The basic deal is that if you found a company, and you want to raise money, you will probably have to give up some ownership of it. When the company is just you and an idea in your garage, you own the whole company. But if the company needs money to build factories or what have you, the people who give you the money will want something in return. Often what they will want is stock. If you hire employees, they will want stock. By the time your company is a big thriving public company, you might own only 10% or 20% of it.

By that point, you might never again need to sell stock to raise money to do the company’s business. It’s thriving, it’s bringing in money, and if it is like the typical big US public company over the last few decades, particularly in the tech industry, it is bringing in more money than it needs to spend on growth. Until very recently, your company was more likely to be buying back stock than selling it.

On the other hand, you might want money, for yourself. Maybe your yacht is getting a bit dingy, or you want to buy all the mansions around your mansion so no one can get too close to you. Probably you don’t have much liquid wealth other than the stock. So if you need a few hundred million dollars for yachts or mansions, the natural thing to do is to sell stock.

But if you sell stock, you will own less of your company. You might, after founding the company and running it for many years and having your identity bound up with it, not like that. Well, tough. The thing that you have that has economic value is ownership of the company, and if you want to turn that into money you have to give up ownership. (Fine: You can borrow against your stock, which is not quite giving up ownership, though it creates some risk that you might have to.) 

There is a traditional compromise, which is that you give up some ownership but not control. Normally this is done with dual-class stock: Your shares get 10 votes per share, everyone else’s shares get 1 vote per share, so you end up with, perhaps, about 13% of the economic ownership of the company but about 60% of the votes. That way, you can pick the board of directors and run the company how you want, even while raising money — for yourself and for the company — from outside investors. If you sell some of your high-vote stock, it converts into low-vote stock, so you get money without giving up much control. You have to get your investors to agree to this, of course, and generally before you go public, but there is a long tradition of tech investors being fine with, and in fact liking, giving founders extra votes.

This more or less works, most of the time. If you own 13% of the company, but your shares all have 10 votes and everyone else’s shares have 1 vote, then (1) you have about 60% of the vote and (2) you can sell a bunch of stock and still keep control. For instance, you can sell 3% of the company; that will take your economic stake down to 10%, but will still leave you with 52.6% of the vote.^[1] Or the company can raise money by selling 10% more stock; that will take your economic stake down to about 11.8% but leave you with 57.3% of the vote.^[2] Plenty of room to maneuver.

But not unlimited room to maneuver. If you sell half your stock, or if the company doubles its shares outstanding, you will lose voting control; if you own 6.5% of the stock and have 10 votes per share, you have about 41% of the votes.^[3] Having 10 votes per share helps you keep control of your company even while (1) it raises money and (2) you raise money, but only over some reasonable range. At some point, tons and tons of one-vote shares can outvote your diminishing stash of 10-vote shares.

The obvious solution to this is to give everyone else zero-vote shares. You get 10 votes per share, everyone else gets zero votes per share, and you can sell as many shares as you like without diluting your voting power at all. As long as you keep one share, everyone else can own umpteen quintillion shares and you can still outvote them.

Some companies actually do this, but it is sort of … awkward? Shareholders like to have voting rights, and they like their founder-controllers to have some skin in the game. “Sure,” investors say, “you should get more votes than us, because this company is your baby and you are the charismatic visionary who should lead it, even after you sell some stock to buy mansions. But not infinitely more votes, and not if you sell virtually all of your stock. You should control the company only as long as you hang on to a big chunk of stock.”

There is no actual magic to the 10-to-1 ratio, and some companies have other ratios. But 10 to 1 is pretty traditional. It’s a nice round number and it produces reasonably intuitive results. The founder needs to keep about 10% of the economic ownership to keep majority voting control. If the founder still owns 10% of the company, that’s a big chunk; she still has a lot of her wealth tied up in the company and good incentives to come into work and make good decisions with her voting control. That’s just the usual deal at big tech companies, the usual amount of economic ownership the founder has to maintain if she wants to keep voting control.

