TFTC: A Bitcoin Podcast

Anthropic's S-1 reveals massive compute commitments and profitability, while Bitcoin resilience follows Binance hack.

39 min Listen to the original Episode page
Anthropic S-1AI InfrastructureBitcoin SecurityBinance HackHyperscaler Cash FlowGeopolitics

Summary

The hosts analyze Anthropic’s S-1 filing, highlighting nearly $1 billion in Q2 operating income and $895.4 billion in conditional compute commitments, while noting that 72% of compute is currently used for training rather than customer inference. They discuss the geopolitical implications of AI labs verticalizing into sectors like pharma and defense, and the financial risks posed by expected junk credit ratings and negative free cash flow among hyperscalers. In crypto news, the speakers examine the $6.3 billion Binance hack, attributing it to a social engineering attack on an Ethereum feature, and note Bitcoin’s price stability above its 50-week moving average. The episode concludes by contrasting Bitcoin’s security model with Ethereum’s complexity, citing AI audits of Bitcoin Core that found only minor bugs.

Episode sections

  1. 0:45 – 5:43

    S-1 Analysis: Profitability and Customer Base

    Hosts analyze Anthropic’s S-1, noting $1 billion Q2 operating income and a strong enterprise customer base of 12,000+ customers paying >$100k/year.

  2. 5:43 – 7:45

    Compute Commitments and Capital Intensity

    Discussion of $895.4 billion in compute commitments, clarified as largely conditional on performance rather than binding debt, despite extreme capital intensity.

  3. 7:45 – 11:52

    Training vs. Inference and ROI Questions

    Analysis shows 72% of compute used for training/research vs 25% for customer service, raising questions about ROI and potential shifts to inference.

  4. 11:52 – 16:47

    Verticalization and Geopolitical Implications

    Speakers discuss AI labs potentially competing in vertical markets like pharma and defense, creating channel conflicts and geopolitical advantages for US firms.

  5. 16:47 – 20:55

    Credit Ratings and Financing

    Conversation on Anthropic’s expected junk credit rating and the need for investment-grade status to finance the $800 billion infrastructure plan.

  6. 20:55 – 23:30

    Market Correction and Hedge Fund Degrossing

    Review of market volatility, Intel’s decline, and record hedge fund degrossing in AI stocks as the 'mystery box' phase ends.

  7. 23:30 – 28:17

    Hyperscaler Cash Flow and Macroeconomic Link

    Analysis of negative free cash flow among the 'Big Four' hyperscalers in Q3, linking AI CapEx to GDP growth and manufacturing PMIs.

  8. 28:17 – 29:19

    Meta vs. Google Advertising and AI Engagement

    Discussion on AI cannibalizing Google Search revenue while Meta leverages AI for ad targeting and user engagement on Instagram.

  9. 29:19 – 32:11

    Brazil Election Results and Geopolitical Implications

    Analysis of Lula’s re-election in Brazil as a setback for US strategic interests in Latin America regarding rare earth minerals and Chinese influence.

  10. 32:11 – 34:40

    Binance Hack and Market Reaction

    Coverage of the $6.3 billion Binance hack via social engineering, noting Bitcoin’s resilience above the 50-week moving average and Binance’s quick response.

  11. 34:40 – 36:02

    Historical Parallels to FTX and Bitcoin's Technical Resilience

    Comparison of the Binance hack to FTX, noting historical patterns of Bitcoin bottoms during election-period exchange failures.

  12. 36:02 – 38:44

    Bitcoin's Security Model vs. Ethereum's Complexity

    Argument that Bitcoin’s simplicity makes it more secure than Ethereum, citing the EIP-7702 exploit and AI audits finding only minor bugs in Bitcoin Core.

