원문 보기
Welcome to Processing Unit in New Show at the intersection of business and technology. What do I have on the docket for you today? Well, let me show you with the help of a little bit of music. All right, so let's start off with the news that Elon is going to give um kids in the US who have a Trump account one SpaceX shares and he's going to increasingly, you know, give away more shares.
And this is a very interesting The typical approach done by most billionaires is the NGO industrial complex, right? It's not the most effective. There are questions around, you know, why are they doing it that way? And I actually found a video from It's from 6 years ago, so it's it's basically a longer documentary by The Economist on giving.
And there's one very critical journalist, and probably 6 years ago, this is a very kind of niche leftist POV. But I think actually, also especially with the rise of the far left, this view on philanthropy will especially will increasingly become commonplace. So, giving directly and giving shares, which gives people a sense of what is capitalism about, right?
They can see the accrual of value from the moment they got the share until the moment they can sell it or or keep it. Um is an important distinction. Um and it will increasingly, I predict, be the way that the ultra-rich will give. But to understand, you know, where the grievances with current philanthropy are coming from, let's listen to this uh journalist first.
What I see is a room full of people who think they are helping, but are working at much greater scale to maintain and entrench a system that frankly dooms the people that they're helping. Real altruism would actually be doing less harm, not running working people into the ground through the pressure they put on the companies they take stakes in.
In the past 30 years, the number of foundations in America has almost tripled. Since 1978, the proportion of overall giving that's come from those foundations has also tripled. But the US Treasury estimates philanthropy will cost it $740 billion in lost tax revenue over the next decade. Anand claims this giving by wealthy Americans is more about tax breaks than charity.
Poor people, people who make $20,000 a year, are paying higher taxes than they otherwise would to subsidize about $50 billion in tax breaks every year that we give people for donating money. You are injecting harm into the society. You are making more money, and then you are going to the Robin Hood gala to donate 1% of what you have stolen from the common good to a fraction of the people whose interests you have harmed.
So, this is a very very cynical take, right? Obviously, you could argue that perhaps some um of the ultra-rich do this for these reasons, but I think you will increasingly hear this POV that traditional philanthropy will come at the expense of tax revenue. There are a lot of questions around how effective is it.
So, it goes through this large machine of foundations who have employees. How much of the actual dollars arrive, you know, actually with the people that need it, and how much of the solutions that are funded are actually measured and effective, and useful. The effective altruists used to, you know, basically uh care a lot about this before moving on to AI safety.
It's a very challenging problem. Um it doesn't mean that people will stop giving in the traditional ways. Obviously, these are the most popular ways of doing philanthropy, and they will be around for a long time. But, you increasingly see people think about these new ways of giving. So, the donor accounts are something that um are an alternative.
Um you could wonder whether again a government-directed approach is the best approach, but again, I think more and more of the wealthy and the ultra rich will continue to think about direct giving. Giving shares, I mean, is very interesting because it gets people a real sense of what um what capitalism is about, how they can have a little part of it and hopefully understand that, you know, once they have enough income, they too can participate in the upside by buying shares.
You could say, you know, you could see them on the other side and say that's a lot that's a stretch. A lot of people will basically, when they get access to their Trump account, basically sell all the equities and use the money to have fun or pay off debt or whatever. But nonetheless, that first interaction will maybe allow 10 to 15% to think like, "Hey, um there's more to this and maybe I shouldn't sell it all.
Maybe I should keep some of it. Uh maybe I should actually participate in this system that has allowed, you know, what will be 12 to 20k roughly, depending on how much contributions they got, to arrive there at the age of 18. So, expect more people to actually think about giving directly. I think um that is something that is still not that popular now, but given Yaron's efforts and the Trump accounts, I expect more people to come around and understand that they just have to or it is better to bypass the NGO industrial complex and just give things to people directly.
Most people know what to do with it. They don't need to be pandered to. They don't need to be lectured. Um they don't need to, you know, be seen as people who need solutions that you finance. Just give them the amount directly and I think that's best way for them to both experience capitalism and not be pandered to.
All right, let's move on to Ken Griffin. Now, Ken Griffin recently said that he didn't say it. He didn't say we're in an AI bubble. Um uh and our journalist sort of suggested it and he kind of went along with it. Let's have a listen first before we uh discuss this one. No, is how as as these models become more intelligent, how much are they able to to demand extremely high pricing because they help to actually solve the problems that have vexed humanity for centuries.
