a16z rebrands to Frontier, CFTC clears equity perps, and debate over 'obliterated' AI models
Summary
Robert Leshner discusses a16z Crypto's rebrand to a16z Frontier, expanding its focus to AI and frontier technologies. The hosts analyze the 2026 midterm election results, noting prediction markets accurately forecasted a Democratic sweep but underestimated its magnitude. They examine the CFTC's clearance of single-stock perpetual futures and the immediate legal challenge by Citadel and SIBO. Finally, they debate the regulation of 'obliterated' AI models, with Leshner arguing that banning them constitutes prior restraint and hinders cyber defense.
Key takeaways
- 1 For anyone who missed the news two weeks ago, A16Z is no longer A16Z crypto. They are A16Z Frontier.
Robert Leshner · 2:51
- 2 They announced their Q3 revenues, News, which was 17 billion versus 11.5 11 and a half billion they made in Q2. So a fifth rough. above 50% quarter over quarter increase. Uh they are back to an operating loss. So they were uh operating profit in Q2, operating loss uh very slightly in Q3.
Unidentified speaker (speaker_2) · 34:33
Episode sections
- 2:51 – 5:40
a16z Crypto rebrands to a16z Frontier
Robert Leshner notes that a16z Crypto has rebranded to a16z Frontier, expanding its mandate beyond crypto to include AI and other frontier technologies. He contrasts this with Paradigm's similar move in 2023, which was met with severe backlash on Twitter, whereas a16z's move was accepted as a 'sign of the times.'
- 5:40 – 14:23
2026 Midterm Election Results and Prediction Markets
The hosts discuss the 'blue mega tsunami' in the 2026 midterms, where Democrats swept control of Congress. Robert Leshner argues that while prediction markets like Polymarket correctly called the outcome, they underpriced the magnitude of the wave because they relied on polls that significantly underestimated the Democratic surge. Tom Schmidt notes that prediction markets resolved the outcome hours before traditional news outlets realized it was over.
- 14:23 – 24:15
Political Implications for Crypto Legislation
Robert Leshner asserts that despite a pro-crypto majority in the new Congress (roughly 2:1), legislative progress on crypto bills like the Clarity Act is unlikely due to Trump's veto threat and the focus on Trump-era corruption. He suggests the next two years will be dominated by investigations into World Liberty Financial and Trump Coin, with any potential legislative action relying on Trump pivoting to improve his approval ratings.
- 24:15 – 29:03
Political Landscape and 2028 Outlook
Speakers discuss the political implications of the midterm elections, noting a unified government opposition to Trump and a shift in prediction markets favoring Democrats for 2028, with Josh Osoff emerging as a frontrunner. Robert Leshner predicts two years of political boredom and less DC-focused crypto regulation.
- 29:03 – 34:24
CFTC Clears Equity Perps and Legal Challenges
The CFTC cleared single-stock perpetual futures for trading on SEC-notice registered DCMs. SIBO and Citadel sued immediately, claiming these are swaps. Tarun Chitra and Robert Leshner debate whether the lawsuit is a principled legal stand or a strategic move to slow down the market to prepare.
- 34:24 – 41:30
Anthropic Financials and AI Compute Economics
Anthropic reported Q3 revenues of $17 billion, up from $11.5 billion in Q2, but returned to an operating loss. The S-1 update added recursive self-improvement (RSI) as a risk factor. Robert Leshner argues that a blue wave slowing data center construction benefits existing hyperscalers and AI labs by reducing CapEx burdens and competition for compute.
- 41:30 – 43:35
Data Center Politics and Cryptoization of AI
The discussion covers the political backlash against data centers, highlighted by Texas Mayor Gina Hinojosa running on a platform to slow new builds. Speakers predict a divergence where non-US compute builds accelerate, leading to a 'cryptoization' of the data center industry with geographically distributed infrastructure.
- 43:35 – 48:03
Obliterated AI Models and Regulation
The Foundation for American Innovation (FAI) published a paper calling 'obliterated' AI models (those stripped of safety refusals) a threat. Speakers argue these models are essential for cyber defense and red teaming, and that banning them constitutes prior restraint, especially given the ease of creating them using larger models.
- 48:03 – 51:29
Obliterated Models and Regulatory Feasibility
Speakers discuss the inevitability of 'obliterated' (unrestricted) AI models existing alongside restricted ones, arguing that regulation cannot eliminate them. They compare this to early crypto privacy tools, suggesting that bans are ineffective and that the focus should be on defensive use and context-based moderation rather than prohibition.
- 51:29 – 53:41
Anthropic's Cybersecurity Refusals and Prior Restraint
Robert Leshner characterizes the push to ban obliterated models as 'prior restraint,' citing the OpenAI/Hugging Face attack where defensive capabilities were refused by frontier labs. He argues that restricting access to these models hinders legitimate defense and red-teaming efforts, drawing parallels to First and Second Amendment rights.
- 53:41 – 55:57
Liability vs. Expression and Geopolitical Reality
The hosts debate the line between protected expression and illegal acts, noting that while liability exists for negligent hosting, the primary users of unrestricted models for defense are Americans, not foreign actors. They conclude that the threat of US providers enabling domestic hacking is theoretical, as actual cyberattacks are predominantly foreign.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| BitcoinBTC | crypto | neutral | I mean, television Taler Rico is pro-crypto. He's like a Bitcoin guy, right? |
| HyperliquidHYPE | crypto | neutral | Hyperliquid, which uh recently entered into this partnership with ICE, might be one of the beneficiaries of this uh this new ruling by the CFTC. |
| EthereumETH | crypto | neutral | crypto was up, Bitcoin was up three, b Ethereum was up six. |
| NVIDIANVDA | equity | bearish | NVIDIA and like most of the iSocks are down, right? Yeah. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
a16z Crypto rebranded to a16z Frontier, expanding its mandate to include AI and other frontier technologies beyond just crypto.
Robert Leshner ·2:51
How to verify: Check a16z's official press release or blog post announcing the rebrand from late October 2026. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceParadigm attempted to rebrand as a crypto and AI firm in 2023, shortly after ChatGPT's release, and faced severe backlash on Twitter.
Robert Leshner ·4:35
How to verify: Search Twitter archives and news articles from late 2026 regarding Paradigm's announcement and the subsequent community reaction. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Maxine Waters is drafting a revamped version of the crypto regulatory clarity act.
Unidentified speaker (UNKNOWN) ·15:05
How to verify: Check the US House of Representatives official website or Maxine Waters' press releases for new legislative text introduced in late 2026. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Trump has stated he will veto any crypto legislation that passes Congress.
Unidentified speaker (UNKNOWN) ·19:15
How to verify: Search for Trump's public statements or press conferences from late October/November 2026 regarding his stance on crypto legislation. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Republicans will realize they need to break with Trump to have a chance at winning in 2028.
Unidentified speaker (UNKNOWN) ·24:15
How to verify: Monitor 2026 Republican primary polling and candidate statements regarding Trump. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Prediction markets are pricing a 68% chance that the 2028 victor will be a Democrat.
Unidentified speaker (speaker_2) ·26:25
How to verify: Check Polymarket or other prediction market platforms for 2026 US Presidential election odds. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Josh Osoff is now the number one most likely candidate for the Democratic nomination, leading AOC and Gavin Newsom.
Unidentified speaker (speaker_2) ·26:45
How to verify: Check prediction market odds for Democratic presidential candidates. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
JD Vance is the front runner on the Republican side, with Rubio a few points behind.
Unidentified speaker (speaker_2) ·27:05
How to verify: Check prediction market odds for Republican presidential candidates. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
The CFTC cleared single-stock perps for trading in the US on DCMs that are SEC notice registered for security futures.
Unidentified speaker (UNKNOWN) ·29:03
How to verify: Review CFTC statement issued by Chairman Rostin Behnam on the day of the midterms (Nov 2026). For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceSIBO and Citadel sued the CFTC within hours of the ruling, claiming equity perps are swaps.
Unidentified speaker (speaker_2) ·29:23
How to verify: Search federal court filings for lawsuits filed by SIBO and Citadel against the CFTC dated Nov 2026. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceAnthropic reported Q3 revenues of $17 billion, up from $11.5 billion in Q2.
Unidentified speaker (speaker_2) ·34:33
How to verify: Review Anthropic's S-1 filing or press release for Q3 2026 financial results. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceAnthropic added recursive self-improvement (RSI) as a risk factor in its S-1 update.
Unidentified speaker (speaker_2) ·35:12
How to verify: Review the risk factors section of Anthropic's S-1 filing. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceTexas Mayor Gina Hinojosa ran on a platform to slow down or impose a moratorium on new data center build outs.
Unidentified speaker (UNKNOWN) ·41:53
How to verify: Review Gina Hinojosa's campaign platform and election results for Texas municipal elections. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
The Foundation for American Innovation (FAI) published a position paper calling obliterated AI models a threat.
Unidentified speaker (speaker_2) ·43:59
How to verify: Search for FAI publications or press releases regarding 'obliterated models' or 'AI safety'. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Polymarket had called the midterm results at 99% by 7 PM.
Unidentified speaker (speaker_1) ·48:03
How to verify: Check Polymarket historical data for the '2026 US Midterm Election' market resolution and timestamp. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Anthropic's S-1 update added recursive self-improvement as a risk factor.
Unidentified speaker (speaker_1) ·48:03
How to verify: Search SEC EDGAR for Anthropic's S-1 filing and review the 'Risk Factors' section for mentions of recursive self-improvement. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Primary sourceThe OpenAI/Hugging Face attack occurred.
Unidentified speaker (speaker_2) ·51:49
How to verify: Search news archives for reports on the 'OpenAI Hugging Face attack' and Hugging Face's defense using open models. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Hugging Face had to defend itself using an open model because OpenAI refused to engage in defensive behavior.
Unidentified speaker (speaker_2) ·51:49
How to verify: Review Hugging Face's official blog or press releases regarding the incident and their use of open-source models for defense. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
OpenAI and Anthropic's cybersecurity programs (Daybreak and Glasswing) refuse some red-teaming requests.
Unidentified speaker (speaker_2) ·51:49
How to verify: Check the official websites or documentation for OpenAI's Daybreak and Anthropic's Glasswing programs for their policies on red-teaming access. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.
Do your own research
- Verify a16z Frontier's official press release and updated investment thesis to confirm the scope of their new mandate beyond crypto.
- Check the CFTC's official order clearing single-stock perpetual futures and the specific legal definitions of 'security futures' versus 'swaps' cited in the ruling.
- Review the S-1 filing update from Anthropic to assess the specific risk factors disclosed regarding recursive self-improvement (RSI) and financial performance.
- Examine the Foundation for American Innovation (FAI) paper on 'obliterated' models to understand the regulatory arguments against unrestricted AI capabilities.
- Research the legal filings from Citadel and SIBO against the CFTC to evaluate the merits of their claim that equity perps are swaps.
- Analyze data from cyber defense firms to determine the actual volume of 'obliterated' model usage for domestic vs. foreign cyber threats.
- Check prediction market odds for the 2028 presidential election to see if Josh Osoff remains the frontrunner as reported.
- Review the OpenAI/Hugging Face incident reports to verify claims about frontier labs refusing defensive use cases for their own security teams.
Full transcriptexpand
- 0:00
- Robert Leshner: What about this? What if we rename the show the chipping block?
- 0:05
- Robert Leshner: That way we can work
- 0:07
- Unidentified speaker (UNKNOWN): the AI
- 0:08
- Robert Leshner: in.
- 0:08
- Unidentified speaker (UNKNOWN): No,
- 0:08
- Unidentified speaker (speaker_1): no, no. But but the chopping block has the view of like a guillotine, right? It's like you're dead in one s one kill. Where the chipping
- 0:14
- Unidentified speaker (UNKNOWN): block is
- 0:14
- Unidentified speaker (speaker_1): like I'm just like slowly beating you up.
