I apologize for the length, but this is yesterday's newsletter from Matt Levine -- the best financial writer in the business.
Matt Levine wrote:
There is so much, but I want to start with Serum.
The Box
If a troubled company has a few days to beg potential investors for a bailout before it files for bankruptcy, and it sends those investors its balance sheet so they can consider investing, and they all pass, and then the company files for bankruptcy, of course the balance sheet was bad. That is not a state of affairs that is consistent with a pristine fortress balance sheet.
But there is a range of possible badness, even in bankruptcy, and the balance sheet that Sam Bankman-Fried’s failed crypto exchange FTX.com sent to potential investors last week before filing for bankruptcy on Friday is very bad. It’s an Excel file full of the howling of ghosts and the shrieking of tortured souls. If you look too long at that spreadsheet, you will go insane. Antoine Gara, Kadhim Shubber and Joshua Oliver at the Financial Times reported on Saturday:
Sam Bankman-Fried’s main international FTX exchange held just $900mn in easily sellable assets against $9bn of liabilities the day before it collapsed into bankruptcy, according to investment materials seen by the Financial Times.
The largest portion of those liquid assets listed on a FTX international balance sheet dated Thursday was $470mn of Robinhood shares owned by a Bankman-Fried vehicle not listed in Friday’s bankruptcy filing, which included 134 corporate entities.
Seems bad, but it somehow keeps getting worse:
A spreadsheet listing FTX international’s assets and liabilities, seen by the Financial Times, point at the issues that brought Bankman-Fried crashing back down to earth. It references $5bn of withdrawals last Sunday, and a negative $8bn entry described as “hidden, poorly internally labled ‘fiat@’ account”.
What.
The vast majority of FTX Trading’s recorded assets are either illiquid venture capital investments or crypto tokens that are not widely traded, according to the spreadsheet, which cautions that the figures “are rough values, and could be slightly off; there is also obviously a chance of typos etc. They also change a bit over time as trades happen.”
What.
In all, the spreadsheet says FTX Trading’s assets were $900mn of “liquid” assets, $5.5bn of “less liquid” assets consisting of crypto tokens, and $3.2bn of illiquid private equity investments. There is also an obscure $7mn holding called “TRUMPLOSE”. There are no bitcoin assets listed, despite bitcoin liabilities of $1.4bn.
What. And yet bad as all of this is, it can’t prepare you for the balance sheet itself, published by FT Alphaville, which is less a balance sheet and more a list of some tickers interspersed with hasty apologies. If you blithely add up the “liquid,” “less liquid” and “illiquid” assets, at their “deliverable” value as of Thursday, and subtract the liabilities, you do get a positive net equity of about $700 million. (Roughly $9.6 billion of assets versus $8.9 billion of liabilities.) But then there is the “Hidden, poorly internally labeled ‘fiat@’ account,” with a balance of negative $8 billion. I don’t actually think that you’re supposed to subtract that number from net equity — though I do not know how this balance sheet is supposed to work! — but it doesn’t matter. If you try to calculate the equity of a balance sheet with an entry for HIDDEN POORLY INTERNALLY LABELED ACCOUNT, Microsoft Clippy will appear before you in the flesh, bloodshot and staggering, with a knife in his little paper-clip hand, saying “just what do you think you’re doing Dave?” You cannot apply ordinary arithmetic to numbers in a cell labeled “HIDDEN POORLY INTERNALLY LABELED ACCOUNT.” The result of adding or subtracting those numbers with ordinary numbers is not a number; it is prison.
But here is the paragraph that drove me insane:
The company’s biggest asset as of Thursday was $2.2bn worth of a cryptocurrency called Serum. Serum’s market value was $88mn on Saturday, according to data provider CryptoCompare, suggesting FTX’s holdings would be worth far less if sold into the market. CryptoCompare’s figures take into account the coin’s liquidity.
