baby bump smuggling or GPU's
(26:56):
sensitive electronics being smuggled with live lobsters. Turns out that
is not a tall tale. That is a thing that happened.
Speaker 2 (27:02):
So, yes, this is the thing people say. Jensen Wong
is cool. If you're cool, you say, yeah, well, our
chips like, we don't like this happening. This is bad,
but doing everything we can't stop it. But yeah, they're
doing James bondshit to get our GPUs, Like, come on,
you could just say that, like they do seem a
little sensitive, and we're we're recording this the day before
(27:24):
their earnings, and I got no idea what's going to happen,
No one does. Everyone's a bit everyone's a bit worried
about it. So it's I think it. The China market
is going to get extremely interesting because they've got the
H twenty band. But now the H twenty band has
been lifted, and now we're in this weird spot where
I don't know if you've heard anything about this yourself,
(27:45):
where there are suggestions that the Chinese government is actively
saying not to buy in video GPUs. If you heard
anything about that.
Speaker 3 (27:52):
Yes, So the Chinese government, I believe it's called the
Cyberspace be a security agency or something like that, but
one of their government agencies put out a statement following
separate statement from a state owned or at least state
(28:12):
you know, controlled in some capacity media report where the
statement said or at least floated the possibility that Nvidia's
H twenties may have some form of spyware in them.
And so the suggestions we're tracking or government back doors,
and so that was kind of the accusation, and Vidia
(28:35):
of course posted a statement saying absolutely not, this is
not a thing, and you know, just to be fair here,
there's not currently, as far as we're aware of, any
firm evidence of this. But I'm also sure that governments
know things about you know, about products from other nations
that we would never know anyway, So who knows. But
(28:55):
that's where it stood. Nvidia said, we don't have these.
China's agency said to companies in China that they should
be wary of using H twenties and advised against purchasing
them for fear of some kind of backdoor or spyware.
Speaker 2 (29:11):
So how realistic would it be for Actually, did you
hear any scuttle about around this of switching to Huawei
chips or something?
Speaker 3 (29:19):
We did ask? Yeah, I think we might have only
kept one of those questions in the actual video, but
I did ask a lot of people Huawei from the users' perspectives,
the people we spoke to, they none of the people
we spoke to are using Huawei components right now, so
(29:41):
they're getting more powerful from what I understand, I think
there's a few hurdles. One of them is. It's kind
of the same reason where the companies aren't even really
using AMD and they've been doing this for a while,
whereas Huawei is pretty new. And a lot of that
comes down to Kuda, where Nvidia is the proprietor of Kuda,
(30:03):
which is a library they can use to accelerate tasks
in which a lot of software is written to use,
I mean even just rendering a video.
Speaker 2 (30:12):
It's the coding language to get accelerate computing, right, correct, Yeah,
specifically on Nvidia GPUs.
Speaker 3 (30:19):
Though, yes, it specifically works on Nvidia and nothing else,
and so AMD would have to use i don't know,
open cl or something else, depending on like for video encoding,
you maybe use open CL. But that's the biggest limitation
where a lot of the sort of so called AI
tasks are reliant on Kuda to operate at any reasonable speed.
(30:43):
And so until there's a big push to move away
from that, it's Nvidia GPUs all the way down, you know. So,
and maybe the push is export control. Maybe the groups
in China that need a higher volume of GPUs than
they're able to easily get of actually say okay, let's
just let's put all of our resources on trying to
(31:05):
use open Seattle or use something out of or PyTorch
or whatever, so that may be the direction that goes.
Speaker 2 (31:11):
Yeah, it just feel it does actually feel like Couder
is the thing that needs to be broken here for
better or for worse. I'm just saying that there's a
lots of talk about oh, making more powerful chips here
and there with AMD or Huawei, but it feels like
couder really is the thing that they actually need to change.
And I guess the probably explaining to the markets what
Couder is is difficult enough, but it feels that that
(31:33):
if they could move past Kuda, that would be the
actual thing that we're doing.
Speaker 3 (31:36):
Yeah, it's kind of like the X eighty six situation
where getting off of X eighty six.
Speaker 2 (31:43):
Just for the listeners, what do you what do you
mean by that?
Speaker 3 (31:46):
Yeah, So your average computer is very likely an X
eighty six computer, which is a specific micro architecture that
has to do with how software runs and interacts with
the hardware and the other so around the system. And
so it's kind of like a prerequisite for you might
have an application that requires Windows to be compatible, and
(32:11):
this is another one of those types of foundational prerequisites. Okay,
X eighty six as opposed to for example, ARM and
ARM laptops made kind of a splash a while ago
because they were trying to basically assert that, hey, look
we can run stuff that typically would only run on
an old X eighty six micro architecture and now we
can run it on ARM. And so it's that is
(32:34):
something where Intel an AMD, although mostly Intel have a
vested interest in keeping everything on x eighty six, and
likewise on Vidia has an interest in keeping things on Cuda.