- 0:18
- Unidentified speaker (UNKNOWN): I mean,
- 0:18
- Robert Leshner: that
- 0:18
- Unidentified speaker (UNKNOWN): sounds like the show. We've
- 0:19
- Robert Leshner: been doing this for years. We're slowly
- 0:21
- Unidentified speaker (UNKNOWN): chipping shopping away.
- 0:22
- Unidentified speaker (speaker_2): Principles are getting chipped away at. That's the
- 0:23
- Unidentified speaker (UNKNOWN): chipping block.
- 0:24
- Robert Leshner: Not a dividend. It's
- 0:25
- Unidentified speaker (speaker_1): a the tale of two Tucon. quans.
- 0:27
- Unidentified speaker (speaker_2): Now, your losses are on someone else's balance sheet.
- 0:29
- Unidentified speaker (speaker_1): Generally speaking, airdrops are kind of pointless anyways.
- 0:32
- Unidentified speaker (speaker_2): Unnamed
- 0:33
- Unidentified speaker (speaker_1): trading firms who are
- 0:34
- Unidentified speaker (speaker_2): very involved.
- 0:35
- Unidentified speaker (speaker_1): Polic.eetha is the ultimate obviously.
- 0:37
- Robert Leshner: DeFi protocols are the antidote to this problem.
- 0:41
- Unidentified speaker (speaker_2): Hello, everybody. Welcome to the chopping block. Every couple weeks, the four of us get together and give the industry insider's perspective on the crypto topics of the day. Quick intro, intros. is first you got Tom, the DeFi Maven and Master
- 0:50
- Unidentified speaker (UNKNOWN): of
- 0:50
- Unidentified speaker (speaker_2): Meme. Memes.
- 0:51
- Unidentified speaker (speaker_3): Hello, everyone.
- 0:52
- Unidentified speaker (UNKNOWN): Next
- 0:53
- Unidentified speaker (speaker_2): we've got Tarun, the Giga Brain, and Grand Pooh Bah at Gauntlet.
- 0:56
- Unidentified speaker (speaker_1): Yo.
- 0:57
- Unidentified speaker (UNKNOWN): And
- 0:58
- Unidentified speaker (speaker_2): we've got Robert, the crypto connoisseur and czar of Superstate.
- 1:01
- Robert Leshner: Good morning.
- 1:02
- Unidentified speaker (UNKNOWN): And I
- 1:02
- Unidentified speaker (speaker_2): am Asib, the head hype man at Dragonfly. We are early stage investors in crypto, but I want to caveat that nothing we say here is investment advice, legal advice, or even life advice. Please see chopping block block.xyz that XYZ for more disclosures. So, boys, I am back. I assume you all miss me. I can I can feel how desperately you guys want to be.
- 1:19
- Unidentified speaker (UNKNOWN): Welcome back, Asiep. Welcome.
- 1:21
- Unidentified speaker (speaker_2): Thank you. That's
- 1:22
- Unidentified speaker (UNKNOWN): it. How was
- 1:22
- Unidentified speaker (speaker_3): it over there?
- 1:23
- Unidentified speaker (speaker_2): Uh it was weird. It was weird. I was not the host. So I was just guessing it up and I realized how nice it is to not have to run the show and like keep everything on track. It's um I I I envy you guys. Listen,
- 1:37
- Robert Leshner: I
- 1:37
- Unidentified speaker (UNKNOWN): will say
- 1:37
- Robert Leshner: I keep everything on track. Take all the credit
- 1:40
- Unidentified speaker (UNKNOWN): you want.
- 1:41
- Unidentified speaker (speaker_3): Please.
- 1:43
- Unidentified speaker (speaker_3): You know, one thing I I you know Rob was great. One thing Rob did, which maybe we should do, is uh
- 1:50
- Unidentified speaker (UNKNOWN): he
- 1:50
- Unidentified speaker (speaker_3): said the frontier topics of today. The frontier
- 1:53
- Unidentified speaker (UNKNOWN): topics. He
- 1:54
- Unidentified speaker (speaker_2): did say that. Um what
- 1:56
- Unidentified speaker (UNKNOWN): do
- 1:56
- Unidentified speaker (speaker_3): you think?
- 1:57
- Unidentified speaker (speaker_2): I I mean look, I I feel like we I feel like we gotta own who we are. Um Uh and this is uh it's the chopping block, it's on Unchained. We talk mostly about crypto. We do talk about other stuff. So obviously there's uh there's stuff in here increasingly about AI, and you know, we're gonna be talking about the midterms, obviously, which is expanding beyond crypto. And
- 2:17
- Unidentified speaker (speaker_2): so I kind of feel like it's um it's it's we're there's there's a sort of free association that's happening within crypto and around crypto. Uh so we will come. cover other topics, but frontier it's like I mean, look, we're not gonna be talking about nuclear.
- 2:31
- Unidentified speaker (UNKNOWN): Yeah, I
- 2:31
- Unidentified speaker (speaker_3): was just joking
- 2:32
- Unidentified speaker (UNKNOWN): just to be clear.
- 2:33
- Unidentified speaker (speaker_2): Okay,
- 2:33
- Unidentified speaker (UNKNOWN): yeah.
- 2:34
- Unidentified speaker (speaker_1): Yeah, I I think I think the only time we can use that word legally is if Chris Dixon is a guest on the show.
- 2:40
- Robert Leshner: Choppy Frontier.
- 2:41
- Unidentified speaker (speaker_2): True. We did not talk about that, did we? Did we did we do uh A6 and C crypto or AC
- 2:45
- Unidentified speaker (UNKNOWN): and Z frontier?
- 2:47
- Unidentified speaker (UNKNOWN): I think yeah, no,
- 2:48
- Unidentified speaker (speaker_1): I thought we did
- 2:49
- Unidentified speaker (UNKNOWN): we sort
- 2:49
- Unidentified speaker (speaker_1): of uh maybe we didn't really talk about
- 2:50
- Unidentified speaker (UNKNOWN): it.
- 2:51
- Robert Leshner: For anyone who missed the news two weeks ago, A16Z is no longer A16Z crypto. They are A16Z Frontier.
- 2:59
- Unidentified speaker (UNKNOWN): What's
- 3:00
- Unidentified speaker (speaker_2): to be clear? A16Z crypto, not A6 and Z. A16Z crypto became A6 and Z Frontier. Yes. It seems like the way in which they expanded their mandate is that it's not just crypto, but it still includes crypto, Uh, but will now include things. like new computing paradigms, agent to agent, something, something,
- 3:20
- Unidentified speaker (speaker_2): something. So it's kind of like the grab bag of cool futuristic stuff that falls through the cracks at the other A6 and Z funds. That's my understanding. But they do have an AI infrastructure fund, they do have um a biotech fund. So presumably it's not supposed to be
- 3:35
- Unidentified speaker (UNKNOWN): yeah,
- 3:36
- Unidentified speaker (speaker_2): infra or biotech, but
- 3:38
- Unidentified speaker (UNKNOWN): in other
- 3:38
- Unidentified speaker (speaker_2): another words, looking way
- 3:38
- Unidentified speaker (UNKNOWN): at the
- 3:38
- Unidentified speaker (speaker_2): to look at it, it's whatever Chris Dixon wants to do. So that's the Sounds
- 3:43
- Unidentified speaker (UNKNOWN): like you're giving
- 3:43
- Unidentified speaker (speaker_3): like a divorce talk. It's still going to include you, but it's
- 3:46
- Unidentified speaker (UNKNOWN): gonna
- 3:46
- Unidentified speaker (speaker_3): include, you know, other people that Daddy wants to meet. And uh, you know
- 3:51
- Unidentified speaker (UNKNOWN): it
- 3:51
- Unidentified speaker (speaker_3): should be fun. It should be a big group.
- 3:56
- Unidentified speaker (speaker_2): Yeah. It hits a little close to home, I have to say.
- 3:59
- Unidentified speaker (speaker_1): I sorry,
- 4:00
- Unidentified speaker (UNKNOWN): wow,
- 4:00
- Unidentified speaker (speaker_1): that was that was like
- 4:01
- Unidentified speaker (UNKNOWN): I was like what
- 4:05
- Unidentified speaker (speaker_2): Chris Dixon. You know, it's um
- 4:08
- Unidentified speaker (UNKNOWN): I
- 4:09
- Unidentified speaker (speaker_2): mean how how how do you how did you guys feel? Not not okay, ha analyzing what it means for Chris, the industry. How did you feel inside seeing that happen?
- 4:19
- Unidentified speaker (UNKNOWN): I
- 4:19
- Robert Leshner: think it's a sign of the times, right? Do you remember when Paradigm two years ago or so decided to
- 4:27
- Unidentified speaker (speaker_2): That was 22. That was 2022 when they did that.
- 4:29
- Unidentified speaker (UNKNOWN): No
- 4:30
- Unidentified speaker (speaker_1): 23, 23. Oh,
- 4:31
- Unidentified speaker (speaker_2): sorry, 23,
- 4:31
- Unidentified speaker (UNKNOWN): 23.
- 4:32
- Unidentified speaker (UNKNOWN): Right after
- 4:33
- Unidentified speaker (speaker_2): Chat GPT. Right after Chat GPT, they did that.
- 4:35
- Robert Leshner: Right. Which, by the way, ahead of the times. Okay.
- 4:38
- Unidentified speaker (speaker_2): Incredible in retrospect, incredible timing that we bullied them out of. Correct. Would have just absolutely nailed any AI investments in that vintage.
- 4:48
- Unidentified speaker (UNKNOWN): Correct.
- 4:49
- Robert Leshner: And
- 4:49
- Unidentified speaker (UNKNOWN): they were
- 4:49
- Unidentified speaker (speaker_2): booed off stage.
- 4:50
- Unidentified speaker (UNKNOWN): For
- 4:50
- Robert Leshner: any anyone who doesn't remember the history and the lore, Paradigm dipped their toe in the water and said, we are no longer a crypto firm, we are a crypto and AI firm. And Twitter lost its mind,
- 5:04
- Robert Leshner: relentlessly and mercilessly bullied them.
- 5:08
- Unidentified speaker (UNKNOWN): And
- 5:09
- Robert Leshner: then they said, okay, we're a crypto firm.
- 5:13
- Robert Leshner: And when I compare the A6. A16Z crypto announcement, that's my frame of reference.
- 5:19
- Unidentified speaker (UNKNOWN): Was
- 5:20
- Robert Leshner: paradigm getting mercilessly bullied? And when A16Z crypto did it three years later, no bullying, no jokes. Everybody said, sign of the times.
- 5:31
- Unidentified speaker (UNKNOWN): Yeah,
- 5:31
- Unidentified speaker (speaker_2): everybody's just like, it makes sense. Makes
- 5:33
- Unidentified speaker (UNKNOWN): sense.
- 5:35
- Unidentified speaker (UNKNOWN): Smart move, good for them.
- 5:36
- Unidentified speaker (speaker_2): Yeah. Yeah. Yeah. Yeah. You
- 5:37
- Unidentified speaker (UNKNOWN): may
- 5:38
- Robert Leshner: as well dabble
- 5:38
- Unidentified speaker (UNKNOWN): in the
- 5:39
- Robert Leshner: frontier.
- 5:40
- Unidentified speaker (speaker_2): That's right. That's right. Well, um, the the big the big change in the frontier, of course, has been the blue wave that we saw during the midterm.
- 5:48
- Unidentified speaker (speaker_1): Tsunami.
- 5:49
- Unidentified speaker (speaker_2): Tsunami. Sorry, mega tsunami, actually. There's a there's a there's a categorization of there's like a market for was it a mega tsunami? And the market ended
- 5:56
- Unidentified speaker (UNKNOWN): up
- 5:57
- Unidentified speaker (speaker_2): uh resolving yes. So um we saw just a total shellacking of reports. Republicans in the midterms, largely anticipated by both polling and by prediction markets. Um I was just looking at some stats.
