A Lineless World: AI, X, Darkness
35:46 Share

GiveWell, a nonprofit that researches and recommends giving opportunities, takes the impact of donations seriously. To ensure their recommendations withstand tough scrutiny, GiveWell had their own researchers spend months trying to identify flaws in their past work. They then published their findings, mistakes and all, for any donors to use for their giving. It's this kind of rigor that can help your donation make a big impact on the world. GiveWell has now spent over 17 years researching charitable organizations

and only directs funding to a few of the highest impact opportunities they've found. Over 125,000 donors have used GiveWell to donate more than $2 billion. Rigorous evidence suggests that these donations will save over 200,000 lives. If you've never used GiveWell to donate, you can have your donations matched up to $100 before the end of the year or as long as matching funds last. To claim your match, go to GiveWell.org and pick

podcast and specify where you heard this ad. Make sure they know that you heard about GiveWell from this podcast. In

In 15th century Florence, the great inventor Leonardo da Vinci dreamt of creating a flying machine. But something kept getting in his way. Admin. Piles of it. Luckily, Leo used the smart buying tools on Amazon Business, so he could work more efficiently. With the extra time, he not only invented the flying machine, but actually built it. Magnifico. Incredibile. Splendido. Whoa. Easy there, Leo. Splendido.

Amazon Business, your partner for smart business buying. Developers like you are building the future, but you need the right tools to move fast and go further, right? That's where Microsoft comes in. With tools like GitHub Copilot, VS Code, and Azure AI Foundry, you have everything you need to push the limits and bring your ideas to life faster. And with security, compliance, and responsible AI built in, you can focus on what matters most.

Building the next big thing. Learn more at developer.microsoft.com slash AI. Bloomberg Audio Studios. Podcasts. Radio. News. Hello and welcome to the Money Stuff podcast. Your weekly podcast where we talk about stuff related to money. I'm Matt Levine and I write the Money Stuff column for Bloomberg Opinion. And I'm Katie Greifeld, a reporter for Bloomberg News and an anchor for Bloomberg Television.

What are we talking about today, Katie? We're going to talk about how the U.S. stock market got deep-seeked. We're going to talk about that sweet, sweet ex-debt. And then we're going to talk about trading in the dark. Sounds good.

Deep Seek. Deep Seek. So I cannot believe, I cannot believe that there is a money stuff from June that mentions Deep Seek because I would say 97% of the people that I talked to on Monday hadn't heard of it before this past weekend. Where did they hear of it? The people I talked to on Monday who hadn't heard of it previously? Yeah. They heard about it when the app shot up in the App Store. Okay.

And then you had the sell side start writing about it. There were a ton of tweets about it over the weekend. Not that Twitter is real life, but in some cases it kind of is. Yeah. Like there's one guy who wrote a long report on like his personal website and argues somewhat plausibly that he influenced a lot of the investor reaction. It's like interesting to see how like...

investor reactions coalesce, right? Because like the model was kind of released like at the beginning of last week. The catalyst is some combination of people getting to think about it over the weekend and like it's shooting up in the app store. But at some point there's like this big shift from like people using the app to everyone having

existential crises about Nvidia. Yeah, it was pretty wild. Anyway, this is a digression. Yes, I wrote about it in June. I'm prescient. I was like, yeah, this is going to be really bad for the future of internet. But I do love, you know, what I wrote about in June is like the founder of DeepSeek

It kind of spun out of his quantitative hedge fund. And I love that the skill sets of quant hedge funds and large language models are like kind of overlapping, right? These are kind of like, you know, using machine learning techniques to predict somewhat unpredictable things, whether that's like the next word in a sentence or the stocks that will go up. And classically,

people made billions of dollars. And by people, I mean, like Renaissance technologies made billions of dollars by like repurposing like people in the business of like natural language generation into predicting stock prices. And it's nice to see that come full circle. And the people who are using machine learning to predict stock prices are now getting back into the natural language game and making billions of dollars that way. Yeah, I mean, everything, everything is cyclical in that sense. I will say I wish we had talked about it on the podcast in June so that

I could have shared in some of this while it was so early, but in any case, it's fine. The thing is not that like DeepSeek is an AI company in China. The thing is that the market lost a trillion dollars of market cap on Monday. I don't want to say that the U.S. economy is based on like...