From an end user standpoint, there are benefits to this
where Kuda, for me, for rendering a video, it just
(32:55):
works better than something else, right, right, And so it's
kind of it is a chain or the egg problem
where it's it's kind of like, well, okay, you know,
on one hand, you don't want a company to develop
the monopoly only because there's some kind of anti competitive
practice that's keeping them in place. On the other hand,
you don't want to punish a company either for inventing
(33:15):
something that's just better. And that The problem we run
into with software is the developers need to choose what
to support, and a lot of times they choose n
video and maybe not something else, because if you have
limited resources, you're going to go with what ninety percent
of the market has. It's kind of a self fulfilling,
you know, it to feedback loop basically. So to expand
(33:36):
it to something like Huawei, they would have to break
that kuda. I don't know if you want to call
it a moat or something like that, but or a
walled garden I guess either one of those or a
walled garden with a moat around it. But they would
have to break that down to really become more viable
and increase the performance.
Speaker 2 (33:54):
So before we go, did you catch any black Well
out there or any signs that black Well would even
make it out there?
Speaker 3 (34:00):
Yeah, So in addition to the fifteen nineties which are Blackwell,
as far as like the high end server stuff, we
did work with. So there's two companies we worked with here.
One of them we physically visited in Taiwan, and there's
another one we were talking to in Stapore. And what
(34:20):
we learned is that there is some intermediary basically passed
throughs that can happen through for example, a Taiwanese testing
agency where a Chinese company might ask them to purchase
Blackwell servers or high end Hopper servers and bring them
into Taiwan under their company performed testing or pre testing
(34:44):
on them, assembly, whatever, packaging, whether it's a facade or not,
basically the stated purposes, make sure it all works, and
then once it does, forward that shipment to their customer,
the Taiwanese company's customer in China, and so that would
allow them to potentially get into China. And so in
that situation, yeah, we were showing a room filled with
(35:05):
for example GB or Grace Blackwell servers that were being
tested and prepped.
Speaker 2 (35:13):
Yeah, well I'm gonna I think we should wrap it there, Steve. So,
the status of the video right now as we record
this is it's offline. But by the time this goes live,
you know, in I think September the third, it will
go up. Hopefully it's back on. Where can people support.
Speaker 3 (35:28):
You gamers and access just the YouTube channel is the
best place to go. And yeah, hopefully it's back up
within It should be within a couple of days of
you posting this, unless Bloomberg submits a files a lawsuit,
in which case they'll hear about you. Everyone will hear
about it.
Speaker 2 (35:47):
Oh and no hear about it from me as well.
This is utter bullshit what Bloomberg is doing. It's ridiculous.
But you have been listening to Better Offline. I'm at
Zetron Steve. Thank you so much for joining us, and
yeah we can do something in a couple months as well,
but always push to having you ma'am.
Speaker 3 (36:02):
Thank you right, thanks so much.
Speaker 1 (36:12):
Thank you for listening to Better Offline.
Speaker 2 (36:14):
The editor and composer of the Better Offline theme song
is Matasowski. You can check out more of his music
and audio projects at Mattasowski dot com, m A T
T O S O W s ki dot com. You
can email me at easy at Better offline dot com
or visit Better Offline dot com to find more podcast
links and of course my newsletter. I also really recommend
(36:36):
you go to chat dot Where's Youreed dot at to
visit the discord, and go to our slash.
Speaker 1 (36:40):
Better Offline to check out our reddit. Thank you so
much for listening.
Speaker 5 (36:45):
Better Offline is a production of cool Zone Media. For
more from cool Zone Media, visit our website cool Zonemedia
dot com, or check us out on the iHeartRadio app,
Apple Podcasts, or wherever you get your podcasts,
Also media, Hello and welcome to Better Offline.
Speaker 2 (00:06):
I'm your host ed ZiT Tron. And of course you
can go into the notes for the episode. You can
take a look at links from a newsletter. You can
go and buy some merch as always, but much more importantly,
(00:28):
today I'm joined by the esteem Steve Burke of Gamers Nexus. Steve,
thank you for joining.
Speaker 3 (00:31):
Us, Thanks for having me.
Speaker 2 (00:33):
So you did an incredible video on the black market
smuggling of in videos AI GPUs into China, and then
there has been a thing that's happened where the video
has been pulled down due to a DMCA complaint from Bloomberg.
What is happening?