- 6:09
- Unidentified speaker (speaker_1): But by one much earlier than the other.
- 6:12
- Unidentified speaker (speaker_2): Yes. So if you look at uh Polymarket published some stats, of course we're investors in Polymarket. Polymarket published that uh 419 of 426 races, the favorite on Polymarket end up winning. Basically by 7 seven p.m. on a midterm night, they'd called it ninety nine percent that there was going to be
- 6:32
- Unidentified speaker (speaker_2): total control of Congress. Tom, go ahead.
- 6:35
- Unidentified speaker (speaker_3): Yeah, I was gonna say, like a at a journalist friend text me last night at like eight o'clock. It was like, Hey, are you watching the midterms? Are you glued to the results? I'm like, what what results? It's it's over. Like if you're watching the prediction market, like it's it's over. There's no more information to consume. It was basically done at like eight.
- 6:50
- Unidentified speaker (UNKNOWN): The only thing
- 6:52
- Unidentified speaker (speaker_1): Only thing everyone fucked up was Mississippi. But honestly, that vote vote isn't still fully tallied, and it's like 50 votes delta. So it's like, I don't know. It's kind of a that one I feel like was one of the ones that like
- 7:06
- Unidentified speaker (UNKNOWN): all
- 7:06
- Unidentified speaker (speaker_1): these people, like, oh, prediction markets, they weren't so Aren't good. They couldn't predict Mississippi. And I'm like,
- 7:11
- Unidentified speaker (UNKNOWN): it
- 7:12
- Unidentified speaker (speaker_1): was an election won by 50 votes. Like it's like kind of hard to hard to hold. I don't think anything is quite at that precision for a human behavioral thing.
- 7:22
- Unidentified speaker (speaker_2): Yeah. So it was a fine night for prediction markets. I think uh I saw many people
- 7:26
- Unidentified speaker (UNKNOWN): on the time
- 7:27
- Unidentified speaker (speaker_2): saying, like, wow, prediction markets called it. Everybody who was uh hand-wringing about oh polling error, blah, blah, blah, blah. You know, maybe the markets can't be trusted. Well,
- 7:37
- Unidentified speaker (UNKNOWN): there was
- 7:37
- Robert Leshner: polling error, trust me. There was polling.
- 7:39
- Robert Leshner: error. The polls all had Democrats forecast for a very good night. The polls did not anticipate mega tsunami, right? The polls did not indicate that the Democrats would take deep, deep, deep, deep, deep red Senate seats.
- 7:58
- Robert Leshner: So the polls were off tremendously, and they understood. underweighted the Democrats. There was a lot of talk about oh, the polls are bogus, they're gonna overweight the Democrats. No, they underweighted the Democrats. The polls were significantly off. Yet again, I think the prediction markets
- 8:17
- Unidentified speaker (UNKNOWN): were
- 8:17
- Robert Leshner: fortunate that the direction of the polling error was in the direction of where
- 8:24
- Unidentified speaker (UNKNOWN): the
- 8:25
- Robert Leshner: polls would have had people predicting. Like if you took the polls at face value, you would have
- 8:29
- Unidentified speaker (UNKNOWN): been
- 8:29
- Robert Leshner: like, oh, the Democrats have a pretty good chance to sweep, right? Oh, this Senate race, pretty good chance for the Democrats. Most of the Senate seats were like,
- 8:38
- Unidentified speaker (UNKNOWN): you
- 8:38
- Robert Leshner: know, 70 to 80% odds for the Democrats going into it.
- 8:43
- Robert Leshner: So there wasn't a prediction market error because it was aligned with the polls, but the polls, the magnitude was way off still. And therefore, I also think the prediction markets underpriced the true odds in that case, because they were quite well aligned with the polls.
- 9:01
- Unidentified speaker (speaker_2): Yes. The the well, so I I'd push back a little bit because it's like, okay, the true odds according to what information, right? Was this information somewhere accessible that you could have known that the margin would be this deep? Or was this in fact the best picture of all the information that was available to everybody, uh, including people with private
- 9:21
- Unidentified speaker (speaker_2): polling data?
- 9:22
- Unidentified speaker (UNKNOWN): Not
- 9:22
- Unidentified speaker (speaker_3): to derail this too much, but I think um I think obviously for US midterms, markets saw a lot of volume, liquidity is pretty good. I think the other election recently that was very strange was obviously the Brazilian election, um, where prediction markets, as well as just like the public market, was like very strong. strongly pricing in Bolsonaro winning. And then Lula won
- 9:42
- Unidentified speaker (UNKNOWN): by
- 9:42
- Unidentified speaker (speaker_3): like 3%. And so I think a lot of people were like saying, oh, you know, it's you're supposed to win 20% of the time if it's a 20% chance. But like that felt like a scenario where like it was huge uh swing, where it wasn't like scraping by by like a few bips. It was like a 3%
- 9:56
- Unidentified speaker (speaker_3): you know, margin. And obviously the public markets also were acting in terms of like how Brazilian stocks and banks were sort of trading. And so I don't know. That that was kind of a strange one for me where I think we've seen globally predictive markets be really good in different regional elections. Like a lot of the same kind of patterns apply. And so it was it was surprising me that was so off for for for this particular election.
- 10:16
- Unidentified speaker (UNKNOWN): Uh
- 10:16
- Unidentified speaker (speaker_2): well, at least for the midterms, I mean we have a lot of independent events that we can calibrate the midterms on with with a lot of betting volume amongst each of them. And I think the answer is that at least for US politics, it came out very, very resoundingly in favor of prediction markets being very good at being able to anticipate what happens in these elections. And to your point, Tom, the I was I tweeted out uh I think
- 10:36
- Unidentified speaker (speaker_2): like at 6 30 or like 7 p.m. prediction markets were already showing 99% that uh it was going to be a clean sweep. And you know, you read the New York Times, you read the Wall Street Journal, you
- 10:46
- Unidentified speaker (UNKNOWN): read
- 10:46
- Unidentified speaker (speaker_2): like any of the mainstream news. I mean,
- 10:47
- Unidentified speaker (UNKNOWN): and they're all like, oh, it's a horse race
- 10:49
- Unidentified speaker (speaker_2): still going. It
- 10:49
- Unidentified speaker (speaker_1): was 80% queen sweep. clean sweep for like a week and a half before.
- 10:54
- Unidentified speaker (speaker_2): Yes. Yes.
- 10:55
- Unidentified speaker (UNKNOWN): It was like I think it was like
- 10:56
- Unidentified speaker (speaker_2): 70.
- 10:57
- Unidentified speaker (speaker_1): High 70s, though. Like,
- 10:58
- Unidentified speaker (UNKNOWN): yeah. And by the
- 10:59
- Robert Leshner: way, individual like Senate races underpriced it, right? So like there was a number of races where like the Democrats were at like Mississippi. Like Mississippi was at like you know seven 7%, percent, right? Before going to 100%, right? There was a number of races that the prediction markets were off just because no one thought the wave was going to be this big, right? Right.
- 11:20
- Robert Leshner: On the aggregate,
- 11:21
- Unidentified speaker (UNKNOWN): the
- 11:21
- Robert Leshner: prediction markets were actually very accurate, right? In the aggregate. And when you said, like, okay, at a national level, was it close? Like, yeah, it was very close. Like the prediction markets seemed convinced that there was going to be a wave. The prediction markets seemed convinced that the Democrats would take the House and the Senate. If you looked at the microstructure of it, though, like
- 11:41
- Robert Leshner: individual races in like deep red states, like,
- 11:44
- Unidentified speaker (UNKNOWN): of
- 11:44
- Robert Leshner: course they didn't predict Mississippi going to the Democrats.
- 11:48
- Unidentified speaker (speaker_2): Yeah. I mean, it's it's crazy because I I grew up in Texas. So seeing Texas go completely blue is um really bizarre. Uh almost every single race in Texas came up blue, which I think it's like 40 years or something. since that has happened. It's just like a completely different map in a way that you see
- 12:08
- Unidentified speaker (speaker_2): in the same way that Democrats were reckoning with what happened in 24 as being like an existential party level moment of like, what do we do? How do we like read how do we reposition ourselves among the electorate? It feels like suddenly last night was that for Republicans.
- 12:27
- Unidentified speaker (UNKNOWN): Yeah.
- 12:27
- Robert Leshner: And by the way, I don't think the Democrats really repositioned themselves. I think Trump dragged down the Republican ballot into the nether regions of hell.
- 12:36
- Unidentified speaker (UNKNOWN): I
- 12:36
- Unidentified speaker (speaker_1): mean, I I will say I was pretty I thought it we're talking about all of this. But then if you read the that J V Dance, sorry, J D Vance, I I I can't I have the spoonerism from all these jokes I kept reading about.
- 12:47
- Robert Leshner: JV Dance?
- 12:49
- Unidentified speaker (speaker_1): Yeah. That was what uh I think uh Al Sayed called him on the campaign trail. But uh but he basically he had this thing where it's like all the data we saw told us. Like, I don't know if you read his
- 13:02
- Unidentified speaker (UNKNOWN): data as of yesterday. All the
- 13:03
- Unidentified speaker (speaker_1): data as of it, and I'm like, we're talking about this prediction market stuff. It's like the prediction market fucking said this like for months. Like, man, like what are you talking about? Like, I there was there's a little bit of like people just like suddenly realizing they lived in like, as I saw in another meme or another tweet, a chud bubble, and like they just had to had to get out of it.
- 13:23
- Unidentified speaker (speaker_2): It seemed like that message was clearly him positioning for twenty eight, which is that he now realized, and everybody in the Republican Party is is is dawning on them now that Trump is toxic. His brand is only going to drag them down. Every single candidate he handpicked lost. pretty resoundingly. And of course, just you know, down ballot, you have all these races that
- 13:43
- Unidentified speaker (speaker_2): really should not have gone to Republicans, or sorry, it should not have gone to Democrats that went blue, uh, many for the first time in in many, many decades. So this is a very clearly an ultimatum on Trump, on the Iran war, on inflation. And it's it felt like almost less of an embrace of Democrats. Uh, I was reading a bunch of uh polling, uh
- 14:03
- Unidentified speaker (speaker_2): exit polling, saying that people still don't like the Democratic Party that much, but they really felt like this was a response to Trump and to Trumpism. And anybody who kowed out to Trump, they were like, screw you, that is not what we want to see. Um, and now we're gonna have a very combative legislature against Trump. Now, for us, uh go ahead,
- 14:23
- Unidentified speaker (speaker_2): Robert.
- 14:23
- Unidentified speaker (UNKNOWN): I
- 14:24
- Robert Leshner: was gonna say this is a crypto show. The one thing I did. judge Trump on is did he make America the crypto capital of the world? And the answer is not really.
- 14:34
- Unidentified speaker (UNKNOWN): Yeah.
- 14:35
- Unidentified speaker (speaker_2): Not really. Okay. Well
- 14:37
- Unidentified speaker (speaker_1): If
- 14:37
- Unidentified speaker (UNKNOWN): we didn't
- 14:37
- Unidentified speaker (speaker_1): launch Trump coin, I think there was a much higher chance. Like I think that's how
- 14:42
- Unidentified speaker (speaker_1): the meme coins basically fucked everything.
- 14:44
- Unidentified speaker (speaker_2): Well, corruption was one of the leading issues in Exit Pulse.
- 14:47
- Unidentified speaker (speaker_1): Yeah.
- 14:47
- Unidentified speaker (speaker_2): Which again is also surprising because corruption does not rank. rate
- 14:49
- Unidentified speaker (UNKNOWN): the colour. Also also
- 14:50
- Unidentified speaker (speaker_1): announced in the first hundred day uh hundred hour policy of the Democrats that they put out today was like one of the top five issued things was going out just subpoenaing all the Trump.