building and powering data centers for AI companies, but the projected incremental cash flows to the US economy. Like a lot of those are like, yes, we're going to build a lot of data centers and a lot of like power plants to power them. And DeepSeek arguably

people think that it undermines that case quite significantly. And so if you thought that all of economic growth would come from AI data centers, then now you're like, ah, the source of economic growth is gone, which is a funny thing to think, but yeah. I know. I think a lot of people actually would agree with that logic because you take a look at what happened in the bond market. There was this...

insane bid into bonds on Monday as well. And sort of that was bonds being a haven. Yeah, I saw you like tweet at DeepSeek as a macro event. Yeah, well, one of the most credible reasons I got was just because people are worried that this is going to shave that incremental bid off of GDP. Because there's been plenty of risk-off events where bonds didn't catch a bid. But this was the one thing that spurred like this haven bid across the treasury curve, which was pretty funny. I love...

AI as a, as a like subject matter, because it is so science fictional and

One thing people talk about sometimes is whether GDP is a bad index of human flourishing. There's this argument that GDP measures understate GDP growth because you're getting all these hedonic benefits from going on social media or whatever, and that's not captured in GDP figures or whatever. The thesis here is something like, if we have to build a lot of buildings and burn a lot of coal or natural gas to...

make really good AI models, then that's good for GDP. And if we can get the same AI benefits for free, then that's bad for GDP. But like, that's better, right? It's clearly better to have those benefits without burning coal than to have them, you know, and burn coal. I wrote something like, you know, if you sort of like fully believe the deep seek thesis and that's like this magic AI thing that is free, that's clearly better for human flourishing, but it's like worse for stock market capitalization because no one can make a profit from it. And I mean-

Can we flourish if the stock market doesn't go higher? I don't know, Matt. I'm scared to find out the answer. It's a real question. There's a Simpsons bit where Homer has an auto dialer and it calls Mr. Burns and it says, hello, friend.

Would you trade $1 for eternal happiness? And Mr. Burns, thinking he's talking to a person, thinks for a minute and says, hmm. $1 for eternal happiness. I'd be happier with the dollar. I think about this all the time. Like, I sometimes feel that way too, right? Like, if your 401k goes down, but like all of your future needs will be met by like a little robot in your pocket, then you don't need the 401k, but like you want the 401k.

That's actually all I could think about on Monday was how I will never financially recover from this trading session. You're like weeping on air. But I was like, no, I need to get to the podcast. You're all in on the video. The show must go on. So you boldly asked the question in a Money Stuff this week, what if this Chinese quant is

hedge funds that, you know, then went on to develop DeepSeek, had a bunch of NVIDIA puts and made a bunch of money that way. And then friend of the show, Bill Ackman, tweeted the exact same question, which was fun. I assume that other people independently came to this idea. I had several readers email me about it before I wrote it. No, no, no. You were the first one to ever write about DeepSeek, first one to ever pose this hypothetical. That's true. And I should also say that the idea of like

disruptors funding their business by giving away their product for free and shorting incumbents. I learned of it from Joe Weisenthal, our fellow Bloomberg podcaster, like a decade ago. He was writing about it before everyone. But in any case, yeah, one, it's a really funny idea. Two, it has really kind of percolated up. Apparently, Howard Lutnick is getting questions about it in Congress this week. Oh, my God. Whether that's happened. There's no evidence for it, but it would be very funny if he did it. Also, yeah.

is it illegal? I mean, is it insider trading? So here's what I'll say. One, this is not legal advice. Two, I think in the U.S. it's like clearly fine. The way it works in the U.S. is like,

You are allowed to trade on your own information, but you're not allowed to misappropriate information from someone else. So if he was buying Nvidia puts in his personal account while running DeepSeek, then that might look bad depending on exactly what his arrangement was with DeepSeek. But if the corporate complex of DeepSeek and the hedge fund was trading on DeepSeek's own information...

to make a profit for that complex, I don't know, it seems fine. It seems like you're not misappropriating any information. You're really only trading on your own knowledge of your own business, right? And your own extrapolation of what that will mean for other businesses. I think it's totally fine. Now, the two caveats to that are one, I'm not sure that every jurisdiction sees it that way. US law is much more about this sort of misappropriation theory, whereas in other places it's more common to just be a fairness focus, like you can't trade on information that the market doesn't have.