But times are changing:

Meta Platforms is getting creative to fund its massive investments in artificial intelligence.

The company reported record revenue for the third quarter and said it would create a new class of nonvoting stock to raise the capital it needs for developing and serving AI models. The new shares will be issued as a dividend to holders of common and supervoting shares, with an effective date of Jan. 19, 2027. The effective 2-for-1 stock split won’t affect Chief Executive Mark Zuckerberg’s voting control or the voting power of other shareholders, the company said.

In a filing, Meta said Zuckerberg would be able to immediately sell 20% of his new nonvoting shares, potentially helping him pay a new wealth tax on California’s upcoming ballot. The company said Zuckerberg faced a possible diminishment of his voting interest if the tax forced him to sell shares, and that he had discussed the possibility of relocating. Zuckerberg has committed to living in the same state as Meta’s headquarters for at least eight years barring unforeseen circumstances, the filing said.

Here are Meta’s announcement, and its 10-Q, and the information statement for the new share class. The way it works is:

1. Right now, Zuckerberg owns about 13% of the stock and about 61% of the votes: He owns almost all of the Class B shares, which have 10 votes per share; everyone else owns Class A shares, which have one vote per share.
2. If he sells down stock from time to time to buy mansions, sure, no problem; he’s got plenty of room. But “if California’s Proposition 40, a statewide ballot measure on the November 3, 2026 general election ballot and commonly referred to as the ‘California wealth tax,’ is enacted as law in California, it is expected that Mr. Zuckerberg could become liable for a significant, multi-billion dollar tax payment based on his ownership of stock in the Company,” and he’d have to sell a lot of stock to pay it.
3. Similarly, if Meta needs to issue a little stock from time to time to pay employees, sure, no problem; Zuckerberg has plenty of room. But if it needs to issue hundreds of billions of dollars of stock to fund its $825 billion of AI spending commitments, that might dilute Zuckerberg’s voting control unacceptably. 
4. Or the combination: If he’s selling stock to pay taxes, and Meta’s selling stock to pay for data centers, then there’s a real risk of him losing control.
5. The solution is a quasi-share split: Meta is creating a new Class C stock, with zero votes per share. Shareholders will get one no-vote Class C share for every one-vote Class A share or 10-vote Class B share that they currently own. Zuckerberg, who owns about 341 million high-vote Class B shares (and like 1 million low-vote shares), will get about 342 million new zero-vote shares; all the other shareholders, who own about 3.4 billion low-vote shares, will get 3.4 billion new no-vote shares.
6. Everyone implicitly assumes that the no-vote Class C shares will be worth about as much as the low-vote Class A shares. (Alphabet Inc. has a similar structure, and its no-vote shares currently trade at about a 1% discount to its low-vote shares.)
7. If Zuckerberg needs to pay taxes, he can sell his new zero-vote shares to raise money without diluting his voting rights. Not too many: He can “immediately sell or transfer up to 20% of the shares of Class C common stock that he receives in the Dividend, representing 10% of his overall economic stake in the Company, for any reason,” but the board “negotiated for and obtained Mr. Zuckerberg’s agreement that 80% of the shares of Class C common stock that he receives in the Dividend be subject to the Stapling Condition,” under which he can’t sell those no-vote shares without also selling some high-vote shares. The basic deal remains intact — he needs to own a meaningful chunk of shares to keep voting control — but there’s a liquidity buffer for taxes.
8. Meanwhile, if Meta needs to raise money to pay for data centers, it can sell zero-vote shares to its heart’s content, raising money without diluting Zuckerberg.
9. Technically all of this requires a shareholder vote, because Meta is amending its charter to create this new non-voting stock. But because Zuckerberg controls 61% of the vote, that’s fine! He already voted for it and it’s a done deal; outside shareholders don’t even get to vote, because their votes can’t change the outcome.^[4]

Fine. Again, Alphabet did something like this, ages ago, back when it was called Google. In fact, Meta tried something like this, ages ago, back when it was called Facebook. Back in 2016, it announced a similar distribution of Class C non-voting stock, for the same basic reason: It wanted Zuckerberg to be able to keep voting control even while selling down some stock. (So that he could give stock to charity.) This ended up not happening, though. Back in 2016, Facebook made the deal contingent on the approval of outside shareholders, the outside shareholders objected (and sued), and Facebook ultimately backed down.