- 15:02
- Unidentified speaker (UNKNOWN): And they're
- 15:02
- Robert Leshner: already they're already talking about
- 15:05
- Unidentified speaker (UNKNOWN): the
- 15:05
- Robert Leshner: legislative side of this. Like Maxine Waters is already floating new draft legislative text for a revamped version of crypto regulatory. clarity acts, whatever you want to call it. It'll come back as the AML bill, like re-crypto. But it it's still on the conversational agenda. Whether that's a good or
- 15:25
- Robert Leshner: a bad thing, unclear. But
- 15:29
- Robert Leshner: Trump, his involvement in crypto, the unfinished business of the last session, all of that like continues into the next session, which is going to be dominated by the Democrats.
- 15:39
- Unidentified speaker (speaker_2): Okay, so let's let's zoom out and just get a sense of what this means for crypto specifically, at least in the legislature. So stand with crypto, which rates candidates on their crypto friendliness. They saw, you know, all the turnover in the House and the Senate. 226 pro-crypto candidates came into the House, 106 anti-crypto. It's actually a very good
- 15:59
- Unidentified speaker (speaker_2): margin that is pro-crypto pro crypto in the House composition for the Senate. 19 pro crypto, eight anti-crypto. So again, basically a two to one margin in terms of crypto friendliness, both in the House and the Senate. Now, Sherrod Brown, who is a very famous enemy of the crypto industry, he he was pushed out previously by FairShake Pack. He ended up winning an election
- 16:19
- Unidentified speaker (speaker_2): where Fair Shake spent tens of millions of dollars to try to
- 16:22
- Unidentified speaker (UNKNOWN): try to oust him and
- 16:23
- Unidentified speaker (speaker_2): prevent him. So
- 16:24
- Unidentified speaker (UNKNOWN): I
- 16:24
- Unidentified speaker (speaker_2): think it was tw double that. Uh
- 16:26
- Unidentified speaker (UNKNOWN): yeah, I think at
- 16:27
- Robert Leshner: this cycle it was like forty million.
- 16:29
- Unidentified speaker (speaker_2): Okay, yeah. They spent a lot to try to uh try to beat him. Sorry, yeah, it's forty forty million is what I'm seeing. And uh he ended up getting back in. Uh and
- 16:36
- Unidentified speaker (UNKNOWN): everyone got upon his
- 16:38
- Unidentified speaker (speaker_2): day. Right, right. Um and so you know, crypto spent a lot of money on the
- 16:42
- Unidentified speaker (UNKNOWN): Hailey Stevens.
- 16:44
- Unidentified speaker (speaker_1): Sorry, that was slightly delayed joke, but
- 16:48
- Unidentified speaker (speaker_2): okay. So we have some wins in terms of what the composition of the legislature looks like. And I think the reality is that the legislature is now younger. I think it's more technology oriented, and there's a lot of pro crypto people who have come into this new Congress. So I don't think we should focus entirely on Trump. There is going to be a post-Trump era where, you know,
- 17:08
- Unidentified speaker (speaker_2): it's it's not going to be entirely about corruption and the Trump family and so on, but it's very clear next two years, the main theme is going to be Trump corruption, World Liberty Financial, Trump coin, all this stuff is going to be repeatedly blaring through TV and through hearings. So Robert, let me start with you. How how
- 17:28
- Unidentified speaker (speaker_2): do you think this plays out? On the one side, a relatively good fact pattern in terms of what this new Congress looks like with respect to their crypto friendliness. Other side is that like Trump corruption rates and it's gonna rate for the next two years.
- 17:42
- Robert Leshner: Yeah, I agree. I mean, in the aggregate, the silver lining for the industry, right, is as you said, a lot of pro-crypto members of the House and Senate were elected, right? Why is that the case? Because it was such a ginormous, insane level of suit blue
- 18:01
- Robert Leshner: tsunami. right? In all of the red districts, they were running moderates for the most part, right? Because like, why even like throw a you know far left candidate in the mix if you know
- 18:14
- Unidentified speaker (UNKNOWN): it's
- 18:14
- Robert Leshner: gonna be like a red district? It's like, no, you you run someone closer to the center. So there's a lot of candidates much closer to the center, who are much more moderate, on the whole, pro like pro-crypto. I mean, television Taler Rico is pro-crypto. He's like a Bitcoin guy, right? Like there's a lot of people that would surprise you who made it through, right? So this is not,
- 18:34
- Robert Leshner: there's a lot less far left seats as a proportion of the electorate than you might guess from last night's election. There's a lot of moderates who've come in.
- 18:44
- Unidentified speaker (UNKNOWN): Oh,
- 18:44
- Unidentified speaker (speaker_1): yeah. I forgot Talarico Talerico voted on the Bitcoin reserve. Yeah,
- 18:47
- Robert Leshner: in Texas. Yeah. He
- 18:48
- Unidentified speaker (speaker_1): was like, yeah, that's right.
- 18:49
- Robert Leshner: Yeah, yeah. So like I actually think this is like gonna be like potentially, once you get past the dirty laundry and all of that, like this is potentially a really pragmatic
- 19:03
- Robert Leshner: Congress. Yes, there's still unfinished business with the Clarity Act. You know, Maxine Waters wants crypto, Elizabeth Warren wants a crypto bill that regulates crypto into obsolescence, right? Like
- 19:15
- Unidentified speaker (UNKNOWN): I
- 19:15
- Robert Leshner: assume Trump will veto any legislation that somehow gets approved, right? The next two years I think is going to be a lot of posturing. It's going to be like trying to identify the contours of the electorate, the contours of the legislative body, positioning for 2028. I don't think anything's getting signed assigned into law. I think the chances that we get
- 19:35
- Robert Leshner: any legislation, basically zero.
- 19:39
- Robert Leshner: I hope that they can work on bipartisan things. Trump has said, like, I'd love to work on Democrats with certain things, but like, let's see.
- 19:48
- Robert Leshner: That being said,
- 19:50
- Robert Leshner: there are a number of presidents who have gotten completely shellacked.
- 19:56
- Unidentified speaker (UNKNOWN): You know,
- 19:57
- Robert Leshner: I was listening to a historian his story this morning while I was at the gym on the train. trim. There have been a number of presidents that have gotten completely.
- 20:03
- Robert Leshner: shellacked prior to the last two years of their term. Clinton, Obama, right? This happens. Presidents who were in the 30 range of approval ratings at temporary lows in their presidency, and who listened and who got the message and
- 20:23
- Robert Leshner: who course corrected. And like when Clinton and Obama left. their terms, they were almost at all-time high approval ratings. They basically said, I messed up, messed up bad. That's not what America wants. Let's pivot hard, right? There's a chance that Trump pivots hard and says, you know what? Sure, ethics provisions and a crypto bill. Let's do it. Like
- 20:43
- Unidentified speaker (speaker_1): I love that you're optimistic that Donald Trump believes in mean reversion.
- 20:48
- Robert Leshner: You know what? I don't know. I am open
- 20:51
- Unidentified speaker (UNKNOWN): minded,
- 20:52
- Robert Leshner: right?
- 20:52
- Unidentified speaker (speaker_2): I mean I it's Trump you're talking about, man.
- 20:58
- Unidentified speaker (speaker_2): Yeah,
- 20:58
- Unidentified speaker (speaker_1): it's
- 20:59
- Unidentified speaker (UNKNOWN): like
- 20:59
- Unidentified speaker (speaker_3): if anything,
- 21:00
- Unidentified speaker (UNKNOWN): he's the mean
- 21:00
- Unidentified speaker (speaker_3): reverting president. How many presidents you know have have a gap and still serve
- 21:05
- Unidentified speaker (UNKNOWN): all
- 21:05
- Robert Leshner: I'm saying is the door is not closed
- 21:09
- Unidentified speaker (UNKNOWN): to
- 21:09
- Robert Leshner: things happening
- 21:10
- Unidentified speaker (UNKNOWN): in
- 21:10
- Robert Leshner: this term that will surprise you.
- 21:12
- Unidentified speaker (speaker_2): I
- 21:12
- Unidentified speaker (UNKNOWN): think his legacy is
- 21:13
- Unidentified speaker (speaker_2): still talking about the reform. reflecting pool. Dude,
- 21:16
- Unidentified speaker (UNKNOWN): like
- 21:16
- Unidentified speaker (speaker_2): he gave a speech today. He's
- 21:18
- Unidentified speaker (UNKNOWN): still
- 21:18
- Unidentified speaker (speaker_2): talking about fucking White House renovations.
- 21:20
- Robert Leshner: The status quo is that his legacy is gonna get worse and worse and worse through the end of the term. And his approval rating will sink. But as somebody who does care about how they're perceived very Pretty much, I could see him potentially hard pivoting.
- 21:38
- Unidentified speaker (speaker_2): I mean, I I'd love to believe that. I would absolutely love to believe that. But like the the list of things, like crypto isn't even on the list of things he needs to hard pivot on, right? He needs to hard pivot on Iran, he needs to hard pivot on tariffs, he needs to like there's so many things that come before crypto. Like crypto is not even on the list, you know? And the reality is that he's so incredibly unpopular and so
- 21:58
- Unidentified speaker (speaker_2): energizing to 2028 that like 2028, they're still gonna be running on. Trump. You know, even you know, Trump, Trump's gone, it's gonna be you know Rubio or or JD Vance, and they're still gonna be talking about corruption and the Trump family and blah blah blah. Because like it's just so it just it rates, it's so animating,
- 22:16
- Unidentified speaker (UNKNOWN): you
- 22:16
- Unidentified speaker (speaker_2): know? And like I think I think what they learned from the midterms is that this is the message for the next two years.
- 22:22
- Robert Leshner: Let's see how the last tiers play out, because at this point, right now. Trump said he's going to veto everything. There's probably no expectation of legislative progress on any topic. It'll be a lot of posturing. It'll be a lot of investigations of Trump. It'll be a lot of impeachment
- 22:42
- Robert Leshner: of Trump. And he's going to continue to wield executive authority. He's going to, I mean, the things that are in his control. right?
- 22:52
- Robert Leshner: That are outside of the legislative process. It's like
- 22:55
- Unidentified speaker (UNKNOWN): the
- 22:55
- Robert Leshner: military, he dictates what happens in Iran. He has a lot of control over how all of the departments are run. He has, there's a lot of things Trump can do for good and for bad in the last two years. I doubt permanent legislation is going to be a part of it. I think there's going to be a lot happening in the courts.
- 23:15
- Robert Leshner: offense and defense there, but it's going to be
- 23:19
- Unidentified speaker (UNKNOWN): a
- 23:20
- Robert Leshner: sight to behold. So I think crypto's going to be quiet.
- 23:22
- Unidentified speaker (speaker_2): You know, it's it's interesting because the the first Trump administration, the last two years, when he was uh a lame duck,
- 23:30
- Unidentified speaker (UNKNOWN): uh
- 23:31
- Unidentified speaker (speaker_2): I I wasn't really paying as close attention to politics. And so I didn't really have a great sense of what he was able and not able to do when he was basically being railroaded by Congress. My sense of Trump is that he has a much better understanding now of how to make things happen.
- 23:46
- Unidentified speaker (UNKNOWN): I mean, you had a pandemic
- 23:47
- Unidentified speaker (speaker_1): in the last year, so it's a little much harder. It's very hard to assess how well.
- 23:54
- Unidentified speaker (speaker_2): Right, right. But when the impeachment
- 23:55
- Unidentified speaker (UNKNOWN): started, to
- 23:56
- Robert Leshner: operation warp speed, like they did shepherd, they did
- 23:59
- Unidentified speaker (UNKNOWN): have a lot of people. I think the
- 24:00
- Unidentified speaker (speaker_2): pandemic also opened the door to a lot. more things being being on the table and and doable because of what an extraordinary time it was. But I think I I I guess I don't have a great mental model of how how much do you think Trump will get slown
- 24:15
- Unidentified speaker (UNKNOWN): down
- 24:15
- Unidentified speaker (speaker_2): by this? He's obviously been very expansive in his reading of his abilities, what he can do with executive orders. I think he's going to be very aggressive with what he the marching orders he gives the agencies. Uh but there's now very united government against him. And I think the Republicans are also going to start fraying because they're going to realize that they need to break with Trump in order
- 24:35
- Unidentified speaker (speaker_2): to have a chance at 28. And so I kind of feel like he's in a worse position now than he was in this second part of his first term.