And then two, if you did this, people would get mad and they would say,

say, oh, that's insider trading. And then you'd say, no, no, it's not insider trading. It's fine. They're like, okay, fine. But it's market manipulation. And like, it's market manipulation because they'll find something you said wrong. Right. And so here in deep secret, there's been a lot of controversy about like, is it really the case that like it's training costs was as low as it was, or was it like distilling models for other AI companies? You know, you find something that you can seize on to be like, oh, this is a misrepresentation. And then it's like market manipulation. You were like saying false things to bring down NVIDIA stock. Yeah.

So it would be a risky thing to do to tank NVIDIA stock in such a high profile way while owning puts on it. Like people would get mad and like might find a thing to charge you with. But like, no, I don't think it's insider trading. Yeah. Well, I mean, again, this is all hypothetical. We have no idea if he actually did this, but I don't think he did. I just like it's fun. He should have done it. I would have done it.

Because like the other thing is like they don't like I don't know what their business model is like public reporting is like some combination of oh he's just you know making money on this hedge fund or he's doing this out of like the goodness of his heart and like I want to want to contribute to AI research because it's like open source. It's cheap. How are they going to make money? I don't know. Like it's shortening video, right?

Yeah. I did also want to bring up that the narrative around this week has been that basically DeepSeek was able to recreate OpenAI with $6 million. And it's just been fun watching that get picked apart. I've spoken to so many NVIDIA bulls in the past couple of days who have made this into a bull case for NVIDIA, which I find really interesting. And there is a lot of skepticism around the numbers here. And one of the news stories that

came about is that Microsoft and OpenAI are taking a look at whether this group used OpenAI's API to basically get a large amount of data, which would violate OpenAI's terms of service. Because OpenAI, as we were reminded this week, isn't actually open source. It's closed source. But Meta's Lama is open source. Here's how I think about this. There are a lot of businesses that

that have been or look likely to be really badly disrupted by AI, right? There's some margin where if you're an accountant, let's say. Podcaster. Podcaster, yes. If you're a podcaster, they're coming for you. And in the next few years,

And AI will be able to sit here and talk into a microphone. I mean, not literally, but like we'll be able to generate voice in a way that is better than I can do for a fraction of my very high price. And therefore I will be out of work. Right. And this is true, like across a range of like sort of knowledge industries. Right. And so AI is like undercutting technology.

a lot of people's jobs in a way that increases abundance for the people who want to listen to podcasts or get accounting services, but that undercuts the earning potential of the people who are providing it. Cheap AI is to

AI, what AI is to humans, right? It's like OpenAI is like, ah, we can do your accounting for a tenth of the price of accountants. And now DeepSync is like, we can do it for a tenth of the price of OpenAI. It's great. But like the other interesting thing is like, you know, you talk about like, were they using OpenAI's APIs to train their model, to like distill OpenAI's model? You really like get a corpus of OpenAI outputs and you use that to train your model. Yeah.

We use it to refine the training of your model. That to me is a little bit analogous to the complaints that

publishers have about open AI using their data to train its models, right? AI synthesizes the output of like all these humans and then comes up with a thing that is, that they can produce that sort of output more cheaply than humans can. And they're like, wow, what? We were just like learning from humans. It's fine. And like, you know, DeepSeek kind of did that to open AI. Maybe, like arguably, like people are suspicious, which I think is

Violates the terms of service, but there's like a certain poetic justice to it. Well, DeepSeek for its part says that it has distilled models for R1 based on other open source systems, not necessarily on open AI, but...

Again, Meta's Llama is open source. It's freely available for use. And I don't know, maybe that's not a bad thing. They're all just building on each other. That doesn't seem terrible. Right. It's like there's this notion that if DeepSeek is piggybacking on the work of other models, then somehow the bear case against US AI infrastructure spending is weaker. But I'm not sure that's true, right? It's possible that it's just like the answer is that you've made it cheaper to scale AI because you can...

build impressively on prior work. And so you don't need all those data centers. I am curious what this means for the state of capex spending when it comes to all these big tech giants. And we do have a lot of tech giants reporting this week. And, you know, it's going to come up on the earnings calls, which haven't happened yet as we're recording this podcast.

We're like two days out from the US government being like, we're going to spend a trillion dollars on AI CapEx, right? Well, you think about last week. I don't know. People are so news-driven. You could imagine everyone being like, ah, we're just kidding about all the CapEx. You can imagine a really hard pivot in the next week. That doesn't seem that much. Well, there's a lot of...