Assets discussed

AssetKindStanceContext
BitcoinBTCcryptomixedNot for Bitcoin either.
MicrosoftMSFTequityneutralI think maybe the most obvious one, right, is and this isn't necessarily beyond tech, but Microsoft announced last week that they're gonna cut headcount or at least offer buyouts for 60,000 roughly employees, which is essentially a quarter of global workforce.
Meta PlatformsMETAequityneutralWe talked about Meta and the offering that they may, the equity offering that they may tap in the future.
AlphabetGOOGLequityneutralI think Google really set the trend earlier this summer when they raised 85 billion in equity for the first time uh in quite a while.
EthereumETHcryptomixedI believe 18,000 Bitcoin were swept, and then a number of other coins across the board, Ethereum, Solana, Tron, a lot of tether and USDC as well.
SolanaSOLcryptomixedI believe 18,000 Bitcoin were swept, and then a number of other coins across the board, Ethereum, Solana, Tron, a lot of tether and USDC as well.
TRONTRXcryptoneutralI believe 18,000 Bitcoin were swept, and then a number of other coins across the board, Ethereum, Solana, Tron, a lot of tether and USDC as well.
TetherUSDTcryptoneutralI believe 18,000 Bitcoin were swept, and then a number of other coins across the board, Ethereum, Solana, Tron, a lot of tether and USDC as well.
USDCUSDCcryptoneutralI believe 18,000 Bitcoin were swept, and then a number of other coins across the board, Ethereum, Solana, Tron, a lot of tether and USDC as well.

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

Anthropic had almost $1 billion of operating income in Q2 on a GAAP basis.

Speaker 1 ·2:00

How to verify: Check Anthropic’s S-1 filing, specifically the Consolidated Statements of Operations for the period ending Q2 2026.

Anthropic has close to 12,000 customers paying over $100,000 annually.

Speaker 1 ·2:57

How to verify: Check Anthropic’s S-1 filing, specifically the section detailing customer concentration and enterprise sales metrics.

Anthropic’s Q3 revenue was around $17 billion, up 1,127% year over year.

Speaker 2 ·5:51

How to verify: Check the amended S-1 filing released in November 2026 for Q3 revenue figures.

Anthropic has cumulative purchase and lease commitments of $895.4 billion.

Speaker 2 ·6:02

How to verify: Check the amended S-1 filing for the 'Commitments' or 'Leases' footnote section.

72% of Anthropic’s H1 compute went to training, research, and internal use, while only 25% went to serving paying customers.

Speaker 1 ·7:54

How to verify: Check the S-1 filing for details on compute utilization or cost allocation between training and inference.

Anthropic is expected to receive a high junk rating from major credit rating agencies after its IPO.

Speaker 2 ·17:01

How to verify: Check reports from the Financial Times or statements from Blackstone regarding Anthropic’s credit rating outlook.

Anthropic expects to spend over $800 billion on computing infrastructure between now and 2045.

Speaker 2 ·17:26

How to verify: Check the Financial Times report cited by the speaker or Anthropic’s S-1 for long-term capital expenditure guidance.

Microsoft announced it would cut headcount or offer buyouts for roughly 60,000 employees.

Speaker 1 ·19:40

How to verify: Check Microsoft’s press releases or earnings call transcripts from October 2026 regarding workforce reductions.

Hedge funds executed the biggest degrossing event on record in October.

Speaker 1 ·21:33

How to verify: Check data releases from Morgan Stanley regarding hedge fund positioning and net exposure changes in October 2026.

Google raised $85 billion in equity earlier in the summer.

Speaker 2 ·23:55

How to verify: Check Google’s (Alphabet) SEC filings (Form S-1 or 8-K) for equity offerings completed in the first half of 2026.

AI LLMs are cannibalizing Google Search revenue.

Speaker 1 ·28:17

How to verify: Check Google's quarterly earnings reports (10-Q/10-K) for search revenue trends and management commentary on AI impact.

Meta's first major AI implementation was with their ad product targeting.

Speaker 1 ·28:44

How to verify: Review Meta's investor presentations or earnings call transcripts regarding AI adoption in advertising.

Lula won the Brazilian election after Bolsonaro had pulled ahead unexpectedly.

Speaker 1 ·29:43

How to verify: Check official Brazilian election results and news reports from October/November 2026.

Binance was hacked for $6.3 billion, the largest crypto theft on record.

Speaker 1 ·32:32

How to verify: Review Binance's official security announcements and blockchain analytics reports from Chainalysis or TRM Labs.

Bitcoin slipped below $77,000 but closed above its 50-week moving average during the hack.

Speaker 1 ·32:41

How to verify: Check Bitcoin price data from CoinGecko or CoinMarketCap for the week of the hack (Nov 2026).

