Money Stuff: The Perfect AI Startup

If you run a big artificial intelligence lab in 2025, your company has two important sources of value:

1. Presumably you are building some sort of AI thing. Probably that thing (1) loses a lot of money now but (2) has some plausible path to making a gajillion dollars eventually. You are in the business of burning a lot of investor cash to build God, which would be valuable if you succeed.
2. You employ a lot of AI researchers, and their market value is, what, at least $10 million each? Much more, if they’re good?

The second point is not entirely satisfying. The AI researchers are worth buckets of money because they could be put to work building AI things, which will then make a gajillion dollars. The first point is the theoretically valid one: You have organized resources to do a risky project whose expected future cash flows are large. Those resources include AI researchers. If your project succeeds, the AI researchers will create billions of dollars of value for your company. If it fails, they will move on to another company that can make better use of them. Their value to you in that case will be approximately zero. California does not enforce non-competes.

On the other hand, the cash flows from your AI project really are extremely speculative — AI is moving fast, “it may be difficult to know what role money will play in a post-[artificial general intelligence] world,” and you might not even be sure what your revenue model will be — while the AI researchers trade in a liquid market with clearly observable prices. You know exactly what Mark Zuckerberg will pay for them because he’s standing outside your office holding up giant novelty checks. If you plug “we have 30 top AI researchers” into a valuation model, and the valuation model does not spit out a value of at least several billion dollars, the model is wrong. If the model asks “but what are the AI researchers doing,” the model is missing the point. [1]

In theory there is an incredible arbitrage here. If you are a top AI researcher, probably a lot of your friends are also top AI researchers. If you all got together, you could (1) hang out and have fun and (2) also do some top AI research. And then you could go out to investors and say “hey, look, top AI researchers here, give us a billion dollars.” And if the investors said “yes but what are you all doing together,” you could give them a disapproving look and take a step toward the door, and they would say “we’re so sorry, we don’t know what we were thinking, to make it up to you here’s $2 billion.” 

And then you’d have $2 billion dollars to hang out with your friends. And do top AI research with them, if you want. You probably do want! That’s how you got to be top AI researchers. But do you need to do boring stuff like have a business plan or pursue revenue? I don’t … I don’t really see why you would?

The broad point is that if you start a company, the two ways to make money are (1) make products and sell them for more than they cost to produce and (2) sell stock. In very hot financing markets, one approach is much much much much easier than the other!

At the Information, Stephanie Palazzolo has an incredible article about Mira Murati’s AI startup:

Not only has the one-year-old Thinking Machines not yet released a product, it hasn’t talked publicly about what that product will be. Even some of the company’s investors don’t have a very good idea of what it is working on. While raising capital for Thinking Machines earlier this year and late last year, Murati shared few details about what it would be building, prospective investors said.

“It was the most absurd pitch meeting,” one investor who met with Murati said. “She was like, ‘So we’re doing an AI company with the best AI people, but we can’t answer any questions.’”

Despite that vagueness, Murati raised $2 billion in funding—the largest seed round ever—at a $10 billion pre-investment valuation from top Silicon Valley VC firms including Andreessen Horowitz, Accel and GV. The investors also made the highly unusual decision to give her total veto power over the board of directors.

I’m sorry but that’s the best pitch ever! That is the platonic ideal of a tech startup pitch! Murati and her employees are extremely valuable, and they monetized that value directly. Why should they have to build a product?

I feel like, if I was a hedge fund portfolio manager and my job was figuring out the correct prices of stocks, and I got paid tens of millions of dollars to do that, I would apply some of my skills, on a strictly extracurricular basis, to figuring out the correct prices of hedge fund portfolio managers. You know? Hedge fund PMs seem to move around a lot in a fairly active market, and I would be keenly interested in marking myself to market and selling high as frequently as possible. And I would presumably be good at things like understanding value and spotting market irrationality and taking advantage of counterparties’ mistakes, all of which could help me maximize my paycheck. 