Because people don't know the answer to that, often the conversation goes to this this bubble one. That maybe this is a bubble. Um Ray Dalio thinks it is. Howard Marks says he's on bubble watch. The BIS says that it has echoes of the dot com boom. And I wonder whether it is knowable if this is a bubble.
And if you have any particular thoughts about that, if it is a bubble and if it one it's one that looks like it's going to burst at least in certain parts of the different AI layers. So so the big picture is there's there's almost certainly bubbles within this moment in time. And and any rapid transformation of our economy is generally intertwined with a bubble.
All right? I remember back in 1999, 2000, and the metrics that we used to value internet companies. And and let's be clear, some of these internet companies had pretty fanciful business model. But I mean, I just wish I'd kept my Amazon stock from that moment in time. It's all about selection. the internet which was, you know, which was the great bubble.
Could it be more of a part of our daily life today than it is? I mean, I I I've told the story before that I was updating my Xbox with my son a few years ago. He was 13 at the time. He And he looks at me and he goes, "Dad, you you lived before the internet existed." So, this is the key point, which is do we need Well, do we need bubbles?
Are bubbles inevitable? To some extent, you could say that you need these cycles, right? These bearing points of innovation where all of this enthusiasm comes together, basically creates momentum in the capital markets, funds a lot of innovation. Some of it, you know, actually sustains itself like Amazon and Google and some of the you know, early web 1.0 companies, and some of it just dies, right?
Like there were some companies like Webvan that didn't make it. And we're in a similar kind of bifurcation moment right now. There's a lot of momentum, there's a lot of excitement, but inevitably, inevitably, there will need to be a bifurcation moment, and you know, the cream of the crop has to be filtered out.
And a lot of the discussion, I presume, in the next months to come, will be about, you know, who do you separate from the pack? Um that discussion has already started a little bit. We're still a bit We're not early, but we're still in the kind of core infra stage. Um this also connects us to our next topic, which is that basically, the bricks, the underlying bricks that build the foundation, are still the going narrative, right?
There was an article in the Wall Street Journal, which uh basically says that uh Micron and Nvidia basically make up 35% of the earnings. Boom, right? The The The indexes already heavily weighed towards tech, and it's these two guys and the AI revolution that are basically kind of holding up the sky.
And we are still very much at the again, the brick stage, right? I We know we'll we'll talk about neo scalers, hyper scalers in bit, but again, there's still demand for the product of these guys because there's still belief there are still a lot of people who believe that the error of omission, which is not investing, is, you know, greater than the error of commission.
Zucker has said something similarly. It's better to bet, and, you know, in his case, because he can bet, you know, tens of billions wrongly, maybe even a hundred billion wrongly, than to miss um the trend. That's just the dynamics in in technology. And this is also the dynamic that eventually creates bubbles.
Not everyone is as savvy, not everyone is going to make it. And that's how you basically get to a bubble. And now we're getting at the stage where more and more people are wondering about, you know, how do you select out the right versus wrong? This is basically what Ken Griffin is also alluding to.
And some who have an interest um in this are already starting to you know, are are kind of flaming that conversation. It started with Elon and his charity. Now, he also, obviously, given who he is, has a say on who to select, how to select. He had some uh takes on the uh neo-scalers that have moved the markets, and uh it's definitely a a controversial take.
I don't know if I fully agree with him, but I think uh more and more people I think there he really has a point, which is an increasing number of capital allocators are going to really try to better understand where value is, right? Build better mental models, scrutinize the financing more. So, I think when it comes to that point, Elon, given even though he has an interest, does have a point.
You need some sort of mental model, some sort of financial rigor as we move into this next phase where a lot of people are building out compute and even though you could say that, you know, bubbles are almost near inevitable you also want to avoid this kind of mindless build out where everybody gets cheap capital and everybody just uh builds over capacity.
There a little bit more scrutiny, a little bit more um direction in terms of who deserves capital and who doesn't um should obviously be very welcome. Um let's listen in on uh the uh the take and what Elon had to say uh uh this week. for Nvidia chips with Apollo backing it is is confirmed. They're they're doing that now, right?
You know, this comes on the back of this weekend when Elon Musk was talking about how he also confirmed that we got this $40 billion deal for them for Nvidia GPUs. Uh but then he also added a couple other things, right? He talked about how debt markets are basically done giving money to neo clouds, right?