- 24:45
- Unidentified speaker (UNKNOWN): I mean,
- 24:45
- Unidentified speaker (speaker_1): that's
- 24:47
- Unidentified speaker (speaker_1): to be true. I mean, it the just even just pure congressional disposition. disadvantages was not as
- 24:53
- Unidentified speaker (UNKNOWN): is much cheaper
- 24:54
- Unidentified speaker (speaker_2): now. Yeah. Yeah. It's like potentially veto.
- 24:56
- Unidentified speaker (speaker_1): But congressional and state, right? Even the state stuff is like very against so like I don't know. This the this feels like you need Trump to like have gone to like the equivalent of like, you know, the the person who's who's like everything they're working on blew up and then they went on like a meditation retreat and they come back. and they're a totally different new
- 25:16
- Unidentified speaker (speaker_1): person and suddenly they're like, I want to be kinder and I want to like have less insults and like stop talking about guys' asses and like whatever, all this shit.
- 25:27
- Unidentified speaker (UNKNOWN): Well,
- 25:28
- Unidentified speaker (UNKNOWN): I don't know, I don't know if that's
- 25:29
- Unidentified speaker (speaker_3): assassination attempt if you remember that. And then uh yeah, yeah,
- 25:32
- Unidentified speaker (UNKNOWN): yeah. Back to normal. I feel like the
- 25:34
- Unidentified speaker (speaker_1): level of aura loss here is so high that it's like un un how do you U-turn? That's like not possible, I think.
- 25:41
- Unidentified speaker (speaker_2): Yeah. It's surprising also, he hasn't challenged any like after seeding all this election integrity stuff, he's been like, Oh yeah, yeah, that happened. Yeah, yeah,
- 25:49
- Unidentified speaker (UNKNOWN): no seems right.
- 25:50
- Unidentified speaker (speaker_2): All correct. Yep.
- 25:51
- Robert Leshner: Yeah.
- 25:51
- Unidentified speaker (speaker_2): Looks like, yeah, definitely. I lost everything.
- 25:54
- Unidentified speaker (UNKNOWN): I
- 25:54
- Unidentified speaker (speaker_3): don't
- 25:54
- Unidentified speaker (speaker_1): know.
- 25:54
- Unidentified speaker (speaker_2): This is a
- 25:55
- Unidentified speaker (speaker_1): guy who flip-flops. We still have until January 3rd. third. So like he could very easily still do something out of the little a little bit.
- 26:02
- Unidentified speaker (speaker_2): term.
- 26:03
- Robert Leshner: Yeah. I mean, the Republicans have two months, right? My prediction is they will completely
- 26:09
- Unidentified speaker (speaker_1): less
- 26:09
- Unidentified speaker (UNKNOWN): actually
- 26:10
- Robert Leshner: those two months. Accomplishly
- 26:12
- Unidentified speaker (UNKNOWN): actually
- 26:14
- Unidentified speaker (speaker_1): a bunch of the the Senate candidates are replacements.
- 26:17
- Unidentified speaker (UNKNOWN): So
- 26:17
- Unidentified speaker (speaker_1): they start in December.
- 26:20
- Unidentified speaker (speaker_2): So
- 26:21
- Unidentified speaker (speaker_1): that And it's it's kind of even more skewed in some sense to n to be nothing.
- 26:25
- Unidentified speaker (speaker_2): Yeah. Okay. Well, it's gonna be an interesting two years until now in the election. I guess the the last question I have for you guys is uh we saw some movement in the twenty twenty eight markets after looking at this election. If you look at prediction markets, it's now pricing sixty eight percent chance that the twenty twenty eight Victor will be Democrat. Uh we also saw
- 26:45
- Unidentified speaker (speaker_2): Osoff Ossoff move up quite a bit in the election odds. He's now the number one most likely candidate. And he's now leading AOC and Gavin Newsome, who are once the the front runners uh for being the Democratic candidate. So it seems like his aura his star has risen. And you know, JD is still the front runner on the Republican side, but it obviously
- 27:05
- Unidentified speaker (speaker_2): the Republican odds have on the whole contracted. Yeah, he is. Um so it's him and then Rubio is uh a few points behind uh and then it's kind of a pretty scattered field. Uh, but it seems like markets are pricing if it's a head to head with a Trump successor, that he's probably gonna get thumped in twenty eight
- 27:25
- Unidentified speaker (speaker_2): uh once again. So how do you guys think twenty twenty eight on is going to look if we get an Osof as president? Because I honestly don't know that much about Osof's views.
- 27:36
- Unidentified speaker (UNKNOWN): I
- 27:37
- Robert Leshner: don't want to talk about post post-2028. twenty twenty.
- 27:39
- Robert Leshner: I feel like we're just like
- 27:41
- Unidentified speaker (speaker_1): Dude, two years in crypto is forever. I like I don't yeah. What is
- 27:45
- Unidentified speaker (UNKNOWN): it?
- 27:45
- Robert Leshner: Okay. I'm
- 27:46
- Unidentified speaker (UNKNOWN): not political
- 27:47
- Robert Leshner: forecasters. Come
- 27:48
- Unidentified speaker (UNKNOWN): on.
- 27:48
- Unidentified speaker (speaker_2): Okay, okay, okay. I I I th I mean it it's something that I'm thinking about is like what is the you know, we're gonna have two years of pain, but what do we have after that? And uh
- 27:57
- Unidentified speaker (UNKNOWN): I
- 27:58
- Unidentified speaker (speaker_1): mean I I think the bigger question for Asof Osof is more the data center stuff than crypto because of all the joy. Georgia build-outs. So I'm not even sure it's like a top policy issue room.
- 28:07
- Unidentified speaker (speaker_2): Yeah.
- 28:07
- Unidentified speaker (UNKNOWN): I
- 28:08
- Unidentified speaker (speaker_1): think we're by the way, he voted in favor of genius cloture vote. So, like I and and
- 28:14
- Unidentified speaker (UNKNOWN): you know
- 28:14
- Unidentified speaker (speaker_1): he voted against clarity, but it sounded like he was pro-Genius. I don't know if that's
- 28:18
- Unidentified speaker (UNKNOWN): if
- 28:18
- Robert Leshner: every single Democratic senator voted against Clarity. So
- 28:21
- Unidentified speaker (speaker_1): Yeah, I know.
- 28:23
- Unidentified speaker (UNKNOWN): I
- 28:23
- Robert Leshner: I
- 28:23
- Unidentified speaker (UNKNOWN): think it's
- 28:24
- Robert Leshner: gonna be two years of boredom.
- 28:25
- Unidentified speaker (UNKNOWN): I think
- 28:25
- Robert Leshner: we're actually gonna stay. stop talking about DC on the chopping block for two years.
- 28:29
- Unidentified speaker (speaker_1): Thank fucking God.
- 28:31
- Unidentified speaker (speaker_2): It's time.
- 28:32
- Unidentified speaker (speaker_1): Jesus Christ.
- 28:33
- Robert Leshner: Close the book. Close the page.
- 28:35
- Unidentified speaker (UNKNOWN): I'm excited to
- 28:36
- Unidentified speaker (speaker_1): just like get a little less suit nonsense.
- 28:43
- Unidentified speaker (speaker_2): No,
- 28:44
- Robert Leshner: we're
- 28:44
- Unidentified speaker (UNKNOWN): done with we're
- 28:44
- Robert Leshner: done with DC for two years.
- 28:45
- Unidentified speaker (speaker_2): We're done with that? Okay. All right.
- 28:47
- Unidentified speaker (UNKNOWN): No, no, we we
- 28:47
- Unidentified speaker (speaker_1): we we we can talk about that. I think.
- 28:50
- Unidentified speaker (speaker_1): After
- 28:51
- Unidentified speaker (speaker_2): this show.
- 28:52
- Unidentified speaker (speaker_2): Well, I don't want to upset Tarun. I know he gets uh he gets triggered every time we bring on regulators.
- 28:58
- Unidentified speaker (speaker_1): It's just, you know, I I just get I just get tired. That's all I know.
- 29:03
- Unidentified speaker (UNKNOWN): You
- 29:03
- Unidentified speaker (speaker_2): just get tired. Oh, you poor thing. Okay, well, we'll we'll very gently, in a way that won't get Tarun too tired, discuss this news. So yesterday, on the day of the midterms, Mike Sealake, Sealig, CFTC chairman, put out a statement saying that CFTC has now cleared equity perps for trading in the US. Single stop perps can now trade on DCMs that
- 29:23
- Unidentified speaker (speaker_2): are SEC notice registered for security futures. Um this uh uh basically Sealig said, look, rather than being forced to turn to offshore and unregulated venues, you can now bring this uh into the US. Within hours, SIBO and Citadel sued. Uh one once again trying to claim that these things are swaps and uh that these should
- 29:43
- Unidentified speaker (speaker_2): be limited in their allowance to be traded. And of course, many people pointed out that uh Hyperliquid, which uh recently entered into this partnership with ICE, might be one of the beneficiaries of this uh this new ruling by the CFTC. Tarun, what's your take on equity perps being waived into the US and challenged?
- 30:02
- Unidentified speaker (speaker_1): I mean, I think the challenge is a little bit weird to me in one way, in that um
- 30:07
- Robert Leshner: Why?
- 30:08
- Unidentified speaker (UNKNOWN): Well,
- 30:08
- Unidentified speaker (speaker_1): I don't know if I'm Citadel, because like I know they're part of this consortium or Citadel and other people.
- 30:18
- Unidentified speaker (speaker_1): That's like that I didn't touch. It just seems like for Citadel SIDEL securities, hedge fund, I don't know, but for Citadel SIDL Securities, I'm like, this seems like a great new business. to get into. So like I don't really understand the consortium members' goals. There might be some other real reason, which is like, which I was trying to think about, which was like rebates
- 30:38
- Unidentified speaker (speaker_1): on stock trade on exchanges, stock trading exchanges, and rebates on commodity exchanges are different. And you maybe you model out that actually with funding rates plus all the fees, fee dealt differences, you actually make less. money if single stock exists. If you're trying to do the single stock perp versus spot arb. I I don't know. I was just trying to come up with some
- 30:58
- Unidentified speaker (speaker_1): I I naively, I'm just like it feels like a new market. You can quote a remote.
- 31:02
- Unidentified speaker (speaker_2): Maybe it's just a principled stand that they believe in the letter of the law.
- 31:06
- Unidentified speaker (UNKNOWN): Have you
- 31:06
- Unidentified speaker (speaker_2): ever
- 31:06
- Unidentified speaker (UNKNOWN): gotten it? But it's like if you're
- 31:07
- Unidentified speaker (speaker_1): a market maker, like you're just like usually like, oh great new market. I'm gonna like try to make
- 31:11
- Unidentified speaker (UNKNOWN): it. True. And I think
- 31:12
- Unidentified speaker (speaker_2): Citadel is trying to protect investors. I think that's what's going on.
- 31:15
- Unidentified speaker (UNKNOWN): here.
- 31:15
- Unidentified speaker (speaker_1): Okay. I I don't
- 31:17
- Unidentified speaker (UNKNOWN): know. I
- 31:17
- Unidentified speaker (speaker_1): don't know. I don't know exactly, but like my intuition is just like if I'm a market maker, this sounds awesome.
- 31:22
- Robert Leshner: I I think
- 31:25
- Robert Leshner: Citadel and SIBO and those who are suing,
- 31:30
- Robert Leshner: if anything, just want to slow this down so they have more time to catch up and prepare. Right. This is happening very quickly.