If you look on the balance sheet, that’s the biggest “deliverable” number: $2,187,876,172 of SRM, the ticker for the Serum token, down from $5,430,110,335 “before this week.” (“Before this week” means, of course, before last week — before the week that FTX was circulating this spreadsheet — and I will use it as a shorthand to mean roughly “before Nov. 8.” On Nov. 8, the trouble at FTX became public, and everything FTX-related crashed.)
As of about 11 a.m. today, CoinMarketCap showed the Serum token having a price of $0.25, a “market cap” of about $65 million and a “fully diluted market cap” of about $2.5 billion. Those numbers would have been a bit higher — say $0.35 to $0.40 per token — on Thursday, when this balance sheet was created. The price crashed last Tuesday, during FTX’s death throes; before that the token traded around $0.80. In round numbers FTX probably held something like two-thirds of Serum’s fully diluted market cap, and roughly 20 times its basic market cap.
In crypto, market cap is (as CoinMarketCap puts it) “the total market value of a cryptocurrency's circulating supply … analogous to the free-float capitalization in the stock market,” while fully diluted market cap is “the market cap if the max supply was in circulation.” So if for instance some company creates a token, and says that there can be 10 billion of the token, and reserves them all for itself, and then sells 1 million of them to outside investors for $1 each, then the market cap of that token is $1 million ($1 times 1 million circulating tokens), while the fully diluted market cap is $10 billion ($1 times 10 billion total tokens), and the issuer’s 9,999,000,000 remaining tokens have a value, on this math, of $9.999 billion. We will come back to this point.
What is Serum? Serum is a “protocol for decentralized exchanges that brings unprecedented speed and low transaction costs to decentralized finance” that runs on the Solana blockchain. Also Serum (the token, ticker SRM) “is the utility and governance token of Serum (the protocol). If you hold SRM in your wallet, you receive a discount on fees” for trading on the Serum protocol. Also, when the protocol collects trading fees, it uses a portion of the fees to buy and burn SRM. The result is that SRM functions a lot like stock in Serum: If the Serum project does well and a lot of decentralized trading happens on its exchanges, then it will collect a lot of fees and use those fees to buy SRM, which will drive up the value of SRM and make SRM investors rich. (Also the SRM investors can vote on how Serum is run.) If you are bullish on Serum as a business, as a platform for decentralized crypto trading, then you should buy SRM, because SRM is more or less a claim on the cash flows of that business.
One thing that is really really really really really important to mention about the Serum protocol is that it was created and promoted by FTX and Alameda Research, the FTX-affiliated crypto hedge fund that was also founded by Bankman-Fried. FTX is a centralized crypto exchange, but a lot of people in crypto do not trust centralized exchanges (for reasons!) and prefer to trade on decentralized exchanges. Serum is, in a loose but meaningful way, the decentralized exchange of FTX.
Go back to what I said above, about a company creating a token, issuing a bunch of it to itself, and selling a little of it into the public market. Something like 3% of the total value of Serum is held by the public and trading on exchanges. The other 97% is not. Something like two-thirds of that 97% is held by FTX and Alameda.
How did they get their SRM? Well, you can look at the distribution of SRM here, but the main point is that they did not buy it on the open market for cash. When FTX’s balance sheet says that, “before this week,” it held $5.4 billion worth of Serum, that does not mean that Alameda or FTX took $5.4 billion of cash (their own, their investors’, their customers’, anyone’s) and used it to buy a lot of Serum tokens in the open market. They got all those Serum tokens for, in round numbers, free, as the initial backers of the Serum protocol. (Presumably they paid some startup costs.)