Speaker 3 (00:50):
Yeah, it's so the video, it's really cool video where
we went all over China finding these GPUs. As part
of that, we included a recap of all the news
of export controls or the US controls GPUs that are
above a certain performance threshold and the allowance to export
(01:12):
those to China by American companies, and so that was
the story we wanted to look into. Okay, but they're
still getting to China and so how is that? And
as part of that recap, go through all the news
we've got clips of you know, various politicians. There was
a clip of Obama in there, and there was a
clip of Trump in there, a couple of them talking
about AI or GPU export control bands. Of course AI,
(01:35):
I know ed is one of your favorite topics. Yes, yes,
me as well. And yeah, so we received a copyright
strike from small, small publication called the Bloomberg. Yeah. So
(01:56):
that strike took the video offline and now we are
currently challenging it. And so the strike was for specifically,
I believe it was minutes twenty two until minute twenty
three point fifteen in the video.
Speaker 2 (02:10):
Of Donald Trump speaking the President of America. Correct, Yes,
that is ludicrous. And you went to New York specifically
to have it out with them, did they Where do
we stand as we speak, because this is going out
in about a week from now. It's August twenty eighth,
so who knows what twenty sixth even, I don't know
what happened next.
Speaker 3 (02:28):
Yeah, if it goes out in about a week, then
that'll be pretty close to when we should have an
update from YouTube.
Speaker 2 (02:35):
Because the video is off right now.
Speaker 3 (02:37):
Correct, the video's gone. And it's very interesting how copyright
strikes work because if basically the way it works is
a company files a strike. They decide if there's infringement
of some kind, and they file it. YouTube then basically,
especially for comedy, Bloomberg size will automatically approve it. It's then
(03:00):
up to the creator to dispute that the videos offline.
During this entire process, any money from the video is
held at least in escrow, and so then you can
dispute it if you want. At that stage, YouTube then
decides if it's going to accept your dispute. It accepted.
Ours is written by a lawyer, so that makes sense,
and then they step back and they're like, Okay, we're
(03:22):
not involved anymore. It's between you guys. And so the
party that filed the claimant, so Bloomberg here, has ten
business days to file proof of lawsuits. So they if
they submit to YouTube proof that they are suing us,
which thus far as we record this, they have not,
(03:42):
then at that stage YouTube would keep the video offline
and they kind of let the two parties resolve that
amongst themselves. If they don't submit any proof of a
lawsuit or they don't file a lawsuit and then show
YouTube that they've done so, again, they haven't at this point.
Then basically in ten business days the video goes back up.
We get fucked over for the ten days in between,
(04:03):
and then the money, in theory should be released back
to us, the ad revenue that was made from it,
if it follows that course. But the biggest damage that's
done is from the loss of momentum in those you know,
total time period maybe thirteen days or so. But why would.
Speaker 2 (04:21):
Bloomberg do this? Are they really that attached to videos
of Donald Trump? Or is it something else?
Speaker 3 (04:27):
I mean, I yeah, we talked about this in our
video where we flew out to try and speak with them.
I have a number of theories. We don't know their
motive precisely, but you know, there's a few strange things
going on where first of all, I've noticed that they
(04:48):
have not claimed any or struck any that I'm aware
of the re uploads of the video. So that's interesting
because there's hundreds of them. It was just hours that
want top initially. Secondly, we have used that clip before
and that hasn't been stricken and so specifically in this
(05:08):
black market video. And yeah, the theories we came up
with were there's some competing content from Bloomberg where they
attempted to do what we did. It was really a
failed attempt. They tried to find smuggled ai GPUs in
China as far as they kind of presented it in
data centers, they were unable to do so they couldn't
(05:30):
get any access. They filmed sand in the desert. They
go home. There's also that's all they go. Yeah, they
drive around at a desert, they point the camera at
some buildings that are being constructed. They say, you know,
there's ai GPUs in them their hills, and then that's
about it.
Speaker 2 (05:51):
It's kind of funny that you you appear to on
your own and with your with your team obviously able
to get this access to these large legacy media people don't,
which it's meant to be the other way around based
on what everyone says.
Speaker 4 (06:05):
So what is it the makes your approach.
Speaker 2 (06:08):
Different to bloombugs, Like why did you get stuff that
they didn't?
Speaker 3 (06:13):
I think rather than just the desert, right, Yeah, Well,
to be fair, we did not film the desert, so
they did have an exclusive there. But I think it
comes down to the approach where we looked at it
like this is an interesting concept, this idea of black
market GPUs, But from my point of view, we approached
(06:36):
it as well, let's just see what are the people
involved in this chain think and do they think it's
a black market or do they just think it's a market.
And a lot of the people we worked with were just,
I think, amused at the concept of being involved in
the media process and wanted to see how it works.
(06:57):
And so I honestly, I think a lot of that access,
especially to the people who are just sort of like
normal people. They're not these big, you know, executives at
some huge corporation. They're they're middleman and they're people transacting
GPUs on the ground. I think for them, if you
go into it without this free supposition of some kind
(07:19):
of particular i don't know, leaning or ethics of whether
what they're doing is or isn't okay or whatever, or.
Speaker 2 (07:29):
Even expecting what you'll find exactly.