Awkward timing here because you think about Stargate, which was announced last week with OpenAI, with SoftBank, $100 billion. Meta last week announced it was boosting its CapEx to up to $65 billion. Microsoft is spending $80 billion. It's just felt like this arms race to see who can spend the most money on this. And then to have everyone again get deep-seeked on Monday was...

As someone with a 401k, brutal, but pretty fun to watch. I'm sorry, this is rude, but it's such a SoftBank story. SoftBank announcing like RUNA has been $100 billion on data centers like 10 minutes before like that becomes an obsolete thesis. That's like, that's a good story. A good story.

GiveWell, a nonprofit that researches and recommends giving opportunities, takes the impact of donations seriously. To ensure their recommendations withstand tough scrutiny, GiveWell had their own researchers spend months trying to identify flaws in their past work. They then published their findings, mistakes and all, for any donors to use for their giving. It's this kind of rigor that can help your donation make a big impact on the world. GiveWell has now spent over 17 years researching charitable organizations

and only directs funding to a few of the highest impact opportunities they've found. Over 125,000 donors have used GiveWell to donate more than $2 billion. Rigorous evidence suggests that these donations will save over 200,000 lives. If you've never used GiveWell to donate, you can have your donations matched up to $100 before the end of the year or as long as matching funds last. To claim your match, go to GiveWell.org and pick Podcast.

This show is sponsored by BetterHelp. BetterHelp has been revolutionary in connecting people to mental health services. Using BetterHelp can be as easy as opening your laptop or your phone and clicking a button, and the session begins.

Clients are able to choose in what way they would like to communicate with me, whether video or on the phone or chat texting. BetterHelp is there when you need it, and that's what makes all the difference. Visit betterhelp.com slash podbusiness to get 10% off your first month. Therapists were compensated. Your customers are important to you, but they won't feel that way if they're stuck messaging a clunky chatbot or waiting on hold for a representative.

Estimated wait time is 25 minutes.

With Sierra, your company can deploy a branded AI agent that engages and delights customers anytime, anywhere. Sierra agents pick up every phone call and personalize every interaction. No more menus, no more hold times. And if you have an issue, Sierra's AI agents solve tough problems. Whether they're helping your customer pick out the person

Always friendly. Always helpful. Always ready. Visit sierra.ai to learn more. That's sierra.ai.

Speaking of ill-timed trades. Man, let's talk about X debt. It seems like it's getting a lot sweeter, potentially. Well, this is also weird. I never know how to interpret this. Because is this sweetener? So X, Twitter, X. The company formerly known as Twitter. Elon Musk bought it in...

2022. And when he signed his deal to buy it, for a variety of reasons, it looked like a better deal than it turned out to be like months later. And so all these banks led by Morgan Stanley agreed to provide him $13 billion of debt to buy Twitter. And by the time the deal closed, you know, ordinarily, like they would sell that debt to investors before the deal closed, but for a variety of reasons, mainly that he was suing to try to stop the deal from closing until the last minute. And so

and so wouldn't help out on the debt sale for a variety of reasons. They didn't sell the debt before the deal closed. And by the time the deal closed, it looked bad, both because Twitter's business had deteriorated and because Elon Musk deteriorated it further and because he was like talking smack about Twitter for the whole time he was trying not to buy it. And so like no one wanted the debt. They couldn't sell the debt and they,

There were occasional news stories being like they tried to sell the debt at like 60 cents on the dollar and they couldn't or whatever. Like they got bids at 60. Yeah. So like numbers like 60 cents on the dollar were floating around. And now they're apparently offloading some of the debt and like numbers like 90 to 95 cents on the dollar are floating around. So pretty good.

But that's like not necessarily apples to apples because it's like they're varying levels of seniority and it's possible that they couldn't sell the worst debt at 60 and now they're selling the best debt at 90, but still. So this XAI stake that Twitter has, I mean, I have questions. I was already questioning, you know, who knows on pricing, but is that worth, you know, up to 30 cents per bond or whatever? And whether it still is after, of course,

Deep Seek came about on Monday. After Elon Musk bought Twitter, the next thing quickly became large language models. Like when he signed the deal to buy Twitter, like no one was talking about open AI or whatever.

But shortly after the closing, like, large language models were the thing. And so he started an AI company called XAI. It has the same first letter as Twitter does now, which is X. And it clearly shared some resources with X. And, you know, he, like, made enough noise about doing AI out of X that...