One way to look at this is that, back in 2016, there was no pressing reason to do this. It was a forward-thinking conjectural nice-to-have: Zuckerberg thought that, over time, he might want to sell or give away a lot of stock for charitable and/or mansion reasons, so he wanted to set up a structure allowing him to do that without losing control. Shareholders rebelled: That’s not the deal they signed up to; they wanted Zuckerberg to keep the usual amount of ownership if he wanted to keep voting control. The counterarguments were weak, mostly just “this is what he wants,” and did not prevail.

In 2026, there is a reason to do this, or rather two. One is that Zuckerberg might be forced, by California law, to sell a bunch of shares to raise cash to pay taxes. (Or he could move, but Meta doesn’t want that. The information statement has a little discussion of how Zuckerberg threatened to list his California house for sale, which “would be public and inevitably noted and commented upon by Company employees and, accordingly, raise questions for management of the Company going forward.” Would he really put his house on Zillow? Maybe for negotiating leverage!) Shareholders are more likely to understand “well, he’s gotta sell for tax” than “well, he wants to sell for charity.”

The other reason is of course AI. I said above that, until very recently, a big public tech company was more likely to be buying back stock than issuing it. Meta has been buying back tens of billions of dollars of stock a year, making it easy for Zuckerberg to maintain voting control. But that seems to be changing: As the big tech companies commit to spending hundreds of billions of dollars on AI, some of that will probably be financed out of equity, and Meta will need to issue more stock. Again, shareholders are more likely to understand “well, we need to sell stock for AI” than “well, we might want to sell stock for unknown reasons.” 

Say a company issues stock. The company is worth $10 million on Jan. 1, and it sells 10% of itself to an outside investor for $1 million. On Dec. 31, the company is worth $100 million, and the investor’s 10% stake is worth $10 million. The investor has made a $9 million profit.

Has the company made a $9 million loss? No, that’s stupid. In some rough economic sense, issuing stock is a bet, by the company, against its stock price. If the investor made $9 million on the bet, then the company lost $9 million on the bet. But that $9 million loss is not reflected in the company’s income statement. The company — the enterprise, the business — hasn’t lost $9 million. Instead, what has happened is that (1) the company’s value has increased by $90 million and (2) its 10% shareholder has a claim on $9 million of that increase. That $9 million claim doesn’t reduce the company’s accounting income; it just changes how it’s divided. (You might say that the other investors, who own the other 90%, have had their claim on that income diluted.)

Change the facts slightly. The company is worth $10 million on Jan. 1, and it enters into a contract in which (1) an outside investor gives it $1 million on Jan. 1 and (2) it promises to deliver 10% of its stock to that investor on Dec. 31. On Dec. 31, the company is worth $100 million, and it gives the investor that 10% stake, which is now worth $10 million. The investor has made a $9 million profit.

Has the company made a $9 million loss? This is a slightly closer question, isn’t it? For most of the year, the company didn’t have that outside shareholder; instead what it had was, effectively, a bet against its stock. Still, this is basically the same as the first scenario; it’s just a delayed issuance of stock. It still feels a bit stupid to say that the company has a $9 million loss that reduces its net income. In particular, it hasn’t lost any money: It’s not paying the investor $10 million in cash; it’s just giving the investor the agreed-upon amount of stock.

One more change. The company is worth $10 million on Jan. 1, and it enters into a contract in which (1) an outside investor gives it $1 million on Jan. 1 and (2) it promises to pay cash equal to 10% of the value of its stock to that investor on Dec. 31. On Dec. 31, the company is worth $100 million, and it pays the investor 10% of that, $10 million, in cash. The investor has made a $9 million profit.