- 31:37
- Unidentified speaker (UNKNOWN): I
- 31:37
- Robert Leshner: think you're correct, Terrell. Tarun. They would all love to participate.
- 31:41
- Unidentified speaker (speaker_1): They would trade tomorrow if it was live, right? Like
- 31:43
- Unidentified speaker (UNKNOWN): if they
- 31:44
- Robert Leshner: were ready. They don't even they're not ready, right? Like they need to slow this down so they can like understand perps.
- 31:51
- Unidentified speaker (speaker_1): Wow, this is like this is like the perp slowdown, like the AI slowdowns.
- 31:54
- Unidentified speaker (UNKNOWN): Yes.
- 31:57
- Unidentified speaker (UNKNOWN): Stop the stop the the frontier. frontier of Stop the pacing. This is
- 32:00
- Robert Leshner: pacing. This This is is pacing pacing the frontier of crypto.
- 32:03
- Unidentified speaker (UNKNOWN): Beautiful.
- 32:04
- Unidentified speaker (speaker_2): The
- 32:04
- Unidentified speaker (UNKNOWN): other
- 32:04
- Unidentified speaker (speaker_3): thing they don't understand. I'm like uh That's
- 32:06
- Unidentified speaker (UNKNOWN): what I'm saying.
- 32:08
- Unidentified speaker (UNKNOWN): Like
- 32:08
- Unidentified speaker (speaker_3): I don't believe that. I
- 32:08
- Unidentified speaker (UNKNOWN): don't
- 32:08
- Unidentified speaker (speaker_3): know what
- 32:09
- Unidentified speaker (UNKNOWN): to do.
- 32:10
- Unidentified speaker (UNKNOWN): I
- 32:10
- Unidentified speaker (speaker_1): don't believe that either. I feel like they could trade it tomorrow. Like if it was legal.
- 32:17
- Unidentified speaker (UNKNOWN): I don't
- 32:17
- Unidentified speaker (speaker_1): know if they trade Binance perps, though. Like I feel like they trade Coinbase perps, maybe, which is a very low liquidity. And they trade few.
- 32:25
- Unidentified speaker (UNKNOWN): I mean
- 32:26
- Unidentified speaker (speaker_2): Is that
- 32:26
- Unidentified speaker (UNKNOWN): that's different? Like they're big to big critical
- 32:28
- Unidentified speaker (speaker_3): investment. Yeah, but I'm I'm like it's not like they need to like understand perps.
- 32:31
- Unidentified speaker (UNKNOWN): Yeah that's I that's where I
- 32:33
- Unidentified speaker (speaker_1): don't believe that there's a reason to slow it down. So like I I'm sort of I I do think like if I remember correctly the Citadel even in the previous kind of don't bring perps on shore thing they were involved. So they they do have some motivation that I have just I personally haven't been able to divine because
- 32:50
- Unidentified speaker (UNKNOWN): all
- 32:50
- Unidentified speaker (speaker_1): the other market makers, anyone I talk to is like, yeah, we fuck we fucking want this. We'll move all our hyperliquid volume to the US if you do if this happens, right?
- 32:58
- Unidentified speaker (UNKNOWN): Or
- 32:59
- Unidentified speaker (speaker_1): like a lot of the people who are trade US firms trading on hyperliquid, they have to have like multiple entity setups and it's kind of annoying and whatever. And they're like, Oh, now we can do this way more easily. Why would we not do this? So, like everyone seems very positive. So it's weird to like SIBO, I get, right? Totally like, hey, that that's like a land, you know, I want, don't, don't
- 33:19
- Unidentified speaker (speaker_1): eat, don't touch my turf type thing. I'm just like a little surprised with a subdult thing. It just it it I'm sure there's a reason somewhere.
- 33:28
- Unidentified speaker (speaker_2): Fair enough. Toby have a take?
- 33:31
- Unidentified speaker (speaker_3): Uh I mean, I I think similar to Tarune. I uh I did see I think Cal Calci She launched, they're not necessarily perps, but they
- 33:37
- Unidentified speaker (UNKNOWN): are
- 33:37
- Unidentified speaker (speaker_3): like singling single-lane US US equity. equities like futures that trade within market hours for like a few single-lame you know uh um US equity,
- 33:46
- Unidentified speaker (UNKNOWN): like
- 33:46
- Unidentified speaker (speaker_3): Costco and McDonald's and stuff. And like classically, it's like you know, doing like 15 mil volume on like 100k in open interest. But like I don't know, maybe things these things will be be popular at some point. I um no, I I think uh, you know, like Teren, Taroon, there's uh our forefathers, you know, wrote into the constitution. a very clear distinction between a future and a swap.
- 34:06
- Unidentified speaker (speaker_3): And uh we have to respect what the no, I'm I'm just I'm just fucking I'm I'm playing with you. Like no, I've I have no idea what the you know opposition could be.
- 34:13
- Unidentified speaker (speaker_2): Fair enough. Well, um, you know, it after the midterms we saw a bunch of stuff go up. We saw markets go up, NASDAQ was up like one 1.8%. point eight percent, crypto was up, Bitcoin was up three, b Ethereum was up six. I guess Ethereum is the blue one. Um, I suppose. Uh, but we also saw anthropic stock in the pre-market
- 34:33
- Unidentified speaker (speaker_2): futures slash perps, uh, also up from the Democrats uh potentially winning in the midterms. And there was also some updates to Anthropic's S1. They announced their Q3 revenues, News, which was 17 billion versus 11.5 11 and a half billion they made in Q2. So a fifth rough. above 50% quarter over quarter
- 34:52
- Unidentified speaker (speaker_2): increase. Uh they are back to an operating loss. So they were uh operating profit in Q2, operating loss uh very slightly in Q3. And they added some new risk factors, including RSI. RSI meaning recursive self-improvement means the ability for an AI model to improve a future generation of the AI model and potentially
- 35:12
- Unidentified speaker (speaker_2): causing some accelerating loop. of self-improvement that uh results in, quote, dramatically accelerating progress.
- 35:20
- Robert Leshner: How is that a risk? That sounds so positive for the business.
- 35:26
- Unidentified speaker (UNKNOWN): Well,
- 35:27
- Unidentified speaker (speaker_2): it is
- 35:27
- Unidentified speaker (UNKNOWN): a very simple thing.
- 35:32
- Unidentified speaker (speaker_2): Well,
- 35:33
- Unidentified speaker (UNKNOWN): we all have
- 35:33
- Unidentified speaker (speaker_2): risks of getting regulated now. Yes, that's right. I mean, you know, with this blue wave, it's interesting to me that like you see in the anthropic perps, anthropic's clearly up over the last 24 hours. Surprising because so many of the states that ended up flipping were like the data center states. Well, and
- 35:48
- Unidentified speaker (UNKNOWN): they
- 35:48
- Unidentified speaker (speaker_2): went blue.
- 35:49
- Robert Leshner: This is actually why. So some of my hedge fund friends were talking about this exact topic.
- 35:55
- Unidentified speaker (UNKNOWN): And
- 35:55
- Robert Leshner: the logic that they were using goes something like this:
- 36:00
- Robert Leshner: New
- 36:01
- Robert Leshner: capacity is going to come. online slower than we previously thought, right?
- 36:07
- Robert Leshner: This disadvantages neo clouds. This theoretically advantages the existing hyperscalers who have already built tons of data centers, whether they fully retrofit that for new AI use cases or
- 36:21
- Unidentified speaker (UNKNOWN): not.
- 36:23
- Robert Leshner: In some ways, it directly benefits.
- 36:27
- Robert Leshner: the competitive dynamic of the frontier companies in the aggregate. Because if there's less
- 36:35
- Robert Leshner: compute to go around
- 36:37
- Unidentified speaker (UNKNOWN): and
- 36:37
- Robert Leshner: less inference to go around, then
- 36:41
- Unidentified speaker (UNKNOWN): they
- 36:41
- Robert Leshner: don't have to spend as much money. Their commitments are lower.
- 36:45
- Unidentified speaker (UNKNOWN): If
- 36:45
- Robert Leshner: it affects everybody equally, it's a good thing. It's almost like the pacing that they were asking for. for,
- 36:52
- Robert Leshner: but in a fair, more uniformly distributed way. And so theoretically, it like reduces their like CapEx burden and it allows them to like continue delivering their models,
- 37:09
- Unidentified speaker (UNKNOWN): but
- 37:09
- Robert Leshner: without having to spend the bajillion dollars that they've committed.
- 37:13
- Unidentified speaker (speaker_2): Right. But they're a lot. of their spending commitments are locked in, but it it means largely they can't grow.
- 37:18
- Unidentified speaker (UNKNOWN): But if they're if that if that
- 37:20
- Robert Leshner: compute's not delivered to them, they don't have to spend for it.
- 37:23
- Unidentified speaker (UNKNOWN): Yeah.
- 37:23
- Unidentified speaker (speaker_3): I I think I think Robert's point is more that like uh, you know, uh it competitors, you know, are now sort of suffocated off from new compute. It's like they bought all the chickens and so you can't start a new restaurant. And even if the chickens don't ever end up happening, you're still not gonna get chickens because they're they're not gonna come.
- 37:44
- Unidentified speaker (speaker_1): But yeah.
- 37:45
- Unidentified speaker (speaker_3): Sure, I was futures kind of famously chicken nuggets, a little ref. But anyway.
- 37:50
- Robert Leshner: Right. They
- 37:51
- Unidentified speaker (UNKNOWN): all bought
- 37:51
- Robert Leshner: chicken futures. And it's possible that the farms that were being built to grow the chickens don't
- 37:58
- Unidentified speaker (UNKNOWN): get built. bus.
- 37:59
- Unidentified speaker (speaker_2): This is getting this is getting let's let's let's let's pick one analogy, not too. two. Chickens.
- 38:03
- Unidentified speaker (UNKNOWN): Um let's keep
- 38:04
- Robert Leshner: chickens.
- 38:04
- Unidentified speaker (speaker_2): But also like it this also feels like it's fundamentally bearish on AI to believe that they will make more money if there's less compute in the world.
- 38:13
- Robert Leshner: It's bearish American AI relative to global competition.
- 38:17
- Unidentified speaker (UNKNOWN): No,
- 38:17
- Unidentified speaker (speaker_2): no, no. I mean that if for for hedge funds to be pricing this and saying,
- 38:20
- Unidentified speaker (UNKNOWN): But look at the hyperscaler stocks today
- 38:22
- Unidentified speaker (speaker_2): around. Right.
- 38:23
- Unidentified speaker (UNKNOWN): Hyperscalers Hyperscaler are is not the
- 38:24
- Unidentified speaker (speaker_1): same.
- 38:24
- Unidentified speaker (UNKNOWN): Hyperscales are up,
- 38:25
- Unidentified speaker (speaker_2): NVIDIA and like most of the iSocks are down, right? Yeah.
- 38:28
- Unidentified speaker (UNKNOWN): Like to me, they're fundamentally down,
- 38:30
- Robert Leshner: demand for chips down.
- 38:31
- Unidentified speaker (speaker_2): Yes. Like to but to me, what that implies is a fundamental belief that like we are hitting some ceiling on the ROI of getting more compute online. And that in fact getting more compute online is just going to dilute the earnings and create more competition among the players who are there, as opposed to what what all the labs are doing right now. is taking that
- 38:51
- Unidentified speaker (speaker_2): compute and investing it into building better models that will ultimately result in more economic value getting created. It seems like largely the hedge funds don't believe that if they believe that less compute coming online is good for these companies. Like I think on net, it should mean that it's a it's a limiter to growth. Even it can mean that they can earn more in
- 39:11
- Unidentified speaker (speaker_2): the short to medium term.
- 39:13
- Unidentified speaker (UNKNOWN): Uh
- 39:13
- Unidentified speaker (speaker_2): that's the the sense in which I say. the hedge funds are bearish on AI in aggregate, in believing that data centers not coming online is good for anthropic.