The upshot is that XAI is not just owned by him personally and people who put money into XAI. And he's raised a lot of money for XAI. It's also partly owned by X, by Twitter. So that company that he bought, one of its assets is apparently...

a $6 billion stake in XAI. That's measured at its most recent valuation, which is like $50 billion. Now, maybe that valuation has gone way down since DeepSeek was released. There's not a lot of public market comps for pure play AI companies. So who knows what the valuation is of XAI or OpenAI or anything else.

But, you know, last time we checked, that stake was worth $6 billion, which sure is worth a lot of money to the debt because the debt is like $13 billion. And like, you know, if they have collateral, then, you know, that's worth, you know, 45 cents on the dollar. So I think that like there was reporting that X's banks were shopping the debt, specifically telling people, hey, you'd have a priority claim on all these XAI shares. Isn't that nice? And what I wrote about this is like,

It's very hard to do a credit analysis of X Twitter because...

If Twitter doesn't pay interest on its debt, the lenders can foreclose. But what good does that do them? Elon Musk can kind of trash Twitter on the way out the door. And Twitter seems like a hard company to run before Elon Musk bought it. And now it's gotten even harder. So it's a company that has a lot of potential upside, particularly for Elon Musk, who can use it to get presidents elected. But in terms of downside protection for lenders, it's like, eh, I don't know. Whereas like

Like, you know, again, a week ago saying we own a big stake in an AI company seems like really great downside protection because one, AI companies are valuable. Two, Elon Musk is like kind of unlikely to walk away from an AI company in a way that he tried to walk away from Twitter. And three, people do credit analysis of AI companies where they did, where they're like, oh, look, it has like all these like NVIDIA chips.

Even if like something goes horribly wrong at this company, those chips are super valuable because there's such an AI gold rush. So like there's really good collateral here. Right. Again, like that has been undermined by the events of the last week. But it does feel like the XAI stake was pretty credit enhancing for the Twitter bonds. Yeah. And I mean, you say that Elon Musk is less likely to walk away from the AI company in the same way he might be tempted to walk away from Twitter. I mean, how confident are you when you say that?

Because who really knows, Matt? Yeah, nobody knows anything. But like, Elon Musk is kind of always sad.

including when he was buying it, that Twitter was not a great business decision, but he thought it was important for the future of the world, blah, blah, blah, blah. And that, by the way, turned out to be right. He's had a ton of political influence with Twitter without it necessarily making him a lot of money. Again, a week ago, everyone was like, wow, AI is a real gusher of money, right? Who knows now, right? But I'm not saying personally I think there's no chance of him getting bored of AI. I think that was a reasonable thing for the market to think a week ago.

Yeah. You know what could be more fun than owning Twitter? Virtually anything. I don't know. I mean, I'm speaking specifically from Elon Musk's shoes. Perhaps owning TikTok, but that's a totally different conversation. Yeah, yeah. I wonder about that, right? If he buys TikTok, can you imagine him doing that separately? Like, that'd be so rude.

It didn't even occur to me because he's been like talked about as a potential buyer of TikTok. I assumed that he would do that out of the vehicle that is X. But I guess there's no law saying that. I mean, whatever. There are like standard views of like corporate, you know, fiduciary duties and corporate opportunities. We're like, yeah, it would be really weird of him to buy a competitor if he owns X. He could do that. And then he could own them separately. I was going to say, very bold of you to assume there are laws here. But go on. Yeah. Yeah.

Right. No, I started by saying there's no law that he can't do it. But of course there is, but there's not. Anyway, if he buys TikTok, the natural thing to do would be to have X buy TikTok. But that doesn't mean he'll do that. Yeah, I feel like TikTok is way more valuable than X, but what do I know? Yeah, but it's like a weird situation, right? Because you're buying the U.S. operations and you're buying it under the gun, right? You're buying it literally like Donald Trump is saying to TikTok,

Either you shut down or you sell it to my friend, right? Like how much leverage do you have to negotiate a price there? That's true. Something that was in the main bar story about this X debt with the XAI sweetener was just on X's annual interest expenses. And I knew that they had gone higher, but I didn't appreciate by how much. So...

With this debt that they were saddled with, the annual interest expense went from around $50 million to well over a billion dollars for X. That is gargantuan. Yeah, it's 13... You know, well, because they didn't have debt before because they were...