For a minute, ignore this nightmare balance sheet, and think about what FTX’s balance sheet should be. Conceptually, customers give you money — apparently about $16 billion in dollars, crypto, etc. — and then you hang on to the money and owe it back to them. In the simplest world, you keep the customers’ money in exactly the form they give it to you: Someone deposits $100, you keep $100 for him; someone deposits one Bitcoin, you keep one Bitcoin for her. For reasons we have discussed— some legitimate! — FTX doesn’t quite work that way, and you could imagine some more complicated balance sheet where a lot of the money and crypto that came in from some customers was loaned to others. But broadly speaking your balance sheet is still going to look roughly like:
Liabilities: Money customers gave you, which you owe to them;
Assets: Stuff you bought with that money.
And then the basic question is, how bad is the mismatch. Like, $16 billion of dollar liabilities and $16 billion of liquid dollar-denominated assets? Sure, great. $16 billion of dollar liabilities and $16 billion worth of Bitcoin assets? Not ideal, incredibly risky, but in some broad sense understandable. $16 billion of dollar liabilities and assets consisting entirely of some magic beans that you bought in the market for $16 billion? Very bad. $16 billion of dollar liabilities and assets consisting mostly of some magic beans that you invented yourself and acquired for zero dollars? WHAT? Never mind the valuation of the beans; where did the money go? What happened to the $16 billion? Spending $5 billion of customer money on Serum would have been horrible, but FTX didn’t do that, and couldn’t have, because there wasn’t $5 billion of Serum available to buy. FTX shot its customer money into some still-unexplained reaches of the astral plane and was like “well we do have $5 billion of this Serum token we made up, that’s something?” No it isn’t!
One simple point here is that FTX’s Serum holdings — $2.2 billion last week, $5.4 billion before that — could not have been sold for anything like $2.2 billion. FTX’s Serum holdings were vastly larger than the entire circulating supply of Serum. If FTX had attempted to sell them into the market over the course of a week or month or year, it would have swamped the market and crashed the price. Perhaps it could have gotten a few hundred million dollars for them. But I think a realistic valuation of that huge stash of Serum would be closer to zero. That is not a comment on Serum; it’s a comment on the size of the stash.
But I do want to comment on Serum, because Serum is not some weird token that FTX cornered for some reason; Serum is a token that FTX made up. To use a loose but reasonable analogy, Serum (the protocol) is sort of FTX’s decentralized exchange subsidiary, and SRM (the token) is sort of the stock in that subsidiary. A little of the stock trades publicly, but it is mostly held by FTX, its corporate parent, as it were. The public market price of the small free float might give a reasonable estimate of the value of the subsidiary. But in the real world, the value of the subsidiary is incredibly tightly linked to the value of FTX’s overall business. If everyone is like “ah yes FTX is a good exchange operator and a leader in safe crypto trading,” then its decentralized exchange protocol has a good chance of being popular and profitable. If everyone is like “ah yes FTX is a careless fraud,” then Serum is going to have a hard time. At the point that FTX is shopping its Serum stake to seek a rescue financing due to HIDDEN POORLY INTERNALLY LABELED ACCOUNT, its huge stash of Serum is toast! Just toast!
Does this sound familiar? This is pretty much exactly what I said last week about FTT, the utility token of FTX’s regular centralized exchange. I wrote:
FTX issues a token called FTT. The attributes of this token are, like, it entitles you to some discounts and stuff, but the main attribute is that FTX periodically uses a portion of its profits to buy back FTT tokens. This makes FTT kind of like stock in FTX: The higher FTX’s profits are, the higher the price of FTT will be. It is not actually stock in FTX — in fact FTX is a company and has stock and venture capitalists bought it, etc. — but it is a lot like stock in FTX. FTT is a bet on FTX’s future profits.
I was writing about reports suggesting that FTX might have loaned a lot of customer money to Alameda and taken Alameda’s stash of FTT tokens as collateral, and I said:
If you think of the token as “more or less stock,” and you think of a crypto exchange as a securities broker-dealer, this is completely insane. If you go to an investment bank and say “lend me $1 billion, and I will post $2 billion of your stock as collateral,” you are messing with very dark magic and they will say no. The problem with this is that it is wrong-way risk. (It is also, at least sometimes, illegal.) If people start to worry about the investment bank’s financial health, its stock will go down, which means that its collateral will be less valuable, which means that its financial health will get worse, which means that its stock will go down, etc. It is a death spiral.