But it’s a less liquid and transparent market than the stock market — there are fewer data points and they are not usually public — so I suppose there’s value in specializing. If you’re a healthcare portfolio manager, you don’t dabble in trading tech stocks, so why would you dabble in trading yourself? You want to get a guy for that. The Wall Street Journal has a story about a guy who’s an agent for hedge fund portfolio managers:

A former portfolio manager at firms including Citadel and Millennium Management, [Ryan] Walsh is now parlaying his own experience to help clients navigate a manic marketplace for hedge-fund jobs. 

“I’m looking to be the Scott Boras of the hedge-fund world,” said Walsh, referring to the baseball superagent who negotiated record-breaking contracts for major-league sluggers like Juan Soto and Bryce Harper. Hanging on the wall of his office is a work of art featuring Ari Gold, the fictional agent from the TV series “Entourage.” …

Walsh, 43, said he has helped 12 clients land jobs as portfolio managers in deals worth a combined $180 million since launching his firm, Laurel Lake Advisors. ...

Walsh saw a need for an agent after peers of his were getting offers worth $50 million and relying on friends to opine on whether they were good deals. Portfolio managers who have been in the same job for years might not know what the going rate is for someone with a given specialty and track record, or which funds have specific needs or have agreed to certain terms with new recruits recently. Walsh hustles to get and share that intelligence.  

I feel like the arc of all financial assets is:

1. First it doesn’t trade: Bonds were mostly bought and held by sleepy insurance companies; investment professionals were mostly loyal to their employer and uninformed of opportunities elsewhere so could not move all that frequently.
2. Then it trades in a discretionary voice market: Bonds were traded by telephoning dealers who know what the market is; now portfolio managers have a guy who “hustles to get and share” intelligence about market levels.
3. Eventually it trades in algorithmic electronic markets. 

We’re probably a long way from that final outcome in the market for portfolio managers. (Or AI researchers.) But will, like, LinkedIn eventually offer an automated market where hedge funds can post bids and portfolio managers can post offers and a machine-learning model constantly scans market data points to create an up-to-the-minute price for every portfolio manager?

Also:

“The job he’s doing, my sense is it requires a certain sense of chutzpah and commercial sense,” said Cliff Sosin, who runs the hedge-fund firm CAS Investment Partners and has been friends with Walsh since they went to the same private high school together in New Haven, Conn. “You have to be willing to call these firms and say ‘F you, this guy is worth more.’” 

That is a valuable service. If for some reason I was in a situation where I could get tens of millions of dollars but only by swearing at Ken Griffin or Steve Cohen, I’d be happy to pay someone else a million dollars to do it.

You know my theory:

1. Prediction markets like Kalshi, which only relatively recently became available in the US, are moving rapidly into offering sports gambling. (Sorry: sports predicting.) They have some huge advantages as sports gambling platforms. They are available in all 50 states: Some states ban sports gambling, but Kalshi has (so far mostly successfully) argued that, as a federally regulated commodities exchange, it is exempt from those bans, so it can offer sports betting everywhere. They are available in your brokerage app: Robinhood has teamed up with Kalshi to offer football bets, so you can bet on football from the same app that you use to invest your retirement savings. Also it seems likely that the tax treatment of football bets on prediction markets is significantly better than the treatment of football bets at a regular sportsbook (you can deduct 100% of your losing bets rather than 90%), so serious gamblers might prefer to do their betting on prediction markets.
2. This is extremely weird, but no one in the federal government seems interested in stopping it, so there you go.
3. Sports betting is a very important business for prediction markets. Prediction markets get a lot of academic and theoretical attention: It seems socially beneficial to have an market that produces probabilities for various important events. (If the market says there’s an 80% chance of war, stock up on canned goods, that sort of thing.) But this theoretical benefit is hard to realize in practice, because prediction markets aren’t always deep and liquid enough to incentivize people to make good predictions. (There are no index funds, and not many recreational bettors, betting on war, so if you do have a highly informed view that war is likely you can’t make all that much money from it on Kalshi, so you won’t bother trading to correct prices.) Traditionally, in the US, prediction markets get a ton of volume and attention every four years when there’s a presidential election, and considerably less otherwise. But sports happen every week, and there are tons of recreational bettors who want to bet on them, [2] so sports prediction markets can be deep and liquid and filled with noise traders for professionals to bet against. So the prediction markets can pay the bills, but they can also attract gamblers and get them addicted and have them branch out into other, less obvious sorts of predictions — weather, geopolitics, school board elections — to finally realize the dream of complete and informative prediction markets.