Yeah. They were like a not particularly lucrative public tech company. They were not running with a lot of debt. Elon Musk bought them. They were just a bird app. Yeah. Yeah. So he bought them with $13 billion of debt, right? Like 13 billion times, you know, 8% is a billion dollars. Yeah. I mean, you touched on it a little bit. Like Elon Musk could probably say I'm good for it. But so Twitter X theoretically is like barely breaking even. I hesitate to speculate. Like the Wall Street Journal reported a memo saying,

from Elon Musk saying that, but then he denied he wrote it. So I don't, I don't even know, man, but like, I don't know. Let's not speculate then. Okay. But right. I mean, like the other thing I was thinking about is like when Donald Trump was elected, there was a lot of writing to the effect of like, this investment looks a lot better for Elon Musk. Right. Because one, like Twitter X might actually be more valuable now because it's now like platform with a lot of political power. You can probably sell more ads, but two, like,

You know, it's hugely increased the value of the rest of his corporate empire, right? Because, like, he now runs a space company that, like, runs the government. He runs a car company that, you know, he has all these powers. So the investment looks really good for him. Does any of that help the debt?

I don't know. I think that Elon Musk being so close to the levers of power is probably a small negative for lenders. Banks don't like to lend to politically exposed persons, right? Because it makes you a worse credit, right? Because if you lend money to Donald Trump and like,

He doesn't pay you, then you foreclose on his property. But you lend money to Donald Trump and he becomes the president, then like you can't foreclose on him, right? So like it makes your credit a little bit worse. And I think there's something of that with Elon Musk where like on the one hand, he's more likely to be good for the money now that he's like so much richer and his businesses are so much better. But on the other hand, if he doesn't want to pay you, like there's not a lot you can do about it. So it's a mixed proposition for the lenders.

GiveWell, a nonprofit that researches and recommends giving opportunities, takes the impact of donations seriously. To ensure their recommendations withstand tough scrutiny, GiveWell had their own researchers spend months trying to identify flaws in their past work. They then published their findings, mistakes and all, for any donors to use for their giving. It's this kind of rigor that can help your donation make a big impact on the world. GiveWell has now spent over 17 years researching charitable organizations

and only directs funding to a few of the highest impact opportunities they've found. Over 125,000 donors have used GiveWell to donate more than $2 billion. Rigorous evidence suggests that these donations will save over 200,000 lives. If you've never used GiveWell to donate, you can have your donations matched up to $100 before the end of the year or as long as matching funds last. To claim your match, go to GiveWell.org and pick Bestowal.

and specify where you heard this ad. Make sure they know that you heard about GiveWell from this podcast. Your customers are important to you, but they won't feel that way if they're stuck messaging a clunky chatbot or waiting on hold for a representative. Estimated wait time is 25 minutes.

With Sierra, your company can deploy a branded AI agent that engages and delights customers anytime, anywhere. Sierra agents pick up every phone call and personalize every interaction. No more menus, no more hold times. And if you have an issue, Sierra's AI agents solve tough problems. Whether they're helping your customer pick out the person

Always friendly. Always helpful. Always ready. Visit sierra.ai to learn more. That's sierra.ai.

In 15th century Florence, the great inventor Leonardo da Vinci dreamt of creating a flying machine. But something kept getting in his way. Admin. Piles of it. Luckily, Leo used the smart buying tools on Amazon Business, so he could work more efficiently. With the extra time, he not only invented the flying machine, but actually built it. Magnifico. Incredibile. Splendido. Whoa, easy there, Leo. Splendido.

Amazon Business, your partner for smart business buying. I'm going to slam this laptop shut at 2.30, just so you know, no matter what happens. I have a horse to ride. I have miles to run in New Jersey.

Let's talk about dark pools. Okay, dark pools. So, according to Bloomberg News, most U.S. equity trading isn't done publicly anymore. Matt Levine, specifically off-exchange activity is on course to account for

a record 51.8% of traded volume in January. That would be the fifth monthly record in a row and the third month running that actually went off exchange volume was greater than half of all volume. Are you scared?