Speaker 3 (07:32):
Yeah, I think if you go in with an open
mind of like we're just gonna hang out with this
guy for a day and just observe his job, people
are you know, they're pretty happy to share that.
Speaker 2 (07:42):
So, yeah, how did you actually start those? So you
you came up with the idea that you wanted to
see how elicit GPUs making it into China. Did you
have contacts in China or Hong Kong and so on
that you talk to in advance and they kind of say, oh,
if you come here, we can show you this. What
was the process?
Speaker 3 (08:02):
Yeah, the process was really cool. We're still kind of
learning how to iterate on this, but each time we're
The big thing we learned this time for this investigation
was really just trying to that first contact matters a lot,
and being able to work with that first person we
(08:22):
find to then find their contacts and kind of follow
the chain down the line. And so the first guy
we found is a professor at the Chinese University of
Hong Kong. He buys and uses these high end GPUs
in servers for educational purposes and research purposes. And as
soon as we made contact with him, we were able
(08:46):
to work with someone who first of all speaks you know,
native level excellent English, so that certainly helped. And then
secondly who has a lot of local knowledge and contacts.
And I think a lot of it is just kind
of knowing which, in our case, which link in the
chain you need next to make the story makes sense.
(09:09):
So in his case, he's a user of these GPUs.
So next we need is some kind of supplier, you know,
and then we need their supplier.
Speaker 2 (09:17):
Yeah, so what it's all quite complex. I did watch
the video and it was it was what three three
hours or so or more even really wonderful. And so
what exactly are the GPUs that are restricted? Was it
just aigpus? Was it like consumer level ones as well?
What is exactly being handed around?
Speaker 3 (09:39):
It's both. So on the high end side, there's some
smuggling and presence of these data center class GPUs, so
that'd be things like EH one hundreds, Hopper one hundreds,
A one hundreds. We actually physically saw several of those
(10:00):
and things like that. On the consumer side, there aren't
as many restricted consumer GPUs, but there are a few.
So the RTX fifty ninety is one of them that
has thirty two gigabytes of video memories. That's very valuable
for these use cases. The ARTAX forty ninety actually is
another one, and that one's kind of amusing because it's
(10:21):
so GPU right, yes, consumer GPU, yeah, and last gen
consumer GPU at that, but it's easily modified to carry
double the video memory as the actual official end video
spec and the skew so they can increase it from
(10:42):
twenty four to forty eight gigabytes of memory, which is
extremely valuable for these training and LLM type tasks where
it may be a last gen So it's technically a
little bit slower, but if you can fit it in memory,
you can still run it, versus if you can't fit
it in memory of the model you made, not be
able to run it at all. And so those are
(11:02):
the consumer cards the forty ninety. That was interesting because
one of the guys we spoke to who is effectively
a buyer and seller of these band GPUs, he was
asking me on camera. He was seeking confirmation of well,
but are you sure the fifty nineties banned? Yeah? And
(11:23):
I said, yeah, I'm sure. Here it is on the list.
And he was so confused because he said, but they're
just all over the market next door. We walk over
there and there's dozens of them.
Speaker 2 (11:37):
So it's so strange as well, because it doesn't seem
like you basically walked off the plane and had a
price sheet from just a guy, Like it wasn't clear
(11:58):
how you got the price sheet exactly. Perhaps you can't say,
but it was just Yeah, these things a band other
than the fact we have just like guys who sell
them literally everywhere. Yeah, getting in.
Speaker 3 (12:10):
So there's a couple of ways. The one of the
ones that we showed in the video is really interesting.
To the very end of the process. We actually this
is one of the reasons we delayed initial publication. We
were able to, with help of a viewer, locate someone
you could actually classify as a smuggler in the US.
So someone from China in the US.
Speaker 5 (12:34):
And.
Speaker 3 (12:37):
He buys GPUs from people in the US. He drives
around all over and buys most through forty nineties, which
are in a high demand in China, and then he'll
strip those downs that could be removing the cooler. I
don't think this particular guy disolders them. I've heard stories
of some people desoldering GPUs and mailing them that would
(12:59):
reduce your shipping costs a lot, increase your margin. But
he buys the forty nineties ships the boards back his
profits about three hundred dollars US per board before taxes.
It wasn't a one hundred percent clear if he does or
doesn't pay taxes. And then at that point, once they
are either shipped into Hong Kong or Macau directly if
(13:23):
they feel gutsy enough, or indirectly through a third country,
a middle country where there is no export control, you know,
then they arrive and they get redistributed from there. There's
sometimes hand carried as well, like by students, international students
who just fly back with a fifty ninety. They bought
it best Buy or something, and they can double their
money roughly if they're lucky.
Speaker 2 (13:44):
That's so, it's wacky that there is such a big
black market, but I guess the export bends just created it,
as did this whole AI, this crazed AI moment. Did
in fact, while you were over over in Hong Kong,
did you actually go to China or can't remember? It
was just in home? So did you? This is an
anecdotal thing. How was the pressure of AI there? How
(14:06):
big was the conversation the advertisement? Did you see it
on the same level? It's fine if you didn't, I'm
just curious.
