DOGE kills its first bill, Zuck vs OpenAI, Google's AI comeback with bestie Aaron Levie

The episode opens with teasing about David Sacks's absence. One participant compares missing him to a breakup, saying "it's hard to get over your ex," and describes the relationship as codependent. The joke turns into a bar-fight scenario with four thought bubbles: one host wants to fight, another is distracted by a woman in the bar, and Friedberg is pictured worrying about how to get out of it unscathed.

The introductions follow. David Friedberg, the "sultan of science," gets asked about his pawn-themed background, which is described as "the impostor syndrome." Someone notes that a small, devoted group of viewers tries to decode his backgrounds. Chamath Palihapitiya is introduced next, and the group mentions plans to ski together over the holidays. The guest fifth seat goes to Aaron Levie, CEO of the publicly traded company Box, introduced as having "the most to lose by coming here."

When someone says Box is at all-time highs, a participant jokes about whether the company has been buying Bitcoin for its treasury like Michael Saylor. The answer is that a company that "didn't get crushed during the post-COVID period" can "just keep keep cranking." The banter then turns to meme stocks and companies that sell convertible notes to buy Bitcoin and then join the Nasdaq 100. Asked about Box's cash, the answer is "probably 600 or 700" after a recent convert. One participant suggests a two-week experiment: put 5% of the treasury, 30 million, into Bitcoin. The reply is that "a very sophisticated audit committee" stopped a similar impulse during "a brief week in 2021."

Housekeeping: the show now owns the allin.com domain and has started an email list, which the host pitches with self-mocking lines about spam and "first dibs on overpriced event tickets." Sacks is out again because he is in Washington, D.C., for meetings.

Asked about Sacks in his new government role, the respondent calls it "a strong pick" on AI and says they are indifferent and less informed on crypto. The argument is that, at this stage of the technology, the role needs someone with a deregulatory, anti-regulation leaning, because the bigger risk right now is slowing progress too much.

The next question goes to the CEO of a public software company: how did they react to the Biden administration's AI proposals and to California's bill, which set constraints based on parameter counts and model size? The respondent names two objections to SB 1047. The first is that state-by-state rules would be unworkable:

you're in a world of hurt if every state has a different approach to this.

The second objection is the bill's underlying philosophy. As it moved from first draft to final version, it treated AI progress as inherently risky and kept raising the consequences for model developers. The respondent focuses on second- and third-order effects. Would Meta want to release the next version of Llama, or even incremental updates, if it carried that much liability? The market currently benefits from intense competition among five or six players. The respondent wants those players moving as fast as possible, without a council reviewing every release out of fear of government lawsuits worth hundreds of millions of dollars if one user misuses a model. The same concern is applied to some proposals for national legislation.

The respondent is milder on federal action under Biden. The earlier executive order is described as lacking "a lot of teeth," closer to watching and studying the space. The current head of OSTP, named in the transcript as Arti Prabhakar, is called strong, technical, familiar with Silicon Valley and not inclined to overregulate. The respondent expects Sacks's efforts to lean even further toward AI progress and away from "accidentally overregulating too early in this journey."

The conversation turns to what Sacks could do on crypto. The ideas raised include protecting consumers from being left "holding the bag," sandboxing projects, know-your-customer rules and a sophisticated-investor test. Chamath is then asked for his short-term priorities.

The answer begins with stablecoins as the most obvious valuable use of crypto rails, especially for real-time payments. The speaker says the U.S. government already uses some stablecoins and that corporate adoption is growing. A stablecoin is defined as a digital token you get in return for putting a dollar into a wallet. The two big issuers are named as Tether, described as based abroad, "somewhere in the Caribbean," and USDC, run by the American company Circle. The issuer is supposed to buy short-term Treasury securities so each token stays backed. The speaker illustrates the risk: a billion dollars of stablecoins backed by only $900 million of assets creates insolvency and bank-run risk.

The discussion also covers why issuing stablecoins is profitable. Issuers hold the deposited money, and when interest rates are 2% to 5%, they are "making billions of dollars," which another voice calls keeping "the float." The speaker then gives an operational example: SpaceX is said to collect Starlink payments in long-tail countries by swapping into stablecoins, sending them back to the United States and converting to dollars, which avoids foreign-exchange risk and wire transfers. The first policy goal, then, is to make these rails standard in the U.S. and "chip away at all of this decrepit infrastructure that the banks use."

The second goal is competition with Visa, Mastercard and American Express. The speaker says they pay Stripe 3%, and that paying through Stripe with stablecoins costs less than 3% but more than zero. The claim is that taking those 300 basis points to zero would be "a boon to global GDP because that's 3% on many tens of trillions of dollars." Another participant agrees the transaction tax on card rails is "obviously insane" and adds a practical condition: it works only "if you could get everybody to kind of coordinate against that."

The discussion adds caveats about issuers. One voice says regulation could move people off Tether, which it says has "a checkered, colorful, sordid past." Another responds that, in fairness, both companies appear pegged one for one "as of today," though their jurisdictions differ.

Sequencing is the political argument. Stablecoin adoption and cheaper transaction rails are framed as uncontroversial, so Sacks would not get caught in "a political quagmire." Bitcoin and long-tail crypto projects come later, once momentum builds. The speaker cites the backlash to their own recent comments on encryption and quantum computing, which some "maxis" piled onto before others understood the technical point. That world is described as "so religious" and animated that it is a poor place to start. The recommended order:

Go to stablecoins, then disrupt the Visa rails and then go to the other stuff.

Another participant would put an accreditation test first, citing the SEC's mandate, so that educated buyers know they could lose their money.

A further view splits crypto into two categories. The first is DeFi and financial crypto, which improves payment rails and, if you treat Bitcoin as a store of value and digital gold, "can actually kinda make sense." The second is the use of tokens as virtual currencies, token equities and tokens for anything. That category raises SEC questions about whether these are securities, whether there is insider trading, and whether anyone can issue and promote them on Twitter. The respondent says Gensler "did not get his arms around this whole thing," and argues that staying with "crypto 1.0" as financial infrastructure would have avoided much of the noise. The respondent also says that can no longer be undone. They doubt you "could get 10 crypto people to agree on how you regulate that second category," since some see NFTs as freely tradable, like aftermarket concert seats, while others treat them as securities. Another voice agrees: "Absolutely not."

The segment ends with banter about rotating the fifth seat among friends of the pod and newsmakers, including a joke about comparing resting heart rates.