Who rely basically on the revenue streams to solely pay for these these data centers. He said, "Quote, that means that 200 plus neo clouds are going to see guppy level death and consolidation. Some will make it though, but not many." his like confidence. Like first of all, he said the deal was done before it seemed to have been actually done.
The other thing I would say that is really important about this $40 billion deal is it hasn't yet priced. Correct. Mhm. So, we don't know how much it's going to cost SpaceX to do this. We do know that over the course of this year since SpaceX has come public and it has done some debt raises, it's had to pay more.
That's right. Right. Right? So, you know, it's going to be real interesting to see what the rate is on this. reporting is that it was down to 8% is what the debt would probably price at potentially. Which is expensive. Would up from what they were expecting, what SpaceX and and whoever else, right? He said it's it's getting pricey.
Right. And if it's getting pricey for him, as he points out Yeah. it's going to be interesting. It's getting really bad. He's also paper I can still get it. Again, he is also trying to make sure that people don't confuse his product with other people's product, right? Right. You guys can't even get it now, you know?
You can't even pay to get it now. Well, but this is the this is the the Neo Cloud side of the core wave, you know, debt getting marked at 15%. Like that's really really expensive, especially compared to a few years ago. That's how it's trading. How it's trading. That's not it. That's not where they're That's not if it's already issued, that's not what you're paying right now.
The other thing is it's trading. The part about the the Musk stuff is that other people who are like supposed AI bulls were chiming in on Musk's tweet and we can agreeing with him. Like yeah, the Neo Clouds are they're going to they're in a tough spot. It's not going to go well for them. One of them being Chamath.
But it's just it's just interesting that even the bulls are saying, "Oh, you know what? There might be some issues here with raising more and more money. Uh sorry, you know." Yeah, you can see the narrative forming around Neo Clouds, right? I mean, especially for the hyperscalers and for Elon, they are an easy target.
Um they are not at least most of them are not of the best quality. Though again, we should all be mindful of you know, basically throwing the child out with the bathwater, right? It's easy to basically say, "Hey, Neo Clouds is Neo Clouds that." But eventually the I would say what I would coin the GPU vigilantes will come for everyone, right?
So I think people should also be mindful in terms of enabling the narrative too much. There is obviously some sort of discipline that's needed, but you could also very easily play into the hand of short sellers and basically throw out the baby with the bathwater. But we're still confident that this will still change our lives and and change business and change everyone's sort of, you know, how you live in the world.
There there is also a there's a little bit of monkey see, monkey do with Musk. Like whatever he says tends to get amplified really heavily, kind of regardless of whether it's correct or not. And the the data would seem that this is a very salient point he's making that the neo clouds are going to struggle.
But kind of regard Yeah, there I do agree. Um it is a data-driven point, right? Um he doesn't just say it in general. He makes some concrete arguments and it's hard to come up It's hard to steal man the other side, right? But again, it's a tightrope for all players in the tech industry. You should be mindful of not simply kind of boogeymanning the neo clouds because then people will or people will likely borrow a lot of those arguments to come after you as well.
So, it's a tricky tricky way of handling the narrative. Also, what he's saying, there is a community of like the Silicon Valley VC side who just don't want to be caught on the wrong side of his comment. Right, it's also self-serving for him. Like we're fine. You want to do business with us. Well, the interesting thing about the So, his his comment was in a reply to a Chamath tweet.
And the last part of what Chamath says is it will be also also be a consolidation year for a few. So, like there are various views of this as we've discussed. It's like, okay, could the neo clouds go out of business? Or are they just going to be bought out by other companies because their the compute that they have in their building is valuable.
So, yeah, the consolidation argument makes sense, right? This is a a similar product um that is useful to a host of different parties. So, um you know, eventually if it turns out that there is significant overcapacity, market forces will consolidate, right? I think we should also be careful with the whole bubble talk because eventually consolidation will figure things out.
I think demand for GPUs is still there. The Anthropic IPO is a good case in point. Um they still need a lot of compute. So, is there overcapacity? Most likely. Is it really significant? I'm not sure. Is it 60% of the current buildout? I don't think so. You could maybe kind of guestimate that there's about, I don't know, 10, 15, 20% overcapacity and you can fix that with consolidation.
But, I would be very, very weary to say that, you know, all neo clouds are basically overcapacity or to play too much into that argument. Ladies and gentlemen, um thank you for watching. This has been Processing Unit episode 1. I hope you enjoyed my takes and uh see you hopefully for episode 2.
Comments 0