18,000 Bitcoin were swept from Binance hot wallets.

Speaker 1 ·33:21

How to verify: Verify via blockchain explorers (e.g., Etherscan, Blockchain.com) for the specific transaction volumes and addresses involved in the hack.

The Binance hack exploited EIP-7702, an Ethereum improvement proposal implemented in May of the previous year.

Speaker 1 ·36:12

How to verify: Review Ethereum Improvement Proposal EIP-7702 documentation and implementation dates on Ethereum.org or GitHub.

16 AI agents spent a week on Bitcoin Core and returned with two minor items.

Speaker 2 ·36:35

How to verify: Check Coinbase's official blog or press releases regarding their AI security audit of Bitcoin Core.

FTX collapse occurred during the week of the US midterm elections and marked a Bitcoin bottom.

Speaker 1 ·34:44

How to verify: Verify the date of FTX's bankruptcy filing (Nov 2022) against the US midterm election date (Nov 2022) and Bitcoin price history.

Bitcoin is currently trading at $77,352.

Speaker 1 ·32:53

How to verify: Check real-time or historical Bitcoin price data from a major exchange or aggregator at the time of the podcast recording.

Do your own research

  • Review Anthropic’s S-1 filing (filed with the SEC) to verify the $1 billion Q2 operating income and the specific terms of the $895.4 billion compute commitments.
  • Check Blackstone and Financial Times reports regarding Anthropic’s expected credit rating post-IPO and its impact on capital costs.
  • Analyze Q3 earnings reports from Google, Meta, Amazon, and Microsoft to verify negative free cash flow figures and AI monetization performance.
  • Examine Binance’s official announcements and reserve proofs following the $6.3 billion hack to assess solvency and fund allocation.
  • Review Ethereum EIP-7702 documentation and security audits to understand the technical vulnerabilities exploited in the Binance hack.
  • Consult historical Bitcoin price data relative to US midterm election weeks and major exchange failures to evaluate the 'bottoming' thesis.
  • Look for independent security audits of Bitcoin Core, including those conducted by AI models, to verify the claim of only minor bugs found.
Full transcriptexpand
0:00
Unidentified speaker: John Arnold, October.
0:02
Unidentified speaker: Not this year.
0:03
Unidentified speaker: Not quite.
0:04
Unidentified speaker: Definitely.
0:05
Unidentified speaker: Definitely not for definitely not for the S ⁇ P and the NASDAQ, that's for sure.
0:08
Unidentified speaker: Not for Bitcoin either.
0:09
Unidentified speaker: Got our hopes up.
0:10
Unidentified speaker: Got our hopes up.
0:12
Unidentified speaker: Thought we were gonna get October for Bitcoin.
0:14
Unidentified speaker: Didn't happen.
0:15
Unidentified speaker: Calendar turn.
0:17
Unidentified speaker: November 2nd, we beat here.
0:18
Unidentified speaker: We're gonna talk about the S1.
0:20
Unidentified speaker: I think we mentioned it briefly in the beginning of last week's episode as it was just dropping.
0:25
Unidentified speaker: We were, this is the first time I'm hearing about this, reacting, shooting from the hip.
0:30
Unidentified speaker: But we've had a week to digest the news, and let's hop into it.
0:35
Unidentified speaker: Uh you titled this week's timestamp The S1 also rises.
0:40
Unidentified speaker: Two ways gradually then suddenly.
0:42
Unidentified speaker: So uh what's your take on all this?
0:45
Unidentified speaker: We'll go to the first slide.
0:46
Unidentified speaker: We showed this image last week, but again, a week to digest.
0:50
Unidentified speaker: What should we take from all this?
0:52
Unidentified speaker: Yeah, the the two ways gradually than suddenly quote is is interesting here uh to me because you know if you're a bitcoiner, you'll you'll know gradually and suddenly well from Parker Lewis's book and his his long-running blog on Bitcoin adoption.
1:06
Unidentified speaker: But um in in context, the the speaker uh of that quote is being asked, you know, how did you go bankrupt?
1:13
Unidentified speaker: And he says two ways gradually then suddenly.
1:16
Unidentified speaker: And so I think it's interesting because that's uh effectively the debate that's being had right now about the the numbers that you're looking at here, whether it seems like we're kind of branching into two potential paths, one being uh you're you're looking at the the IPO and the emergence of one of the most dominant uh franchises and um institutions in the history of of mankind.
1:35
Unidentified speaker: Uh, and the other path is this is all horribly unsustainable and you know is is 12 months away from blowing up.
1:41
Unidentified speaker: So again, uh continues to be uh we'll talk about this a little more on the in the next slides, but a Rorschach test here.
1:48