Has the company made a $9 million loss? Definitely, right? That’s a real cash payment; the company just made a cash-settled bet against its stock, and lost. Pretty clearly this reduces its net income.

The first scenario is obvious: If you issue stock, and the stock goes up, that doesn’t reduce your income. The third scenario is obvious: If you make a cash-settled bet against your stock, and the stock goes up, that does reduce your income.

The second scenario is weird. Intuitively, promising to issue stock, and then issuing it, shouldn’t create an accounting loss. For one thing, it doesn’t cost you any money. For another thing, it’s conceptually very close to the first scenario, just on a delay. For a third thing, the explanation for the accounting loss here would be “our stock went up, which reduced our net income.” That’s just weird! You should not have to, like, subtract the increase in value of your company from your net income. “Becoming more valuable” is not normally a way to lose money.

Companies do raise money by promising to issue stock in the future: They sell convertible bonds or warrants, giving the buyer the option to get stock in the future. If a company sells a $1,000 bond that converts into 50 shares of stock when the stock is at $20, and then the stock goes to $50, then the bond is worth $2,500. The bondholder has made $1,500. Has the company lost $1,500? Does it have to reduce its net income by $1,500?

The actual answer under US generally accepted accounting principles is “it depends.” Roughly:

1. If the company is just promising to issue stock in the future, then it’s an “equity” transaction, and it doesn’t affect net income. (It is not “marked to market through the income statement.”) If, whenever the bondholder or warrantholder settles up, the company can deliver the agreed-upon amount of stock, that’s stock.^[5]
2. But if the company ever could have to pay cash, then it’s a “derivative” transaction that is marked to market through the income statement. It’s like my third example, just a cash-settled bet on the company’s stock. And in that case, if the company’s value goes up, it has losses on that bet, losses that are marked to market through the income statement.

We talked about this back in 2021. Special purpose acquisition companies normally issue warrants, and back in the SPAC boom, there was a controversy about warrant accounting. Basically what happened is that SPACs all issued a lot of warrants, and they all assumed that the warrants got equity accounting: They were just promises to deliver stock, they could never be settled in cash, so they did not have to be marked to market. But the US Securities and Exchange Commission, as part of a general crackdown on SPACs, said: Not so fast, actually there are weird scenarios where the warrants might have to be settled in cash. (Part-cash part-stock tender offers, for instance.) These scenarios were fabulously unlikely, but they were technically enough to require derivative accounting. And so (1) lots of SPACs had to go around restating their financial statements and (2) those financial statements now had to show losses every time their stocks went up.

Anyway:

Anthropic … said in the draft prospectus [for its initial public offering that] it recorded a net loss of more than $80 billion in the first half of this year, a total that largely reflects an accounting adjustment related to financing agreements. …

The vast majority of that loss, $72 billion, resulted from accounting charges tied to changes in the value of convertible notes and other instruments the company issued to raise funding.

Anthropic had a big net loss in the first half of this year because the value of its stock went up a ton during that half. It lost money by becoming more valuable. Could be worse!

We have talked a couple of times about some students at the University of Chicago who are protesting for the right to trade onion futures. Right now US commodities regulation allows for trading of derivatives on pretty much every commodity except onions (and motion picture box office receipts). This is a weird historical contingency — some guys cornered the onion market once in the 1950s — that was quaint when there were, like, 50 sorts of commodity futures, but is much sillier now that every possible bet on sports, politics, the weather, or anything else you could name (except onions) counts as a “commodity derivative.” 

And so now some Chicago students are demanding the repeal of the Onion Futures Act of 1958 and the legalization of onion futures trading. I assumed these students were protesting out of some combination of (1) genuine fanatical University of Chicago-style devotion to unregulated free markets and (2) it’s very funny to parody fanatical University of Chicago-style devotion to unregulated free markets by marching around chanting and handing out onions.