One way to think about it is the Grossman-Stiglitz paradox: To get an efficient market (here, prediction markets that accurately predict important events), you need to create conditions where informed traders can make a lot of money making prices correct, which means that you need some noise traders willing to trade at incorrect prices to entice the informed traders. And you entice the noise traders with football.

Anyway here’s a Wall Street Journal article about Kalshi’s efforts to disrupt football betting:

“This is the No. 1 thing that we talk about now,” said Jordan Bender, a gaming-industry analyst at Citizens Capital Markets and Advisory. “It’s this new, unknown opportunity. It feels like 50% of the conversations these days are around prediction markets.”

Kalshi operates in a regulatory gray area. As an exchange licensed by the Commodity Futures Trading Commission, Kalshi makes its sports contracts available nationwide—even in states such as California and Texas, where online sports betting is illegal. Kalshi says its federal status means it doesn’t need to abide by local regulations such as restrictions limiting gambling to those 21 and older.

In June, the attorneys general of 34 states signed a brief arguing that Congress never intended to create an “events-contract loophole” allowing Kalshi to offer sports wagers, and complaining that Kalshi was ignoring state safeguards against gambling addiction. …

So far, Kalshi has faced little opposition from the administration of President Trump, whose oldest son, Donald Trump Jr., became an adviser to Kalshi in January. This month’s start of the National Football League season powered record trading volumes at Kalshi, defying skeptics who expected the platform to struggle outside of election years. More than three-quarters of Kalshi’s volume is now sports-driven.

Right the classic thing that people want to bet on is sports. But if you hook them with sports, maybe you can get them to bet on other stuff too.

Elsewhere in prediction markets:

• On Sunday evening, the National Football League announced that Bad Bunny will headline the Super Bowl halftime show in 2026. Kalshi and Polymarket had contracts on this question, though Polymarket’s is technically “Who will perform at Super Bowl halftime show?” (so it can resolve to more than one performer) and Kalshi’s is technically “Who will headline the Pro Football Championship halftime show?” (because they don’t license “Super Bowl” from the NFL). Both markets jumped up significantly on Sunday, before the announcement. We have talked about insider trading in prediction markets and, uh, I have a feeling it is going to keep coming up a lot. The people with academic and theoretical interest in prediction markets tend to like insider trading, because the social purpose of prediction markets is to create accurate predictions, so incorporating inside information is good. But the people with sports-betting interest in prediction markets tend to dislike insider trading, because it scares off casual bettors who provide all the revenue. Also it is not like there is all that much social benefit in knowing that Bad Bunny will perform at the Super Bowl hours before it is announced.
• Also on Sunday, Eric Adams dropped out of the New York City mayoral race. A few weeks ago, we talked about Bill Ackman’s proposal that Adams should (1) insider trade on Polymarket by betting that Andrew Cuomo would win that race, (2) drop out and throw his support to Cuomo and (3) profit when Cuomo’s contract rallied. “There is no insider trading on Polymarket,” Ackman tweeted. We discussed several problems with that trade, of which the ambiguity around prediction-market insider trading rules was the least important and the lack of liquidity was the most important. As it turns out, Cuomo’s contract on Polymarket rose from about 11% to about 14% so it wouldn’t have worked very well anyway.