Unidentified speaker: Um you can kind of see that on uh the the left-hand side, the top, top of the page.
1:52
Unidentified speaker: Uh a lot of statistics about from the S1 that I think are should be quite surprising to a lot of a lot of bears.
2:00
Unidentified speaker: I think anyone who's being honest will tell you they did not see, they did not expect Enthropic to have almost a billion dollars of operating income in Q2 on a gap basis, right?
2:09
Unidentified speaker: So that's including stock comp, no funny business.
2:12
Unidentified speaker: That's kind of a generally real number.
2:14
Unidentified speaker: I think there's some you know nuances that you can argue about of how sustainable it is.
2:19
Unidentified speaker: And Ed Zitron you know suggested that, well, that's just getting flattered because the they had effectively discounted SpaceX compute at the time that's gonna have to re-rate higher.
2:27
Unidentified speaker: So, okay, maybe first half free cash flow, positive, huge cash balance currently.
2:33
Unidentified speaker: Now, whether that's enough to um help them fund the hundreds of billions of you know compute commitments over the next 20 years, you know, we'll have to see.
2:41
Unidentified speaker: Uh, I think maybe the most interesting data point out of the whole thing for me was um the statistics they disclosed on on customers.
2:48
Unidentified speaker: And we hit on this a little bit last week, but uh if you go and kind of look uh in the middle of the S1, there's so there's a breakdown of the users they have at the enterprise level.
2:57
Unidentified speaker: And you've got now, as of the S1, close to 12,000 customers paying over $100,000 annually, and then uh a decent amount as well at higher and higher tiers, including a surprising amount even at you know 10 million plus.
3:09
Unidentified speaker: Um that's up substantially year on year.
3:12
Unidentified speaker: We we mentioned net revenue retention and customers just paying more and more over time, you know, on a same account basis.
3:18
Unidentified speaker: Um, and I think there's also some detail in there about uh the penetration that they have thus far into you know Fortune 500 accounts.
3:24
Unidentified speaker: And I can't remember off the top of my head, but it's it's material in terms of account penetration.
3:27
Unidentified speaker: And within those accounts, you know, the the actual kind of dollar penetration that they're seeing is still very early.
3:32
Unidentified speaker: So, you know, if you're a bully, you're gonna argue like they're getting the traction and they're kind of just in early innings of you know, land and expand in these big accounts, which is meaningful, I think, relative to a popular narrative over the last couple you know months that if you looked at, say, the ramp data, you know, you would see, and I think Apollo pointed this out and kind of was running victory victory laps around it a few weeks ago, that you know, there's a huge amount of spend coming from basically VC funded companies that are kind of in the the leading edge ecosystem.
4:00
Unidentified speaker: And it's this kind of like circular flow of you know, VC dollars paying, paying VC customers uh or paying VC vendors.
4:08
Unidentified speaker: And you know, we saw that to some extent in 2020, 2021 with with the SAS boom.
4:13
Unidentified speaker: And that would suggest, you know, that's just more capital needs to keep coming in just to keep the wheel spinning.
4:17
Unidentified speaker: But uh, if you look at kind of the customer breakdown, I think there's an argument that uh it's quite a lot more, it looks quite a lot more sustainable and and you know, stronger and sturdier than you know that you know narrative that we were seeing like five or six years ago in that space.
4:31
Unidentified speaker: So yeah, I think we talked about the chart on the right uh last week a little bit too, kind of telling the same story.
4:36
Unidentified speaker: On the bottom left of the slide, if you're watching, you can see you know a few bare cases you can still make here, without a doubt.
4:42
Unidentified speaker: You know, notably no Q3 data.
4:44
Unidentified speaker: So it uh I think we just got someone in a minute S1 this literally 10 minutes before we got on here.
4:49
Unidentified speaker: I have not looked at it, but you know, Iceman was making the the point that you know this is the lack of Q3 disclosure here is very intentional.
4:56
Unidentified speaker: I think you're right on the edge where you could have said, well, they you know they haven't really closed the books on Q3 yet, but um I think it's it's reasonable to say that.
5:02
Unidentified speaker: No doubt the compute commitments here are are uh significant.
5:06
Unidentified speaker: I think there's some nuance to apply to like over the time period over which those are paid and which them which of them are actually kind of binding and non-cantable versus just kind of MOUs, but no doubt it's a huge number.