But of course the real answer is that they’re sponsored by Kalshi. The Chicago Maroon reports:

Approximately 30 UChicago undergraduates and Kalshi affiliates gathered on Monday morning outside the Loop headquarters of the Chicago Mercantile Exchange (CME) Group—the world’s largest financial derivatives trading marketplace—to protest the 1958 Onion Futures Act.

Kalshi, an online prediction market platform, paid for a bus to bring students from campus and provided branded merchandise and breakfast for the group, according to Adam Ash, a third-year undergraduate and co-founder of the For Our Futures student group. …

Kalshi also promoted the event on social media, with co-founder and CEO Tarek Mansour reposting multiple For Our Futures posts on X alongside calls for the repeal of the Onion Futures Act. A video from the demonstration reposted by Mansour has since garnered more than 1 million views.

“This isn’t about prediction markets broadly,” Ash said. “This is about whether an onion farmer should be allowed to hedge his group, just like corn, wheat, and soy [farmers.] We welcome any support from Kalshi, any other individual, nonprofit, whatever.”

Kalshi’s mission is to have betting markets on everything, and onions are things, so here we are. The moral of the story is that nothing is funny and cool and everything is sponsored by gambling companies. 

Early in my career, I was a mergers and acquisitions lawyer, which was a lot of work. I left that job to be a corporate equity derivatives investment banker at Goldman Sachs Group Inc., which was not not a lot of work, but less work. That’s why I left. I could not continue to work the hours I was working in M&A law, so I looked for a new job; I took the job at Goldman because I expected the hours to be better, and they were.

Sometimes people do not believe me when I tell this story, because stereotypically investment bankers work very long hours, so they figure I must have worked very long hours. But (1) my particular group and role weren’t that bad and (2) it’s all relative, and I came from working really long hours. But I do realize that this sounds like a joke, and I lean into it a bit. I frequently say that “I am the only person ever to go to Goldman Sachs for the lifestyle.”

Turns out I’m wrong, though. Today Bloomberg’s Paige Smith has a profile of Jackie Reses, who among other things helped Square (now Block Inc.) become a bank, and who now runs a bank called Lead Bank that “powers some of the world’s best-known financial-technology companies, including Nubank, Stripe and Affirm.” But before that she worked at Goldman, where she learned about having a chill relaxed lifestyle:

Reses had a start in life unlike most bank CEOs. At 16, she rented two stalls on a Jersey Shore boardwalk, one for a milk-can toss, the other for flukey ball. Running the carnival games paid her way through the University of Pennsylvania’s Wharton School — sometimes night classes, because they were cheaper — and was her first job after helping her parents at their pharmaceutical businesses as a kid.

“I didn’t know that people worked five days a week until I went to Goldman Sachs,” she said. “I just assumed everyone worked on Christmas because that’s what my family did my entire life.”

Honestly “I didn’t know that people worked five days a week until I went to Goldman Sachs” is much better than my line.

Musk Tries to Lock In His Tesla-SpaceX Dream Before Trump’s Exit. Barclays to cut 5,000 jobs as bank steps up AI adoption. Musk Says Tesla Aims to Begin Humanoid Robot Sales Next Year. Scott Bessent shrugs off ‘systemic’ AI risks at closed-door investor conference. AI Just Found a Bug That Puts up to 70% of the World’s Websites at Risk. Buyout Firms Look to Swap Out Portfolio Company CEOs as Holds Lengthen. Apollo, Blackstone Have NFL Aspirations, Seek Approval to Buy Stakes in Teams. London credit hedge fund Arini’s losses mount. All-or-Nothing Biotech Bets Dented by Rising Treasury Yields. “The vibes put people at ease and helped evangelize private investments, attendees said.” Andreessen Horowitz to Drop ‘Crypto’ From Name of Flagship Blockchain Fund. Italy to Approve Trucker Aid After Mozzarella Output Disrupted. 

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!