Last week I was shocked that one of the main assets of FTX — one of the main assets it relied on to be able to pay out customer balances — was a token it had just made up. But I was wrong! It was two tokens that it had just made up! FTX’s two largest asset balances, “before this week,” were $5.9 billion of FTT ($553 million at post-crash prices last Thursday) and $5.4 billion of SRM ($2.2 billion post-crash). Something like two-thirds of the money that FTX owed to customers was backed by its own tokens that it had made up.
The third-biggest asset, incidentally, was SOL, the token of the Solana blockchain. Solana is notsomething that FTX made up, and has an existence independent of FTX. But it is certainly associated withAlameda, FTX and Sam Bankman-Fried; they have been big backers of the Solana ecosystem. It’s not that Solana is “the blockchain of FTX,” but it’s a little bit like that. There is wrong-way risk there too.
And another big asset is $616 million worth of the MAPS token ($865 million “before last week”). MAPS is the token of Maps.me 2.0, a sort of Serum spinoff that was also launched by FTX; its market cap, according to CoinMarketCap at about 11 a.m. today, is about $3 million. Again, FTX’s MAPS holdings were two hundred times the total value of MAPS actually tradable in the market. It’s the same story as Serum, but worse, though on a smaller scale. (There’s a lot of this all over the balance sheet: Bloomberg’s Annie Massa reported on these projects today, under the headline “Sam Bankman-Fried’s Magic Money Box Enriched Vast Crypto Network.”)
In round numbers, FTX’s Thursday desperation balance sheet shows about $8.9 billion of customer liabilities against assets with a value of roughly $19.6 billion before last week’s crash, and roughly $9.6 billion after the crash (as of Thursday, per FTX’s numbers). Of that $19.6 billion of assets back in the good times, some $14.4 billion was in more-or-less FTX-associated tokens (FTT, SRM, SOL, MAPS). Only about $5.2 billion of assets — against $8.9 billion of customer liabilities — was in more-or-less normal financial stuff. (And even that was mostly in illiquid venture investments; only about $1 billion was in liquid cash, stock and cryptocurrencies — and half of that was Robinhood stock.) After the run on FTX, the FTX-associated stuff, predictably, crashed. The Thursday balance sheet valued the FTT, SRM, SOL and MAPS holdings at a combined $4.3 billion, and that number is still way too high.
I am not saying that all of FTX’s assets were made up. That desperation balance sheet lists dollar and yen accounts, stablecoins, unaffiliated cryptocurrencies, equities, venture investments, etc., all things that were not created or controlled by FTX. And that desperation balance sheet reflects FTX’s position after $5 billion of customer outflows last weekend; presumably FTX burned through its more liquid normal stuff (Bitcoin, dollars, etc.) to meet those withdrawals, so what was left was the weirdo cats and dogs. Still it is striking that the balance sheet that FTX circulated to potential rescuers consisted mostly of stuff it made up. Its balance sheet consisted mostly of stuff it made up! Stuff it made up! You can’t do that! That’s not how balance sheets work! That’s not how anything works!
Oh, fine: It is how crypto works. This might all sound familiar not just because we talked about FTT last week, but because we talked about the collapse of TerraUSD and Luna earlier this year. Terra was a blockchain system run by Do Kwon, and it raised billions of dollars by selling dollar-denominated tokens — TerraUSD — that were supposed to keep their value because they were backed by a variable amount of another token — Luna — that Kwon had also invented. For a while people thought the Terra ecosystem was promising, so the Luna token was worth a lot, so Terra could go around saying its TerraUSD tokens were extremely safe, because the billions of dollars of TerraUSD “debt” were backed by more billions of dollars’ worth of Luna. And then one day people changed their minds, and the price of Luna — which was just a bet on Terra’s future — collapsed, so TerraUSD was unbacked, and the whole thing collapsed. The FTX situation is not the same, but it rhymes. The role of TerraUSD — the “debt” — is played here by FTX’s customer balances; the role of Luna — the backing token — is played by FTT and SRM. In both cases, confidence in the business collapsed, and it turned out that the debt was actually backed by nothing.