5:18
Unidentified speaker: And then we mentioned kind of the the SpaceX um you know nuance that maybe makes profitability on a GoFord basis look a little um less impressive.
5:26
Unidentified speaker: But I I think all this should say, like anyone who was expecting that we were gonna get the S1 and suddenly uh everything about all would be revealed and and the this whole scheme would come you know crashing down and and that would be it, is I think probably, if they're honest, fairly disappointed by by what they've seen.
5:43
Unidentified speaker: And and there's you know more than there are more than enough kind of fundamental positive points here for for the bulls to chew on.
5:49
Unidentified speaker: You're on mute.
5:50
Unidentified speaker: There we go.
5:51
Unidentified speaker: I said we have the uh amended S1, uh I guess the the headline numbers out of there, Q3 revenue around 17 billion, up 1,127% year over year, quarter over quarter.
6:02
Unidentified speaker: I think it's still marginal.
6:04
Unidentified speaker: It would be aligned with that that flattening, and then nine-month revenue at 33.2 billion in cumulative purchase and lease commitments of eight hundred and ninety-five point four billion, I believe somewhere around 500 million or billion of which will be committed to by 2030.
6:22
Unidentified speaker: And yeah, I think that's I think when we get into it, particularly with these like lease agreements and commitments of 895.4 billion.
6:32
Unidentified speaker: I correct me if I'm wrong, but there's a lot of misconceptions around these specifically, too, because the the leases that that they'll sign into it's very conditional.
6:42
Unidentified speaker: Like if we get to this level, we will commit to spending X amount.
6:46
Unidentified speaker: And so it's not like it's fully committed, have to spend it.
6:50
Unidentified speaker: It's conditional on performance of the person hosting the compute.
6:56
Unidentified speaker: Yeah.
6:56
Unidentified speaker: So there is, I was actually doing over the last week kind of a taxonomy of this that I don't have finished, but yeah, there is like a lot of nuance like that around many of the agreements.
7:06
Unidentified speaker: I can't confidently say, you know, 895, you know, even if it's even if it's a third of that, even if it's like, you know, 10 or 15 percent of that, like that's a big number, right?
7:15
Unidentified speaker: So and I I I think we shouldn't understate that there is like this is very obviously it, you know, shouldn't be news to anyone now, a uh a capital and either a capital-intensive industry or at least an extremely capital in capital-intensive phase of the emergence of this industry.
7:30
Unidentified speaker: But yeah, I mean, I do think there are, if you kind of dig into the weeds, there are nuances around, like I said, both the timing of of that ramp and you know what is truly kind of debt-like, you know, non-cancelable liability that they're gonna have to deal with one way or the other.
7:45
Unidentified speaker: Yeah.
7:45
Unidentified speaker: And speaking of what does all this CapEx go to?
7:48
Unidentified speaker: I think it perfectly segues into the next slide, which is what is training actually for?
7:54
Unidentified speaker: So the S1 cost slide implies 72% of anthropic's H1 compute went to training, research, and internal use, and only 25% to serving paying customers.
8:03
Unidentified speaker: And so I think when we talk about training and the build-out, particularly in the context of AI, that's one thing people really need to understand too, is a lot of this CapEx is being spent on training and advancing the models and the IQ of the models, which begs the question how smart do these models need to be?
8:23
Unidentified speaker: And at what point will the frontier labs and hyperscalers recognize that hey, the return on investment of of training a new model is not worth the squeeze.
8:33
Unidentified speaker: So we'll stop training and move all of our compute to inference, which is much higher margin.
8:38
Unidentified speaker: Yeah, I think this is a really important point for the um for a lot of things, obviously for the the businesses themselves, but also for kind of the way that we think about how this flows through the the rest of the points that we talk about on the show a lot, the kind of geopolitical great game and uh industrial reshoring and the you know the the US's position vis-a-vis China.
8:58
Unidentified speaker: Yeah, I think this is this is important for that because you you have you have like an interesting, I mentioned a Rorschach test earlier.
9:04
Unidentified speaker: We've used that that concept in prior episodes, but um I think you really see it most acutely kind of here.