Speaker 3 (14:12):
I saw it more in Taiwan, which we went to
as well. So yeah, after the China trip, we went
over to Taiwan, and I don't think it made it
into the cut, but we shot a short clip of
just a bus, like a public bus with a gigantic
AI the laptop advertisement on it. So I would say
it's kind of at least there, it's similar to what
(14:34):
you see here where it's you know, if you're anywhere
in the vicinity of a best Buy or a city
that has technology dealers, then you're gonna see AI ads
In China, Yeah, I saw it in the tech markets
for sure, some banners, but it just it. I don't know.
That's I only understand Chinese if I'm actively trying to
(14:58):
understand it, you know. So it's I think a lot
like it's filtered out for me, right, Yeah, I was good.
Speaker 2 (15:03):
Yeah, I assume you speak some You seem to speak
some Chinese. I can't speak anything but English, so I
could not correct your Chinese, doll.
Speaker 3 (15:10):
You do speak more proper English than I do, though,
so I'll.
Speaker 2 (15:13):
Get Yeah, yeah, I speak all kinds of it. These days.
It's not going well. People criticize my accent all the time.
So as far as how the government sets these limits,
one thing he did really eloquently towards the beginning of
the video as well, was he kind of went through
how arbitrary the limits seemed to be and how in
Video seems very capable of adapting around them to the
(15:34):
point they almost doesn't make sense.
Speaker 3 (15:37):
Yeah, yeah, they they've modified the regulations a few times.
And it's it's so this particular part I don't really
blame and video for for this, which is just trying
to you know, the government sets some regulations and a
(15:57):
company says, okay, we'll comply with these, and then they
comply with them, and then the government kind of says
no weight not like that.
Speaker 2 (16:04):
Yeah, that's very standard for pretty much or regulation. It
maybe it's good or bad what have you, but it's not.
It happens all the time with companies.
Speaker 3 (16:14):
Yes, And in this case, one of the things that
as the I guess AI world developed and the applications,
one of the things that changed was this understanding that
memory capacity is basically the most important aspect of it.
And the initial versions of the formula didn't consider memory
(16:36):
in any scenario. What I remember, it was based on
the marketed flops of performance on the spec sheet multiplied
against the bit length of the operation or something like that,
and they later added memory bandwidth. Memory capacity though is
still as far as I'm aware right now, is still
(16:58):
not considered in the formula.
Speaker 2 (17:01):
Yeah, it just it feels just arbitrary. But I imagine
also it's quite difficult to find a way to do this,
especially when the people setting the regulation might not have
the best handle on computers.
Speaker 3 (17:13):
Yeah, I think I think the best way to do
this would probably be if they wanted to and I'm
not I don't have a particular one way or the other,
but if they wanted to control it, I think the
best way is probably some kind of benchmark or set
of benchmarks, just like you would do for a review,
where you know, you say, okay, if it exceeds some
performance threshold in this real world use case or multiple
(17:36):
of them in aggregate, then it's control, it's export controlled.
But instead they just kind of calculate based on the
spec sheet and the companies that make these things. Obviously,
you know, full appropriate credit to Nvidia. The company is
extremely confident with making GPUs. They know what they're doing,
and they're going to know a lot more about how
(17:58):
to tweak those dials to comply then a government agency
will so.
Speaker 2 (18:05):
So on the subject of Nvidia knowing things, do you
think that do they seem aware of any of this,
like any of any of this smuggling happening. Do they
seem to Did you get any comment out of them?
Speaker 3 (18:17):
We asked a lot of the people in the chain
what their perception was of Nvidia's awareness, and so the
professor at the Chinese University of Hong Kong, he had
a great answer when I asked him, do you think
Nvidia knows? And he kind of he paused for a second,
and then he said, how could they not? You know,
(18:38):
like these things are very expensive. You've got thirty plus
thousand dollars GPUs and it just seems like they would
all be tracked. So what was the term?
Speaker 2 (18:48):
There was a specific terms like kind of like basically
equivalent of ignoring him.
Speaker 3 (18:55):
Yeah, so the Chinese has done eg and b egn,
which means open one eye, close one eye, or turn
a blind eye? Is the idiot?
Speaker 2 (19:04):
Yeah, And that's awesome.
Speaker 3 (19:08):
It's just very cool. Yeah, it's super I mean the
coolest part about that was multiple people said that exact
idiom and it's it's just it's such a specific choice
of words. It'd be the same as if you asked
like three people a question, you know, they all said, oh,
(19:28):
they turn a blind eye. At some point you're like, Okay,
this seems to be a really consistent belief here, But yeah,
I think from a video standpoint, it's whether or not
they want to actually control what gets in. It's hard
to say. Certainly, from a pure financial standpoint. The more
(19:49):
gps they sell, the more money they make that they
probably don't really care who they sell them to as
long as they sell them to someone and then where
they end up after that. I mean, maybe to the
extent that video doesn't want to piss off the United
States government. They you know, they try to comply as
much as they reasonably can. But how much do they
actually try to control ingress into a country that's export controlled.