Hey it’s a new largest leveraged buyout ever:

Electronic Arts Inc. agreed to sell to a group of private investors in a deal that values the company at about $55 billion, marking the largest leveraged buyout on record.

The consortium, which includes Silver Lake Management, Saudi Arabia’s Public Investment Fund and Jared Kushner’s Affinity Partners, agreed to pay $210 per share in cash for the Redwood, California-based company, according to a statement Monday. That represents a 25% premium to where its shares traded before the talks leaked on Friday.

The deal will be funded by a $36.4 billion equity commitment, plus a $20 billion debt commitment from JPMorgan Chase & Co. The press release and 8-K do not seem to say how that equity commitment for the largest LBO in history will be split among Silver Lake, PIF and Affinity.

One can make some guesses however. Silver Lake, the press release says, “is a global technology investment firm, with more than $110 billion in combined assets under management,” and Dan Primack reports that it is investing out of a $20.5 billion fund. The Saudi PIF has $925 billion of assets and “the gaming and esports industry is one of its priority sectors.” Also PIF owns 9.9% of EA already, worth about $5.2 billion at the deal price. It seems likely that PIF, with $925 billion of assets and an existing investment in EA, is contributing a majority of that $36 billion, and that Silver Lake — a big private equity sponsor with a lot of experience leading tech leveraged buyouts, but an order of magnitude smaller than PIF — is contributing less. (The equity check is almost twice the size of Silver Lake’s entire $20.5 billion fund.)

Meanwhile Affinity has “over $5.4B under management,” which is even smaller. If it put 20% of its assets in just this one deal, it would write about 3% of the equity check. Also the Saudi PIF is a big investor in Affinity. So the PIF, which was already a big investor in EA, had an existing relationship with management, and seems to be leading the deal, will presumably also invest in the deal through Affinity, a relatively small asset manager. 

The New York Times notes, unrelatedly:

The buyout would need the approval of the Committee on Foreign Investment in the United States, a panel of government agencies that reviews international deals for security concerns. Some lawmakers have previously called for scrutiny of the Saudi fund’s investments in sports for national security reasons.

“People don’t often think about video games and national security together, but these are platforms that reach millions of Americans and often collect a lot of personal data,” said Aaron Bartnick, a former official in the Biden administration who worked on national security reviews of foreign investments who is now a fellow at Columbia University. He suggested that the committee would “want to take a close look even if they ultimately end up signing off.”

And Bloomberg News comments:

Most large deals with a foreign investor require some form of government approval. Affinity Partners — founded by Kushner, the president’s son-in-law, during the first Trump administration — is participating in the deal. Affinity is backed by foreign investors, including some from the Middle East.

Just a series of facts there.

Look I don’t like it either:

A group of Disney investors is alleging that the company prioritized “improper political” considerations over the best interests of stockholders with its brief suspension of Jimmy Kimmel last week. ...

Zoom in: In a Wednesday letter to CEO Bob Iger, lawyers representing the American Federation of Teachers, AFL-CIO and Reporters Without Borders said they are seeking answers about the suspension of “Jimmy Kimmel Live!” ...

The investors seek board minutes, internal communications, affiliate agreements and analyses of the suspension's financial impact. While the group applauded ABC for doing the “right thing” in bringing back the show, the letter noted the company still faces negative repercussions given President Trump's threat to “test ABC out on this.” 

The lawyers noted that “Disney’s stock suffered significant declines in response to the Company’s abrupt decision to suspend” the show. ...

“There is a credible basis to suspect that the Board and executives may have breached their fiduciary duties of loyalty, care, and good faith by placing improper political or affiliate considerations above the best interests of the Company and its stockholders,” the lawyers wrote.

It’s not really “securities fraud,” but rather the technically distinct but practically related “breach of fiduciary duties.” [3]  But you know what I say: Every bad thing a public company does is also securities fraud, and cancelling Jimmy Kimmel is arguably a bad thing (the stock went down!), so. So.