9:10
Unidentified speaker: You can see uh, you know, if you were watching last week, you'd see bullish reads on data like this on the basis that you know the huge amount of compute, like you said, is still just going to basically zero revenue RD, right?
9:24
Unidentified speaker: You could think of it as this is currently like non-productive spending, like in in in this accounting period.
9:30
Unidentified speaker: And it's gonna pay dividends over time.
9:33
Unidentified speaker: And yet, even burdened with all that, look how profit businesses, you know, look how like how high inference margins are.
9:39
Unidentified speaker: Um, it's kind of you could think of it like, well, it's a dial and they just flip it.
9:43
Unidentified speaker: They can titrade it to uh when necessary, you know, serving more compute to inference because this is such a the demand for intelligence is infinite and the capacity to to serve that demand is is just fundamentally limited.
9:56
Unidentified speaker: And so when whenever they want to, you know, they can just kind of clip some of this, the some of this training over to you know, inference, and it's obviously reductive, but basically like the mix could shift um at their leisure, and you know, the business could be like the most profitable business in the world, right?
10:09
Unidentified speaker: And I think there's something interesting in that.
10:11
Unidentified speaker: The the flip side is that there, I mean, a couple of things like on a on an AI infrastructure trade basis, this is kind of potentially telling you that there's a lot of like latent supply available for inference.
10:24
Unidentified speaker: And so the we may have the if that dial can flip you know so quickly, then you know, we've we're kind of in this market where we think that uh there's there's there's no compute to no slack compute to satisfy demand.
10:37
Unidentified speaker: And you know, that's that creates the shortage of uh semiconductors and racks and power and all these different things.
10:42
Unidentified speaker: Um but if we could just shift that to inference kind of on a dime, uh particularly if you see lower expected ROI on training, then suddenly the market can get flooded a lot more quickly, you know, like almost instantly, with you know, more more inference capacity, and that can not be not be good for the overall kind of infrastructure build-out.
10:58
Unidentified speaker: Um, and even if that didn't happen, you're also kind of saying if if you keep training at these levels, especially not just at the mix level, but this mix plus the dollar levels we're talking about, you really need, especially if you think token prices generally kind of come down generation over generation and period over period.
11:17
Unidentified speaker: Um, and you know, you can have your nuances on what that means.
11:19
Unidentified speaker: But if you think intelligence is just getting cheaper and cheaper on a token basis, on an inference basis, then kind of P is going down.
11:26
Unidentified speaker: So you really, really need Q to kind of ramp up hard, which means you need uh a lot more kind of downstream users to kind of come in to use these tokens, right?
11:37
Unidentified speaker: I think if you go back up one, if you go back up the the interesting thing to me between that whole debate, and I think you can make interesting cases both ways, but like the is this this piece on the bottom, the the this third rail of well, what you gotta ask step back and ask like, well, what actually is the training like for?
11:52
Unidentified speaker: Is it RD for models that you know the labs or any like compute owner, right?
11:57
Unidentified speaker: Any model weight owner will use to sell you and me tokens as users for like our end uses, like whether that's in subscription plans or kind of API plans, whatever.
12:08
Unidentified speaker: Is it is all of this gonna be monetized by like you and me doing whatever we want with with tokens and building our businesses and doing our analysis and you know making our slot videos or whatever?
12:17
Unidentified speaker: Or is it actually going to be you know verticalized and effectively insourced?
12:22
Unidentified speaker: And is is should you view this RD as effectively like Pfizer's RD or like you know, Raytheon's RD, right?
12:29
Unidentified speaker: Where they're doing RD for their own products that they're gonna sell to their own in in end users and they're gonna capture that value, right?
12:37
Unidentified speaker: And I think increasingly when you look at you know this past week, um Anthropic had an announcement on you know working on its own kind of internal drug discovery efforts.
12:43
Unidentified speaker: Um if you look in the S1 as well, there are comments about you know, we we may selectively compete with our end users in certain channels.