(20:13):
It's hard to say, But certainly I think they are
aware of this happening, and I just I have a
really hard time believing that they are unaware their gps
are making it into China when they shouldn't be.
Speaker 2 (20:26):
And I imagine it must be quite hard to stop.
Like I hate to have any sympathy for in video,
but it does. It does feel like something that would
be kind of a moving target. Because one thing that
was consistent across the whole video was the ingenuity of
the people involved. It does make you see like American
entrepreneurs sometimes and they're like, oh, yeah, it's tough to
(20:48):
you know that the hours are very tough, and you know,
I was covering all not and all of these guys
were just like, yeah, I've got a thing in the
back of my car full of wires. Yeah I've got
I have a whole rig.
Speaker 3 (21:01):
Yeah it's crazy, and that is I mean again, you know,
from that particular side of things, I don't suspect. I mean, look,
it's a government regulation, and if the government wants to
regulate it, it's kind of up to them. Obviously the
companies need to comply with it, but I wouldn't expect
(21:22):
a company, even Nvidia, to send in any kind of enforcement.
Maybe what they would do is if they wanted to
enforce it, you know, they might identify, Okay, this purchaser,
there's evidence of this purchaser exporting to China anyway, and
so we're just gonna stop selling to them or something.
But that's kind of I think the most you might
(21:44):
expect of them. But yeah, as far as the ingenuity,
it was pretty crazy. Like the guy in the US
who buys the cards, he he just got a Prius
with a massive like battery bank in the back, like
a ups with a lithium ion battery in it. He's
got a test bench hooked up to it, an ATX
test bench. Pops the trunk open. He's got a spare
(22:05):
license plate in the trunk for some reason. I don't
know why. Uh, that's not to ask. Yeah, I don't
need to know. And uh yeah. And if you want
to sell him your card, you're just some American new
found him on Facebook Marketplace. He at the time we
were talking with him, he was paying two thousand dollars flat.
He makes three hundred when he sells it, and you
(22:27):
bring them the card, meet him wherever he tests it,
and in the back of the you know, in the
trunk of the car, he just runs a benchmark, make
sure it's good, and then buys it and ships it
back or carries it back, depending on how secure he feels.
And then also to your point of ingenuity, the shops
(22:47):
that are capable of modifying cards, where they're just they're
repair shops, and they just happen to be really good
at board level BGA or ballgrid array device repair or
or swapping, so they can pull memory modules. They can
add memory modules, swap the GPU between PCBs, and they're
(23:08):
able to keep Silicon in service where you know, they're
just repair shops, so it's kind of like from their perspective, look,
they're not doing anything illegal in China, they're just keeping
boards in service. But for people who need these high
capacity GPUs for AI tasks and they having to be
export controlled, a shop like this would be capable of
taking a card that has died, you know, through through
(23:31):
use or just needs to be upfit to be better,
and they can do that for pretty cheap and basically
in a couple hours.
Speaker 2 (23:40):
And it's not illegal to sell them in China once
they are there, right, They've like that is not a
like China is not prosecuting people for any of this.
Speaker 3 (23:50):
Correct that it is not illegal to own or sell
these GPUs or buy them in China. It's legal if
you are an American company, if you're Nvidia or your
best buyer or a partner or whatever. It would be
illegal to sell the export controlled cards to a company
(24:12):
or an entity or person in China. But once it's there, yeah,
there's no real control over it. The only the only
place that we found that China steps in is if
there's some kind of undeclared ingress through port. So like,
for example, there were CPUs and GPUs at various points
(24:34):
in the last several years that were smuggled in with
creates of live lobsters or prosthetic baby bombs.
Speaker 4 (24:40):
But was that real?
Speaker 2 (24:41):
Was the lobes the thing real?
Speaker 3 (24:43):
The lobster things real? Despite what Nvidia said.
Speaker 4 (24:59):
Hell yeah, I love the So what actually describe the full.
Speaker 2 (25:03):
Loves the situation? I think everyone wants will want to
hear it over lobs.
Speaker 3 (25:08):
Yes, So the loster situation as it were is basically
there was a Hong Kong uh port, so customs intercepted
a white van I think it was. There's like a
customs post they post photos of the things that they
(25:28):
are able to snipe being brought in improperly. So Hong
Kong's a free trade port, but there's still certain declarations
you have to make and sometimes people won't make those.