Or it might sound familiar because Bankman-Fried said it himself, to me, on a now-infamous episode of Bloomberg’s Odd Lots podcast last year. I asked him a question about yield farming, and in the course of his answer he said:
You start with a company that builds a box and in practice this box, they probably dress it up to look like a life-changing, you know, world-altering protocol that's gonna replace all the big banks in 38 days or whatever. Maybe for now actually ignore what it does or pretend it does literally nothing. It's just a box. So what this protocol is, it's called ‘Protocol X,’ it's a box, and you take a token. …
So you've got this box and it’s kind of dumb, but like what's the end game, right? This box is worth zero obviously. … But on the other hand, if everyone kind of now thinks that this box token is worth about a billion dollar market cap, that's what people are pricing it at and sort of has that market cap. Everyone's gonna mark to market. In fact, you can even finance this, right? You put X token in a borrow lending protocol and borrow dollars with it. If you think it's worth like less than two thirds of that, you could even just like put some in there, take the dollars out. Never, you know, give the dollars back. You just get liquidated eventually. And it is sort of like real monetizable stuff in some senses.
The box, it turns out, was FTX (and Serum). It looked like a life-changing, world-altering business that would replace all the banks. It had a token, FTT (and SRM), with a multibillion-dollar market cap. You could even finance it, or FTX/Alameda could anyway: They could put FTT (and SRM) tokens in a box and get money out. (From customers.) They could take the dollars out and never, you know, give the dollars back. They just got liquidated eventually. And those tokens, FTT and SRM, were sort of like real monetizable stuff in some senses. But in other senses, not.
But where did it go?
I tried, in the previous section, to capture the horrors of FTX’s balance sheet as it spiraled into bankruptcy. But, as I said, there is something important missing in that account. What’s missing is the money. What’s missing is that FTX had at some point something like $16 billion of customer money, but most of its assets turned out to be tokens that it made up. It did not pay $16 billion for those tokens, or even $1 billion, probably. Money came in, but then when customers came to FTX and pried open the doors of the safe, all they found were cobwebs and Serum. Where did the money go?
I don’t know, but the leading story appears to be that FTX gave the money to Alameda, and Alameda lost it. I am not sure about the order of operations here. The most sensible explanation is that Alameda lost the money first— during the crypto-market meltdown of this spring and summer, when markets were crazy and Alameda spent money propping up other failing crypto firms — and then FTX transferred customer money to prop up Alameda. And Alameda never made the money back, and eventually everyone noticed that it was gone.
So Reuters reported last week:
At least $1 billion of customer funds have vanished from collapsed crypto exchange FTX, according to two people familiar with the matter. The exchange's founder Sam Bankman-Fried secretly transferred $10 billion of customer funds from FTX to Bankman-Fried's trading company Alameda Research, the people told Reuters.
A large portion of that total has since disappeared, they said.
And the Wall Street Journal reported over the weekend:
Alameda Research’s chief executive and senior FTX officials knew that FTX had lent its customers’ money to Alameda to help it meet its liabilities, according to people familiar with the matter. ...
Alameda faced a barrage of demands from lenders after crypto hedge fund Three Arrows Capital collapsed in June, creating losses for crypto brokers such as Voyager Digital Ltd., the people said.
In a video meeting with Alameda employees late Wednesday Hong Kong time, Alameda CEO Caroline Ellison said that she, Mr. Bankman-Fried and two other FTX executives, Nishad Singh and Gary Wang, were aware of the decision to send customer funds to Alameda, according to people familiar with the video. …
Ms. Ellison said on the call that FTX used customer money to help Alameda meet its liabilities, the people said.