I wouldn't say I'm scared, but it is like a little bit like the index fund tipping point, right? Off exchange volume is sort of two exchanges, what like index funds are to active management, right? It's like you trade on the exchange, you produce value,

a public good, you produce information. The particular thing you produce is like stock prices, right? Like when you trade on an exchange, like you trade against a lit order and the order book on the exchange is public. People can see what the stock is trading for and you're producing information. If you trade off exchange, the off exchange mechanism, you know, if you're a hedge fund, it's called the dark pool. If you're an individual, it's called like a wholesaler or an internalizer. But the off exchange place where you trade,

is kind of free riding on the public exchanges, right? Like they're looking at the exchanges for pricing and then they're probably giving you a better price, right? There's probably some reason that you're on the off exchange venue and it's probably that you're getting a better price, right? If you're a retail trader, like it's pretty straightforward. People get mad about it, but like the idea of payment for order flow is that high frequency trading firm

doesn't take nearly as much risk trading with retail traders as they do trading with fancy hedge funds. And so they're willing to give you a better price. It's called the price improvement, right? If you're on a dark pool, you're an institution looking for a better price than you get on the exchange. And you might get that for some variety of reasons. One of which is just like the fees might be lower. And so, you know, you're trading in the dark to get a better price. And so you're not producing the sort of public good byproduct of that, which is like making prices for everyone else. And

If everyone does that, then it can get kind of weird. Bloomer's article about this quoted Larry Tabb saying that the more trading that goes off exchanges, the fewer orders there are on exchange competing to determine the best price. And this means the pricing on and off exchange could get worse, right? Like if there's no one publicly trading, then like no one knows what the price is. And so the off exchange pricing deteriorates. I don't think there's any reason to think that 50% is any sort of magical tipping point, but it's interesting. Yeah.

Yeah. And I'm glad you framed it as a public good. It's a public good that we would be trading on exchange. But you could get worried about price discovery, maybe not at 52 versus 48 percent. But it is kind of a fun thought exercise to imagine a world where you have 90 percent of trades happening in the dark, that their prices are based and extrapolated on the 10 percent that's happening on exchange. I don't know what that world looks like.

It's the same thought experiment as index funds, right? It's the same thing where people are like, you know, if 30% of the stock market is owned by index funds, that's fine. But 90% it gets weird, right? Like, you know, there's no real way to know what the magic number is, but it's the same idea of like, there is this like informational good and it's efficient for a lot of people to free ride on it. But at some point it becomes a problem.

Yeah. Well, I think there is an important caveat here, and this was mentioned lower down in the Bloomberg News article, which was written by Catherine Doherty. It's super good. So you're talking about dark pools when it comes to institutional investors and hedge funds. But when it comes to retail trading in penny stocks, if you strip out penny stocks from this data, according to Jeffries, then off-exchange trading volume remains below 40%.

which is less scary. But it is interesting that it's sort of both sides of the spectrum chipping away at on-exchange volume. It's like you have dark pools over here with all the fancy hedge funds, and then you have retail playing in penny stocks as well. You think of hedge funds as being sophisticated investors, and they are. But I think if you're a hedge fund, you often think of yourself as...

An innocent victim of the public stock market. I think at some time frame, like there are particular firms who are like,

good at getting the last penny of price on the stock exchange. And that's a particular skill set. Those firms are called high-frequency traders, right? And there are a lot of hedge funds who are really mad about high-frequency traders, who feel like they're being front-run by high-frequency traders, who feel like the public stock markets are an evil and predatory place. And you see this in Michael Lewis's book, Flash Boys, where big-time hedge fund managers are like,

He goes into their office and they're like, "Look at these high-friction traders. They're fleecing me." People get really mad. Not unsophisticated retail investors, like sophisticated investment managers whose timeframe is longer than five seconds. Those people feel like they're getting fleeced by the people whose timeframe is less than five seconds. Retail institutions are in the same boat here, where they both feel like the public markets are predatory, and so they want to find a more protected place where they won't be subject to the predators of the public markets.

And the more that happens, the more the public market is just full of predators. It's just full of the most sophisticated trading firms trading against each other and trying to make up off each other. And that makes them less and less appealing for both retail and for a hedge fund that has a fundamental investment thesis. Yeah.

Sorry, I'm still thinking about the comparison to passive versus active. And we did reach that tipping point where passive is now the majority of invested assets. Yeah, but it's fine. I mean, people say it's not fine. People get mad, right? I think there's a reasonable case that some aspects of capital allocation efficiency have been undermined by the rise of indexing. Again, not because 50% was the tipping point, but just because it's bigger than it used to be. But so far, it seems like there's a lot of competition to fall.

find the right price for securities. And there are a lot of hedge funds who make really quite a lot of money trying to make prices efficient. So it's not obvious that indexing has ended that. You could probably tell a similar story for the sort of microstructure level of public stock markets, where again, the trading firms that trade in public markets do pretty well. So it seems fine. Yeah.