12:51
Unidentified speaker: You can kind of see like the writing on the wall, and I don't know how quickly this will happen.
12:56
Unidentified speaker: And it it there are natural channel conflicts with this that that may limit the degree to which they want to do this aggressively.
13:02
Unidentified speaker: Um, you can think about the Alex Carp you know Palantir argument of, you know, he was making this case like six months ago, very vocally on CNBC.
13:09
Unidentified speaker: But in any case, like this is as you think about how this ties in then to the sustainability of the trade and the even more importantly than that, kind of the geopolitical implications of it all.
13:20
Unidentified speaker: Um if this training is effectively for like US domicile, the the top US domiciled corporations, whichever ones ultimately get to be like the winners of this, to have kind of their own gate-kept you know frontier that basically, you know, a handful of companies are actually able to use.
13:37
Unidentified speaker: It's not for you and me, it's for like just a few users to go out and generate like 100x value on you know pharma pipelines or weapons manufacturing or whatever.
13:46
Unidentified speaker: Then maybe the whole like debate, this whole like bull versus bear debate is is almost like irrelevant, right?
13:51
Unidentified speaker: Like then the question becomes like, are they actually gonna be in a position to you know go launch their own kind of verticalized offerings into these you know very, very powerful end markets, big end markets, and you know, have like uh you know a permanent intelligence advantage, you know, in doing that?
14:07
Unidentified speaker: And what would that mean then for like uh the US versus the rest of the world?
14:12
Unidentified speaker: I think the the implications there might be might be pretty clear.
14:15
Unidentified speaker: Yeah, do you do you have any sense on what the window of time that they have to begin executing on that business line to and to show the market, like, hey, we're doing this before people fall out of favor with anthropic specifically or open, I guess you can apply this to open AI too.
14:32
Unidentified speaker: I'd imagine they have similar ambitions as well.
14:35
Unidentified speaker: I mean, I would imagine so as far as timing, like, you know, it's it's it's a high wire act because they do have to like the this whole like selling you and me tokens and and selling metered intelligence business, like is for now, like that's how that's how the you're generating operating income and free cash flow.
14:52
Unidentified speaker: And that's in the very near term, like that's how you keep the wheel spinning and and financing, you know, coming in.
14:59
Unidentified speaker: Um and I would guess so they do have to continue doing that.
15:02
Unidentified speaker: And like I said, then there are like channel conflicts there because everyone knows that like they're doing that, and all their you know, this this supplier, this vendor effectively is competing with me for my alpha.
15:12
Unidentified speaker: So then um what what does that mean I want to do for uh self-hosting or uh you know sovereign deployments and owning you know much more data and uh leaking less and less to the labs?
15:22
Unidentified speaker: Um, that's gonna be like a delicate, a delicate advance and delicate balance.
15:26
Unidentified speaker: I imagine the first like real iteration of this is something that looks like you know, JVs and like partnerships and you know upside sharing of uh if you're if you're working with Pfizer, right, on some drug discovery pipeline or some biotech or working with a defense contractor on like you know leading edge kind of weapons technology or uh semiconductor industry or something like that.
15:46
Unidentified speaker: Like uh you can have kind of a potentially like win-win scenario where you know the the labs get to potentially uh keep some level of like uncapped upside of you know whatever results from using the the true like gate kept frontier tokens.
16:01
Unidentified speaker: I imagine that's probably how you'd see it first play out.
16:04
Unidentified speaker: And it may ultimately be that may be ultimately like the the sustainable equilibrium that keeps everyone kind of happy enough.
16:12
Unidentified speaker: And so maybe it never becomes like you know, Dario has said reportedly that eventually like Anthropic is going to be like the one company that exists in the world.
16:20
Unidentified speaker: So I don't that would be like the the extreme vision of this, that they just because they have like you know 100x advantage in intelligence, they therefore just get to own every industry.
16:29
Unidentified speaker: I don't really think that's how it would play out for a variety of reasons.
16:31