And so there was a van bringing in crates of
undeclared live lobsters driving on I think it was the
(25:49):
Macau or the jew Hai Bridge or something. But anyway,
they got intercepted and alongside the live lobsters were GPUs
and they were older, so I think they were quadros,
but they're brought it all being brought in improperly. And yeah,
so customs post the thing. They're like, we found this,
you know, and I think that's just kind of like
(26:09):
the here's us doing our job and maybe scared people
off or whatever. And then later AI startup Anthropic released
a statement talking about how there needed to be more
strictly enforced export control in the US, and they mentioned
(26:31):
smuggling of GPUs with lobsters, and Vidia responded and I
don't have their statement in front of me right now,
but it was something along the lines of it would
be better if American you know, AI companies would focus
on innovation rather than the part I do remember was
tell tall tales of prosthetic baby bump smuggling or GPU's
(26:56):
sensitive electronics being smuggled with live lobsters. Turns out that
is not a tall tale. That is a thing that happened.
Speaker 2 (27:02):
So, yes, this is the thing people say. Jensen Wong
is cool. If you're cool, you say, yeah, well, our
chips like, we don't like this happening. This is bad,
but doing everything we can't stop it. But yeah, they're
doing James bondshit to get our GPUs, Like, come on,
you could just say that, like they do seem a
little sensitive, and we're we're recording this the day before
(27:24):
their earnings, and I got no idea what's going to happen,
No one does. Everyone's a bit everyone's a bit worried
about it. So it's I think it. The China market
is going to get extremely interesting because they've got the
H twenty band. But now the H twenty band has
been lifted, and now we're in this weird spot where
I don't know if you've heard anything about this yourself,
(27:45):
where there are suggestions that the Chinese government is actively
saying not to buy in video GPUs. If you heard
anything about that.
Speaker 3 (27:52):
Yes, So the Chinese government, I believe it's called the
Cyberspace be a security agency or something like that, but
one of their government agencies put out a statement following
separate statement from a state owned or at least state
(28:12):
you know, controlled in some capacity media report where the
statement said or at least floated the possibility that Nvidia's
H twenties may have some form of spyware in them.
And so the suggestions we're tracking or government back doors,
and so that was kind of the accusation, and Vidia
(28:35):
of course posted a statement saying absolutely not, this is
not a thing, and you know, just to be fair here,
there's not currently, as far as we're aware of, any
firm evidence of this. But I'm also sure that governments
know things about you know, about products from other nations
that we would never know anyway, So who knows. But
(28:55):
that's where it stood. Nvidia said, we don't have these.
China's agency said to companies in China that they should
be wary of using H twenties and advised against purchasing
them for fear of some kind of backdoor or spyware.
Speaker 2 (29:11):
So how realistic would it be for Actually, did you
hear any scuttle about around this of switching to Huawei
chips or something?
Speaker 3 (29:19):
We did ask? Yeah, I think we might have only
kept one of those questions in the actual video, but
I did ask a lot of people Huawei from the users' perspectives,
the people we spoke to, they none of the people
we spoke to are using Huawei components right now, so
(29:41):
they're getting more powerful from what I understand, I think
there's a few hurdles. One of them is. It's kind
of the same reason where the companies aren't even really
using AMD and they've been doing this for a while,
whereas Huawei is pretty new. And a lot of that
comes down to Kuda, where Nvidia is the proprietor of Kuda,
(30:03):
which is a library they can use to accelerate tasks
in which a lot of software is written to use,
I mean even just rendering a video.
Speaker 2 (30:12):
It's the coding language to get accelerate computing, right, correct, Yeah,
specifically on Nvidia GPUs.
Speaker 3 (30:19):
Though, yes, it specifically works on Nvidia and nothing else,
and so AMD would have to use i don't know,
open cl or something else, depending on like for video encoding,
you maybe use open CL. But that's the biggest limitation
where a lot of the sort of so called AI
tasks are reliant on Kuda to operate at any reasonable speed.
(30:43):
And so until there's a big push to move away
from that, it's Nvidia GPUs all the way down, you know. So,
and maybe the push is export control. Maybe the groups
in China that need a higher volume of GPUs than
they're able to easily get of actually say okay, let's
just let's put all of our resources on trying to
(31:05):
use open Seattle or use something out of or PyTorch
or whatever, so that may be the direction that goes.
Speaker 2 (31:11):
Yeah, it just feel it does actually feel like Couder
is the thing that needs to be broken here for
better or for worse. I'm just saying that there's a
lots of talk about oh, making more powerful chips here
and there with AMD or Huawei, but it feels like
couder really is the thing that they actually need to change.
And I guess the probably explaining to the markets what
Couder is is difficult enough, but it feels that that
(31:33):
if they could move past Kuda, that would be the
actual thing that we're doing.
Speaker 3 (31:36):
Yeah, it's kind of like the X eighty six situation
where getting off of X eighty six.
Speaker 2 (31:43):
Just for the listeners, what do you what do you
mean by that?