And also like spreads are tight and everything like that. Yeah. Well, I don't know if passive will ever take up 90% of invested assets. And I'm not saying that we could get to 90% of trading happening in the dark, but it does seem, I mean, taking a look at, you know, the various people that already close in this piece that obviously this trend has been years in the making and maybe

Now we're at 52% roughly. Who knows where that goes? This is something that the SEC under Gary Gensler did try to address and try to push more of that trading to the public exchanges. And the wild card here in terms of the trajectory of where this goes is Paul Atkins. I don't know what his view is. He seems...

Like an interesting guy. He seems like a contrarian. I don't know where he lies on this, but it'll be fun to find out. I think some people in the industry would have comments about the idea that the Gensler SEC tried to push more trade into public exchange. That's true. But he also tried to wildly complicate aspects of public exchanges in ways that might have made them less appealing. But no, I think broadly speaking, that's true. And my guess is that in general, there is not a ton of incentive for the SEC to do a ton of

huge market structure overhauls, because while no one kind of loves US equity market structure, it seems to work fine and every potential change would be very complicated. And in fact, the Gensler SEC proposed pretty radical changes and got pushed back pretty hard and didn't end up actually doing anything. Not doing much in the very radical auctioning kind of ideas. And it's hard for me to imagine.

will surely be a very strange SEC being like, we need to reform dark pool trading. But maybe, who knows? I don't know. I would imagine it's not high on the priority list. Yeah, there is like a weird populist

appeal to saying, we're going to end payment for order flow and send all of your orders to the stock exchange. It's almost certainly bad for retail investors, but if you say it, people are like, oh yeah, that'll be good for retail investors. So there's some populist appeal to that as a platform, and this is arguably why Gary Gensler tried to do it. But you get bogged down very quickly in the weeds, and it's hard to actually do it. Well, to

Taking a look at just the ETF filings that are coming across, there's a lot of hope and dreams that Atkins is going to be a huge crypto bull. But I don't know. We'll see. Well, that seems true. But does it? I don't know. I remember you think about Gary Gensler. People are like, oh, my God, he taught an MIT course on the blockchain. And then turns out he turned into like enemy number one for the crypto crowd. I just think that like.

Paul Atkins has orders from the top to be crypto-friendly. And I also think, you know, you talk about ETF filings, like you could file an ETF for any meme coin in the world. And people are. People are. But in particular, they're filing for like Trump coin and Melania coin. Now, is the SEC going to say, well, these tokens are too subject to manipulation and like don't have a real investment thesis. So we can't approve an ETF on them? No.

Well, well, actually. Are you slamming your laptop closed? Wait, hold on. Okay, here we go. Actually, to that point, there were filings for double leverage Melania and Trump ETFs, but they. Come on. They were withdrawn. That was a joke.

What do you mean? I mean, like, I wish I had filed a double leverage of Milani ETF as a joke because, like, that's very funny. But, like, imagine trying to launch that as a product. I don't think this issuer was joking. I don't think this issuer was joking. You're probably right. But I'm going to just choose to believe that the issuer was joking. Yeah. Well, assuming that they weren't, and I'm pretty sure that they weren't joking, they did withdraw the filing. Well, then they were joking. It's fine. Problem solved. No, they definitely weren't joking. Anyway, that suggests...

they got a call from the SEC that said, maybe this is too far. So we'll find out where the lines are. I am not convinced that in the next four years we'll find out where any lines are. Wow. I suspect that like, you'll be like, is the line over there? Like, nope, not here. A lineless world.

And that was the Money Stuff Podcast. I'm Matt Levine. And I'm Katie Greifeld. You can find my work by subscribing to the Money Stuff newsletter on Bloomberg.com. And you can find me on Bloomberg TV every day on open interest between 9 to 11 a.m. Eastern. We'd love to hear from you. You can send an email to moneypod at Bloomberg.net.

Ask us a question and we might answer it on air. You can also subscribe to our show wherever you're listening right now and leave us a review. It helps more people find the show. The Money Stuff Podcast is produced by Anna Masarakis and Moses Andon. Our theme music was composed by Blake Maples. Brendan Francis-Nunen is our executive producer. And Sage Bauman is Bloomberg's head of podcasts. Thanks for listening to The Money Stuff Podcast. We'll be back next week with more stuff.