Alameda had taken out loans to fund illiquid venture investments, the people said.
Here we are in the realm of pure speculation, but you could imagine a number of ways this could have gone:
1. Crypto prices, and firms, crashed earlier this year, and Alameda spotted a huge opportunity. It deployed as much capital as it could into buying great assets at bargain-basement prices. But since there was a crypto winter, it couldn’t deploy all that much capital, and was getting calls from its own lenders. So Ellison and Bankman-Fried conferred and sensibly decided that they couldn’t miss this opportunity, and that they would deploy FTX customers’ money against it. They’d make a fortune in short order on can’t-lose trades, and pay back the customer funds with interest. Then, oops, they were wrong. This story is bad — none of these stories are going to be good! — but understandable. If you run an opaque business in a lightly regulated industry, and customers trust you with their money, and you use it to make what you think are good bets, and those bets turn out wrong, well, that happens sometimes.
2. Crypto prices crashed earlier this year, and Alameda was caught out. It lost money and was facing calls from its lenders. Ellison and Bankman-Fried realized that Alameda would go under without help, so they took FTX customer money to prop up Alameda, and gambled on redemption. This story is not so different from the previous one, though it is worse, but also very understandable. It is the typical way these things go, the default assumption for why someone would use customer money. No one wants to fail, no one wants to admit that they lost money, and if there’s a poorly guarded pot of money they can use to paper over losses, sometimes they will.
3. Crypto prices, and firms, crashed earlier this year, and FTX/Alameda were like “we are in a confidence business, and if we let these firms crash then investors will lose confidence in crypto exchanges, which is bad for our business.” Either in a good, public-spirited, we-want-crypto-to-thrive way, or in a bad, we-need-suckers way, or a bit of both. So they bailed those firms out with customer money. Here is a video of Bankman-Fried and me discussing this possibility at the Bloomberg Crypto Summit in July, in which he said: “The explicit working principle we had” in these bailouts was that “we are incinerating a relatively small-ish amount of money in doing this,” in order to keep the crypto ecosystem healthy. Alameda/FTX was willing to lose money bailing out other firms, if doing so improved confidence in crypto generally. Of course we did not talk about the possibility that FTX was doing this with customer money.
4. Crypto prices, and firms, crashed earlier this year, and FTX/Alameda spotted an opportunity to acquire new customer deposits cheaply and use them for nefarious purposes. Like, you pay zero dollars for the equity of some busted crypto lending platform, you roll the customers over to become FTX customers, you cash out anyone who wants to cash out, you assume that most people will trust FTX (their savior) and not cash out, and then you use their deposits to fund your wild speculations. If FTX/Alameda were already using customer deposits for bad reasons, and losing them, then acquiring more customer deposits would be a way to keep things going.
5. FTX/Alameda were funneling customer money into lavish lifestyles for their executives. This one does not seem likely here — they slept on beanbag chairs in the office, etc. — but it is in general a very common explanation of missing customer money, and you’d want some accounting.
6. FTX/Alameda were funneling customer money into effective altruism. Bankman-Fried seems to have generously funded a lot of effective altruism charities, artificial-intelligence and pandemic research, Democratic political candidates, etc. One $500 million entry on the desperation balance sheet is “Anthropic,” a venture investment in an AI safety company. At that same Bloomberg Crypto Summit, I asked Bankman-Fried: “You are sort of in the business of funneling money from people who are going to use it poorly on gambling to, like, animal charities and pandemic preparedness and Joe Biden. Is that too cynical a view, or is that not cynical at all, or what?” My question assumed that FTX and Alameda made a lot of money on fees and spreads from running a crypto exchange and market-maker, so they were legitimately taking money from gamblers and using it for charity. But “not cynical enough” might have been the correct answer.