Speaker 3 (31:46):
Yeah, So your average computer is very likely an X
eighty six computer, which is a specific micro architecture that
has to do with how software runs and interacts with
the hardware and the other so around the system. And
so it's kind of like a prerequisite for you might
have an application that requires Windows to be compatible, and
(32:11):
this is another one of those types of foundational prerequisites. Okay,
X eighty six as opposed to for example, ARM and
ARM laptops made kind of a splash a while ago
because they were trying to basically assert that, hey, look
we can run stuff that typically would only run on
an old X eighty six micro architecture and now we
can run it on ARM. And so it's that is
(32:34):
something where Intel an AMD, although mostly Intel have a
vested interest in keeping everything on x eighty six, and
likewise on Vidia has an interest in keeping things on Cuda.
From an end user standpoint, there are benefits to this
where Kuda, for me, for rendering a video, it just
(32:55):
works better than something else, right, right, And so it's
kind of it is a chain or the egg problem
where it's it's kind of like, well, okay, you know,
on one hand, you don't want a company to develop
the monopoly only because there's some kind of anti competitive
practice that's keeping them in place. On the other hand,
you don't want to punish a company either for inventing
(33:15):
something that's just better. And that The problem we run
into with software is the developers need to choose what
to support, and a lot of times they choose n
video and maybe not something else, because if you have
limited resources, you're going to go with what ninety percent
of the market has. It's kind of a self fulfilling,
you know, it to feedback loop basically. So to expand
(33:36):
it to something like Huawei, they would have to break
that kuda. I don't know if you want to call
it a moat or something like that, but or a
walled garden I guess either one of those or a
walled garden with a moat around it. But they would
have to break that down to really become more viable
and increase the performance.
Speaker 2 (33:54):
So before we go, did you catch any black Well
out there or any signs that black Well would even
make it out there?
Speaker 3 (34:00):
Yeah, So in addition to the fifteen nineties which are Blackwell,
as far as like the high end server stuff, we
did work with. So there's two companies we worked with here.
One of them we physically visited in Taiwan, and there's
another one we were talking to in Stapore. And what
(34:20):
we learned is that there is some intermediary basically passed
throughs that can happen through for example, a Taiwanese testing
agency where a Chinese company might ask them to purchase
Blackwell servers or high end Hopper servers and bring them
into Taiwan under their company performed testing or pre testing
(34:44):
on them, assembly, whatever, packaging, whether it's a facade or not,
basically the stated purposes, make sure it all works, and
then once it does, forward that shipment to their customer,
the Taiwanese company's customer in China, and so that would
allow them to potentially get into China. And so in
that situation, yeah, we were showing a room filled with
(35:05):
for example GB or Grace Blackwell servers that were being
tested and prepped.
Speaker 2 (35:13):
Yeah, well I'm gonna I think we should wrap it there, Steve. So,
the status of the video right now as we record
this is it's offline. But by the time this goes live,
you know, in I think September the third, it will
go up. Hopefully it's back on. Where can people support.
Speaker 3 (35:28):
You gamers and access just the YouTube channel is the
best place to go. And yeah, hopefully it's back up
within It should be within a couple of days of
you posting this, unless Bloomberg submits a files a lawsuit,
in which case they'll hear about you. Everyone will hear
about it.
Speaker 2 (35:47):
Oh and no hear about it from me as well.
This is utter bullshit what Bloomberg is doing. It's ridiculous.
But you have been listening to Better Offline. I'm at
Zetron Steve. Thank you so much for joining us, and
yeah we can do something in a couple months as well,
but always push to having you ma'am.
Speaker 3 (36:02):
Thank you right, thanks so much.
Speaker 1 (36:12):
Thank you for listening to Better Offline.
Speaker 2 (36:14):
The editor and composer of the Better Offline theme song
is Matasowski. You can check out more of his music
and audio projects at Mattasowski dot com, m A T
T O S O W s ki dot com. You
can email me at easy at Better offline dot com
or visit Better Offline dot com to find more podcast
links and of course my newsletter. I also really recommend
(36:36):
you go to chat dot Where's Youreed dot at to
visit the discord, and go to our slash.
Speaker 1 (36:40):
Better Offline to check out our reddit. Thank you so
much for listening.
Speaker 5 (36:45):
Better Offline is a production of cool Zone Media. For
more from cool Zone Media, visit our website cool Zonemedia
dot com, or check us out on the iHeartRadio app,
Apple Podcasts, or wherever you get your podcasts,
Host
Ed Zitron
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
NFL Daily with Gregg Rosenthal
Gregg Rosenthal and a rotating crew of elite NFL Media co-hosts, including Patrick Claybon, Colleen Wolfe, Steve Wyche, Nick Shook and Jourdan Rodrigue of The Athletic get you caught up daily on all the NFL news and analysis you need to be smarter and funnier than your friends.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.