- 39:23
- Unidentified speaker (speaker_3): Yeah. I mean, I think it's it's more like concentration of of of power than than anything else, right? Like you can you can still believe that like compute coming online is good, but it's basically like this computer's already are been pre-sold. And so to Rob's point, like if it gets built or not is almost like you know irrelevant. Like we're kind of discussing the hypothetical futures compute that may or may not get built. You know,
- 39:43
- Unidentified speaker (speaker_3): is that good or bad for anthropic? I think it's you know, in some ways bad for anthropic if like they are now one of N in a competition versus like one of one or one of two. Um, I mean, that's just classic kind of you want to be a monopolist, uh monopoly capitalism. And and so I think that's probably what what this is looking at, which is like if we if we stop all increments.
- 40:02
- Unidentified speaker (speaker_3): incremental compute from being built today beyond what Entropic is already paid for, open is already sort of you know paid for, that seems very good for for the people who actually have it.
- 40:10
- Unidentified speaker (speaker_1): My kind of contrarian take to this is
- 40:14
- Unidentified speaker (speaker_1): that the non-US compute build out is now gonna is I mean, you already see like Crusoe and Corviv We've going on TV and podcasts like saying this. And I would say on the private side, you see this a lot too. Is like the non-US build-out speed up is like 100x in the last two weeks versus like where it was
- 40:34
- Unidentified speaker (UNKNOWN): in
- 40:34
- Unidentified speaker (speaker_1): terms of like how much capital commitments are coming in, people issuing debt in other countries. Like I basically think the non-US build-up will catch up, perhaps probably exceed the US now, and that compute will get sold, most likely not to anthropic in some ways, because like
- 40:51
- Unidentified speaker (UNKNOWN): I
- 40:51
- Unidentified speaker (speaker_1): think they, you know, for instance, don't like buying non-US compute as much as other people do.
- 40:57
- Unidentified speaker (speaker_2): Right. It's not as enterprise friendly to have uh your your GPUs outside the US. Um, but the demand from neo clouds and from consumer apps and whatever, I think they're gonna be a lot less sensitive to where the compute's gonna live.
- 41:10
- Unidentified speaker (speaker_1): Exactly. And so I think that's that's you're gonna see the and that's gonna look like crypto. geodistributing data centers, versus like the anthropic stuff is gonna look like Trad hyperscaler. I think we're getting into a tale of two worlds soon from this.
- 41:23
- Unidentified speaker (speaker_2): Yeah.
- 41:24
- Unidentified speaker (speaker_1): And
- 41:25
- Unidentified speaker (UNKNOWN): we
- 41:25
- Unidentified speaker (speaker_1): didn't have to be. We could have had it had more here,
- 41:28
- Unidentified speaker (UNKNOWN): but
- 41:29
- Unidentified speaker (speaker_1): alas.
- 41:30
- Unidentified speaker (speaker_2): Alas. Okay. Well I
- 41:32
- Unidentified speaker (UNKNOWN): I mean,
- 41:33
- Unidentified speaker (speaker_1): also like the other thing we didn't talk about, which I think is important to this, us, is like all the propositions on data centers was like one direction. right? There was not even like
- 41:40
- Robert Leshner: oh yeah, widespread opposition.
- 41:42
- Unidentified speaker (UNKNOWN): We
- 41:42
- Unidentified speaker (speaker_1): didn't talk about that, right?
- 41:43
- Unidentified speaker (speaker_2): Yeah, yeah. The new Texas mayor. So Texas is the number one state in the country for data centers. Uh the new Texas mayor, I'm forgetting her name, Hina Hina Jo Gina.
- 41:53
- Unidentified speaker (UNKNOWN): Gina Hina Joza.
- 41:54
- Unidentified speaker (speaker_2): Gina Hina Jose. She explicitly ran on basically slowing down or I think uh maybe even a moratorium on on new data center build outs
- 42:02
- Unidentified speaker (UNKNOWN): in
- 42:02
- Unidentified speaker (speaker_2): Texas.
- 42:03
- Unidentified speaker (UNKNOWN): She ran on she
- 42:04
- Robert Leshner: ran on a populism campaign in Japan.
- 42:06
- Unidentified speaker (UNKNOWN): Yeah.
- 42:06
- Robert Leshner: Like she ran, by the way, she ran on a platform which you would run on if you were running in Texas and you did not anticipate it was going to be a super duper extreme blue tsunami above all expectations. I don't think she thought she was going to win, right? I don't think she thought there was a chance.
- 42:25
- Unidentified speaker (UNKNOWN): But
- 42:26
- Robert Leshner: her platform was essentially
- 42:29
- Robert Leshner: populism,
- 42:31
- Robert Leshner: no social issues, right? Anti-talking about social issues, no progressivism, like nope, not about social, only about tech companies are the enemy.
- 42:43
- Unidentified speaker (UNKNOWN): Hmm.
- 42:45
- Unidentified speaker (speaker_2): Interesting. I I have not followed that much of Texas Texas politics besides all the stuff about Telerico. But yeah, it does seem like the environment, the mood in the country, especially at the state level, is going to change quite a bit with respect to data centers.
- 42:58
- Unidentified speaker (UNKNOWN): I,
- 42:59
- Unidentified speaker (speaker_1): for one, am excited for the cryptoization of the data center industry. We're just going back to our roots. Then then the show will be, you know, all this AI talk we had will suddenly just be really about crypto again.
- 43:09
- Unidentified speaker (speaker_2): Yeah, indeed. Well, this
- 43:10
- Unidentified speaker (UNKNOWN): like really reminds
- 43:11
- Unidentified speaker (speaker_1): me of like the US miners versus the international miners and like how that that's
- 43:15
- Unidentified speaker (UNKNOWN): what about what
- 43:16
- Robert Leshner: about this? What if we rename the show the chipping block? That way
- 43:22
- Robert Leshner: that way we can work
- 43:24
- Unidentified speaker (UNKNOWN): the thing out
- 43:25
- Robert Leshner: in.
- 43:26
- Unidentified speaker (UNKNOWN): No,
- 43:26
- Unidentified speaker (speaker_1): no, no. But but the chopping block has the view of like a guillotine, right? It's like you're dead and one's one kill. Where the chipping
- 43:32
- Unidentified speaker (UNKNOWN): block is
- 43:32
- Unidentified speaker (speaker_1): like I'm just like slowly beating you up.
- 43:35
- Unidentified speaker (UNKNOWN): I
- 43:36
- Robert Leshner: mean that sounds
- 43:36
- Unidentified speaker (UNKNOWN): like the show.
- 43:37
- Robert Leshner: We've been doing this for years. We're slowly
- 43:38
- Unidentified speaker (UNKNOWN): chipping away.
- 43:39
- Unidentified speaker (speaker_2): Principles are getting chipped away at. That's the chipping block. Yeah. I like that. Okay. Well, speaking of the cryptoization of AI, one of the stories that I wanted to cover, uh, a little bit of a of a deep cut, but so there's this group, this uh what what is their name? FAI, which stands for Yes. What does it stand for? Foundation for American
- 43:57
- Unidentified speaker (UNKNOWN): American.
- 43:58
- Unidentified speaker (speaker_3): Yeah. Yeah, that's right.
- 43:59
- Unidentified speaker (speaker_2): Foundation for American Innovation. There was a a a a position paper they put out. called the Obliteration Threat. So for those of you who don't know, obliteration is a technique that you can use. Uh so taking an AI model, an open source AI model, and you know, AI models will often refuse things that they have been trained to say are unsafe or are um you know unsavory or just
- 44:19
- Unidentified speaker (speaker_2): inappropriate for normal users. Uh obliteration is a technique that you can use to remove the behavior of refusing from an open source model. So if you go on Hugging Face, you can find a lot of obliterated versions of open source models like DeepSeek and GLM 5.3 and so on. So these obliterated models have become quite
- 44:39
- Unidentified speaker (speaker_2): popular online, and largely from a lot of security researchers who use these obliterated models for cyber defense, for red teaming, uh, which are often things that the models off the shelf will refuse to do because. they kind of have been trained. Oh, this looks like cyber. I'm not supposed to do this. Even open source models will often refuse cyber related tasks. But if you use an
- 44:59
- Unidentified speaker (speaker_2): obliterated version, they will say yes to whatever you ask them to do. So this these two guys, Murphy and Huang at FAI, uh, they basically wrote a paper saying that obliterated models are a threat, that they enable bad actors to do bad things, and therefore that they should be regulated and the company. companies that enable them should be taken offline, should be, you know,
- 45:18
- Unidentified speaker (speaker_2): given some kind of have the book thrown at them with respect to liability. Now, I should note that we're investors in Venice, which uh lists obliterated models. There's a a company that has gone very big called obliteration.ai, which does serve a lot of obliterated models. NIR NIAR also serves obliterated models. Uh, and Venice serves some of these obliterated models from some of these third-party third
- 45:38
- Unidentified speaker (speaker_2): party obliterators, I guess would be the term. And so this is, I think, something that actually is quite near and dear to crypto this idea that obliterated models serve an important purpose in giving people the freedom and the ability to use their AIs locally in ways that they choose, that they understand actually they would need a model that
- 45:58
- Unidentified speaker (speaker_2): doesn't refuse, especially in high high-pressure pressure situations such as cyber attacks. Tarun, curious to get your take on this.
- 46:04
- Unidentified speaker (UNKNOWN): I mean, I've
- 46:04
- Unidentified speaker (speaker_1): tweeted about
- 46:05
- Unidentified speaker (UNKNOWN): this a bunch.
- 46:05
- Unidentified speaker (speaker_1): I'm I'm very pro-obliterated models. I generally think AI models are sort of some form of freedom of speech and expression and all obliter like removing the blocks of like the model not respecting your kind of expression demands that maybe are placed upon you by someone you
- 46:25
- Unidentified speaker (speaker_1): disagree with. That feels like a very real thing that, like, if it's an open source model, you should be able to do
- 46:31
- Unidentified speaker (UNKNOWN): it. I
- 46:31
- Unidentified speaker (speaker_1): think the other interesting thing is like obliterated models. Yeah, there's a defense aspect of like you're getting attacked. There's sort of the way to prevent yourself. On the other hand, obviously, I I've heard lots of kind of conjecture and sort of head kind of sketchy kind of conjecture that, like, for instance, in North Korea. Binance
- 46:51
- Unidentified speaker (speaker_1): hack use obliterated models somewhere. So it goes both directions. I don't think, I think inevitably that comes with the the territory here. But
- 47:00
- Unidentified speaker (UNKNOWN): I
- 47:01
- Unidentified speaker (speaker_1): think the more interesting question is:
- 47:04
- Unidentified speaker (speaker_1): if you have this universe of these models, how do you
- 47:09
- Unidentified speaker (speaker_1): how do you think they should be distributed? Slash, do you, you know, what what do you think happens with them? And I think at the end, end of the day, there's always going to be a way to make them. There's not really I don't think there's a way of like, oh, you can prevent safety, people will tell you that. But as you've seen with models constantly being able to uh
- 47:28
- Unidentified speaker (speaker_1): optimize themselves and optimize
- 47:31
- Unidentified speaker (UNKNOWN): other
- 47:31
- Unidentified speaker (speaker_1): models, they can easily you you you you can use larger models to remove obliteration from smaller models very easily right now. That's the main way a lot. of people do
- 47:39
- Unidentified speaker (UNKNOWN): it. I
- 47:40
- Unidentified speaker (speaker_1): don't think you're getting around that. I think you're better off just trying
- 47:43
- Unidentified speaker (UNKNOWN): to
- 47:45
- Unidentified speaker (speaker_1): focus on kind of I I think like the stuff NIR is doing where they're kind of running it privately encrypted, so like you can't kind of interfere with it, but you can also make sure that's like doing the right thing. That that to me is like one very reasonable way of of dealing with them. I think
- 48:03
- Unidentified speaker (speaker_1): inevitably you don't want obliterated models for formal verification. You kind of, in some sense, you want the model to have those restrictions if it's like trying to prove certain properties. And so I think we're going to end up in this world where you have obliterated models, like these free models, restricted models, but they have very specific uses and they're going to interact with one
- 48:23
- Unidentified speaker (speaker_1): another, and people are going to make routers built around them separately. So
- 48:28
- Unidentified speaker (UNKNOWN): I
- 48:28
- Unidentified speaker (speaker_1): don't know. I I think it's it's it's a fact of nature that you can't really get rid of them. There's almost like a law of physics around this. So don't like it's not worth trying to you you can I don't think you're gonna ever regulate this type of thing out of existence, no matter what Anthropic tells
- 48:43
- Unidentified speaker (UNKNOWN): you. Tom,
- 48:45
- Unidentified speaker (speaker_2): what's your take?