The main point that I want to make here is that this stuff is perfectly symmetrical. If Walt Disney Co. had not suspended Kimmel, (1) President Trump would have said mean things about it, (2) the stock would have gone down at some point (and up at other points!) and (3) some different collection of not-purely-economically-motivated shareholders would have hired lawyers to send Disney a letter like this. “There is a credible basis to suspect that the Board and executives may have breached their fiduciary duties of loyalty, care, and good faith by placing improper political or wokeness considerations above the best interests of the Company and its stockholders,” etc.; the letter would have been almost exactly the same.

We have talked about this before. Back before Paramount Global paid Donald Trump to settle an absurd “60 Minutes” lawsuit, I wrote:

Paramount is apparently considering settling the lawsuit with some large payment to Trump. If it doesn’t pay Trump and the deal is not approved by the FCC, that is bad for shareholders. If it does pay Trump there is at least a theoretical risk that it will get in trouble for violating bribery laws, as some US senators have threatened, and of course that would be bad for shareholders. Doing the thing that Trump wants will get you accused of corruption (you allegedly bribed him); not doing the thing that Trump wants can get you accused of corruption (“election interference” is apparently the problem with CBS News); either thing can be bad for shareholders and can get you in trouble.

Similarly, here, there are people who think that you have a fiduciary duty to suspend Jimmy Kimmel, and there are other people who think that you have a fiduciary duty not to suspend Jimmy Kimmel, and all of them are looking to sue.

We talked a month ago about Eric Jackson, the fund manager and meme-stock cheerleader who has led Opendoor Technologies Inc.’s recent meme-stock rise. Specifically we talked about how he is standing outside Drake’s house in Toronto with signs asking Drake to buy Opendoor. Why? Wrong question, buddy! “I’m sure that there is some reason for this,” I wrote, “though I don’t know what it is and don’t especially want to find out.” Just, like, meme stocks, man. You do silly stuff, people notice, the stock goes up. Don’t overthink it.

This weekend Bloomberg News had a story about Eric Jackson standing outside Drake’s house, which is definitely a thing he is doing, though not, like, for a reason:

“All these people are telling me to stop,” Jackson said in an interview in front of Drake’s house, on what he called Day 29 of his campaign to enlist the celebrity. “I’m just going to ignore them. I’m going to trust my gut.”

Okay.

Jackson bristles at seeing Opendoor through the meme-stock lens. Instead, he prefers to think of it as a “cult stock,” like a great film that’s ignored on its release but attracts legions of fans as the years go by. 

Okay.

Drake still hasn’t bought in, as least as far as Jackson knows. He hasn’t spoken to the rapper. Rob Schneider, the former Saturday Night Live comedian and star of the 1999 film Deuce Bigalow: Male Gigolo, became self-proclaimed “believer” in the company, though.

Others are curious. Outside of Drake’s residence, Jackson suddenly received a cellphone call from the Toronto rapper’s head of security, Bucky, who saw him through the cameras. Bucky wasn’t chasing Jackson away. He wanted to talk more about Opendoor and said he wished he had caught the rally.

If you can get Rob Schneider and Bucky, Drake’s head of security, interested in a stock, is that something? Does that make the stock go up?

Citi’s Wall Street Rebound Turns Raghavan Into CEO Contender.  First Brands Files for Bankruptcy After Weeks of Lender Scrutiny. Apollo Sidestepped Blacklist to Short Ailing First Brands’ Loans. The Credit Market Is Humming — and That Has Wall Street On Edge.Investors Pile Into Funds Betting on Elusive Market Volatility. Hedge Funds Turn Bearish on London Cocoa as Supplies Improve. “Never before has an individual seeking a US president’s goodwill channeled so much money to the president’s family.” Why Microsoft Has Lower Borrowing Costs Than the U.S. A New Front Opens Between Zuckerberg and Musk Over Robots. White Men Make a Comeback in America’s Boardrooms. AI can now pass the hardest level of the CFA exam in a matter of minutes.

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!