- 48:46
- Unidentified speaker (speaker_3): Yeah, I mean it does just feel crypto, these are the same kind of your you arguments know people. levied against Bitcoin way way back in the day, or or sort of uh um privacy solutions on on chain where it's like, oh, you know, some bad people will use this for bad things, so therefore they shouldn't not be allowed. And obviously that's there's uh sort of outweighed by the number
- 49:06
- Unidentified speaker (speaker_3): of you know good actors that need this thing. And and so everything is a tool and it you know can be used for for for good purposes and bad purposes. But it feels like right now it's actually so lopsided where. these things ex exist in the wild and so tons of bad actors do end up having them. And so the question is, how can you defend yourself? I I don't think the answer is like everyone has to go through a frontier lab
- 49:26
- Unidentified speaker (speaker_3): and get access if they can, you know, assuming they're even in the US and and they're able to pay for it. You know, the answer is okay, we just have something that is, you know, available for everybody to use to to defend themselves and you know, coming on in into the future and obviously it would also be like like super just unprecedented to like ban like I don't know, possession of these these models information,
- 49:46
- Unidentified speaker (speaker_3): let alone like serving them. And so I think there obviously are places to put put safeguards you know around these, the same way we have you know safeguards around you know social media social media sites or something like that. But like you don't we don't make the act of like uh running a server, you know, illegal. It's a it's a similar kind of kind of concept. And so I I think we're also actually starting to see, you know, in uh in some ways it also reminds me a
- 50:06
- Unidentified speaker (speaker_3): little bit of of social media, where you know there is sort of one set of censorship concepts or sort of allowed discussion that maybe makes sense for one country or or one locale, but like obviously you don't apply globally. And so social media sort of went through that experience you know over the past decade and learning how to do that. And kind of the end you know, result that they've they've ended up on is okay, well,
- 50:26
- Unidentified speaker (speaker_3): yeah. we actually just like, you know, add context, or maybe we we derank stuff that that's not good. But like, we're not gonna go and try to be, you know, uh uh sort of the censorer for like every single country in the world. That's just not possible. And I think similarly, we're seeing people run into all these issues with like Claude. For example, right now, like Opus has all these like random refusals in it. Um,
- 50:46
- Unidentified speaker (speaker_3): or where they they recently added a refusal. People were like, you know, decompiling or recreating video games or like, They're hey, well, actually that's that's like that okay. I I actually ran to this. I went as um Claude Hanson for for for Halloween. So if I could play on Chris Hansen and it refused
- 50:59
- Unidentified speaker (UNKNOWN): to make
- 50:59
- Unidentified speaker (speaker_3): me a uh you know how Chris Hansen comes with like a a transcript of like, you know, the the person on the show and like the and it refused to make me a transcript. It's like even a pretend grooming conversation for the purposes of of this costume, I I will not do. And so maybe for some places, for some people, that's okay. That's where you want to draw draw the line, but I just think
- 51:19
- Unidentified speaker (speaker_3): it's possible to do that globally. And obliteration is one sort of thing where it's you know nothing's ever sort of sort of black and white. And so the end result is you let people sort of police themselves.
- 51:29
- Unidentified speaker (UNKNOWN): Yeah,
- 51:29
- Unidentified speaker (speaker_2): I think obliteration is a perfect example of prior restraint, which is we can tell in advance that the only reason you would use an obliterated model is for a bad reason. And I think that seems uniquely obviously wrong right now, given the cyber apocalypse that we see going around everywhere. You know, very famously when when the
- 51:49
- Unidentified speaker (speaker_2): OpenAI Hugging Face attack took place, Hugging Face had to defend itself using an open model. And because literally OpenAI was refusing As to engage in defensive behavior, and and to be clear, this is true. Even if you go on daybreak. or Glasswing, which are OpenAI and Anthropics cybersecurity programs, even still, you will get refusals
- 52:08
- Unidentified speaker (speaker_2): sometimes. And you will not necessarily get whitelisted for all of the kinds of things that you want to do, such as red teaming. They won't always allow you. They won't always
- 52:16
- Unidentified speaker (UNKNOWN): say,
- 52:16
- Unidentified speaker (speaker_2): Yes, you're good for red teaming. They might
- 52:18
- Unidentified speaker (UNKNOWN): say,
- 52:18
- Unidentified speaker (speaker_2): Oh, you're only good for analyzing logs. You're only good for you know this, this uh you know, config analysis or whatever. So all of that tells me that like this, I think you can. draw a straight line from both the First Amendment and maybe even the Second Amendment to why it is that obliterated models, you cannot say in advance you are not allowed to do this. Now, the liability
- 52:38
- Unidentified speaker (speaker_2): argument, yes, absolutely. If you're doing if you're running hosted models, uh, there may be liabilities. If uh if you're running a model and it's being used to hack people and do all sorts of bad things, and you're serving inference tokens to somebody and you're not monitoring. what's going on and you're being negligent. Okay, there should be liability for that. That's that's understood. But it's obviously not
- 52:57
- Unidentified speaker (speaker_2): North Korea that's using, you know, Venice or, you know, whatever. North Korea's hosting that shit themselves. It's people who are using these models defensively in the US who are signing up with their credit cards and paying for these kinds of services.
- 53:11
- Unidentified speaker (UNKNOWN): I
- 53:11
- Unidentified speaker (speaker_1): think we're in for another kind of like the Supreme Court crypto case. of the 90s around this. Because it is like it sort of feels like a constitutional expression thing. And like end of the day, that's going to be the
- 53:23
- Robert Leshner: Yeah, but
- 53:24
- Unidentified speaker (UNKNOWN): there's a
- 53:24
- Robert Leshner: difference between expression and like wielding a super weapon, right? And like that is a blurry line.
- 53:31
- Unidentified speaker (speaker_1): Yeah, but it's it's at the same time, it's like I can talk about the super weapon without being censored.
- 53:36
- Robert Leshner: Totally. Totally. And maybe it's on use and intent and whatever.
- 53:39
- Unidentified speaker (speaker_1): Yeah, exactly. Exactly. And I do
- 53:41
- Unidentified speaker (UNKNOWN): think these
- 53:42
- Unidentified speaker (speaker_3): things are all crimes that we're
- 53:43
- Unidentified speaker (speaker_1): discussing, by the way. So
- 53:43
- Unidentified speaker (speaker_3): there obviously is liability at the
- 53:45
- Unidentified speaker (UNKNOWN): same time. So the
- 53:46
- Unidentified speaker (speaker_1): liability part is certainly true. And I think inevitably what will happen is like
- 53:51
- Unidentified speaker (UNKNOWN): kind
- 53:51
- Unidentified speaker (speaker_1): of what happened in crypto, right? Where like the US regulated validators are like banning all the OFAC stuff and tornado cash cache and whatever. And then the non non-US US random. random validators are running kind of will do anything. And you're gonna have the same thing with inference providers. You're already starting to see this where like a lot of the
- 54:11
- Unidentified speaker (speaker_1): inference writers are like, okay, we're guaranteeing you US only and only within this these jurisdictions will be where you run and stuff like that. Right. So
- 54:19
- Unidentified speaker (speaker_2): yeah. I think it's not just that. It's like, yeah, I mean but we see all these stories. that Anthropic actually they publish quite a bit of an analysis of bad actors using Claude to attack things and and you know a lot of this kind of OSINT type drops that they that they give us. And the number one thing that we learn is that most of these bad actors are foreign. These aren't Americans. It like when you think
- 54:39
- Unidentified speaker (speaker_2): about all the cyber attacks and the damage that's being done in this unique period that we're in right now, almost all of it is foreign. Almost all of it is like Chinese, Russian, you know, whatever. It's it's it's not Americans. It's not Americans doing this. If it's Americans doing this, they will go to jail. They know that like this is super against the law and you can find them and toss
- 54:59
- Unidentified speaker (speaker_2): them in jail. If
- 55:00
- Unidentified speaker (UNKNOWN): you use a gun to
- 55:01
- Unidentified speaker (speaker_1): kill someone, you will go to jail. It like
- 55:04
- Unidentified speaker (speaker_2): yes. And
- 55:05
- Unidentified speaker (UNKNOWN): not right.
- 55:06
- Unidentified speaker (speaker_1): It's the same thing.
- 55:07
- Unidentified speaker (speaker_2): Agreed. But like overwhelmingly, this problem is a very is a theoretical one. The idea that US inference providers are enabling Americans
- 55:15
- Unidentified speaker (speaker_2): to just flagrantly go around hacking people and doing terrible things.
- 55:18
- Unidentified speaker (UNKNOWN): Or not just Americans not
- 55:20
- Robert Leshner: enabling anyone.
- 55:23
- Unidentified speaker (speaker_2): Yes. Overwhelmingly, these Americans are using obliterated models for defense. They're using it to protect themselves against the scourge of all this stuff happening uh over overseas. So um
- 55:36
- Unidentified speaker (UNKNOWN): I think
- 55:36
- Unidentified speaker (speaker_2): we're at time, so we gotta we gotta wrap it up here. But good to catch up with all you guys and be back on the show. And we will be back next week.
- 55:45
- Unidentified speaker (speaker_1): Thanks
- 55:45
- Unidentified speaker (speaker_2): everyone.
More from Unchained
Is Ethereum's $240 Billion Market Cap Justified With Only $6 Million in Monthly Fees?
Episode discussion: Jeff Dorman argues Ethereum's $240B market cap is unjustified by its $6M monthly fees, labeling most value as 'hopium'.
Speakers’ claims: Jeff Dorman contends that Ethereum's value is predominantly social rather than financial, estimating only $50 billion of its $240 billion market cap is tangible based on $6 million in monthly fees and idle wallet utility. He contrasts this with application-layer tokens like Hyperliquid, which he claims generate significantly more revenue, arguing that the 'fat protocol thesis' is wrong and value accrues to apps, not layer-one networks. Dorman dismisses Ethereum's 'store of value' thesis as misguided in a multi-chain future and proposes that Ethereum should tax applications to capture revenue. Laura Shin and Speaker_1 challenge these views, noting that Dorman's proposals conflict with Ethereum's decentralization principles and that the network's value may stem from factors beyond immediate fee capture.
Uneasy Money: Why Ethereum's 'Ultrasound Money' Meme Finally Ran Out of Road
Episode discussion: Speakers critique Ethereum's 'ultrasound money' narrative, discuss core dev bottlenecks, and analyze AI inference arbitrage opportunities.
Speakers’ claims: Hosts Austin Griffith and Kain Warwick, along with guest Greg Markou, argue that Ethereum's 'ultrasound money' thesis has lost momentum and lacks a compelling institutional value proposition beyond meme-driven narratives. The discussion highlights growing tensions between core and application developers regarding gas repricing and the challenges of testing upgrades in a monolithic codebase. Markou criticizes Ledger's SDK for privacy-invasive telemetry and advocates for clear signing standards to improve security. Finally, the panel explores MegaETH's pivot to an AI inference marketplace, identifying significant arbitrage opportunities between subsidized subscription costs and API rates.