Unidentified speaker: But uh yeah, I would expect you'd see something like you know, JVs, partnerships, profit sharing, that kind of thing.
16:36
Unidentified speaker: Yeah.
16:37
Unidentified speaker: And I wanted to pull this up too, because I think it's uh pertinent to the discussion just on a lot of conversations, particularly in the realm of the build-out of the data centers.
16:47
Unidentified speaker: A lot of the developers and people who have power and land assets are looking for credit wrappers, and the ideal credit wrapper comes with an IG-rated counterparty.
16:58
Unidentified speaker: Um and there was news on that per the Financial Times this week.
17:01
Unidentified speaker: Anthropic is now expected to receive a high junk rating from major credit rating agencies after its IPO, which could come as soon as November.
17:08
Unidentified speaker: Details include Blackstone has told investors that Anthropic likely will not receive an investment grade rating immediately after its IPO.
17:15
Unidentified speaker: Blackstone is working with banks that stumped up financing on selling debt onto other investors.
17:21
Unidentified speaker: An investment grade rating is central to Anthropic's ability to continue its AI build-out.
17:26
Unidentified speaker: And the company says it expects to spend over $800 billion on computing infrastructure between now and 2045.
17:31
Unidentified speaker: So I think another important data point to be paying attention to.
17:35
Unidentified speaker: I don't think anybody expected them to get IG rating out of the gate, but high junk is looking like likely according to Blackstone and the Financial Times.
17:45
Unidentified speaker: And so it'll be interesting too to see if they have a few quarters of positive momentum, they're they're increasing that that cash flow and that income, and they're becoming more efficient with their compute spend.
17:58
Unidentified speaker: Could they get IG rating, get lower costs of capital, and then accelerate the flywheel will be interesting to see.
18:04
Unidentified speaker: Yeah, it'll be important for sure.
18:05
Unidentified speaker: Yeah.
18:06
Unidentified speaker: And so where's the ROI?
18:07
Unidentified speaker: I mean, you mentioned it briefly alluded to it, but I think digging deeper into the fine details of what will enable these companies to actually be successful and thread the needle of bringing in and winning the AI race.
18:23
Unidentified speaker: Yeah, I mean, ultimately, we'll get to, I think the the next slide will maybe have a little color on how how it this is all actually being perceived.
18:31
Unidentified speaker: But like the where's the ROI is like the real question now.
18:34
Unidentified speaker: You know, we're we're past the point of kind of dream the dream, you know, honeymoon phase of this is a crazy technology, it's gonna change the world, et cetera.
18:40
Unidentified speaker: You know, at this, at this scale, you know, where you're talking about percentage points of GDP in terms of you know the capex that's gonna be required to stay sustain this and really make it work.
18:49
Unidentified speaker: You know, this is the you're you're seeing this all over the place of um executives and you know, private equity and hedge fund analysts, et cetera, asking asking this question, putting out white papers to this effect.
18:59
Unidentified speaker: I think Apollo again had a had a paper last week basically saying, you know, they they're not seeing any ROI beyond tech.
19:06
Unidentified speaker: And I think there was a decent, there were several data points last week that I thought uh whether whether anthropic and open AI ultimately kind of own all of the alpha or whether they, you know, they blow up and that alpha gets much more broadly distributed to kind of the existing companies that you know operate in these verticals, regardless of of who kind of wins.
19:23
Unidentified speaker: I think you saw some really interesting data points last week from uh Goldman and a few others suggesting they are actually seeing clear signs that AI is becoming, is driving notable ROI and kind of measurable ROI beyond uh beyond just kind of tech and the San Francisco bubble.
19:38
Unidentified speaker: Um, you can see a couple stats on here.
19:40
Unidentified speaker: I think maybe the most obvious one, right, is and this isn't necessarily beyond tech, but Microsoft announced last week that they're gonna cut headcount or at least offer buyouts for 60,000 roughly employees, which is essentially a quarter of global workforce.
19:55
Unidentified speaker: Um, so it's a it's a massive, uh, would be a massive cut kind of explicitly to help fund AI investment and kind of depending on what the ultimate packages are like and depending on who's actually being replaced, you know, this is uh potentially like tens of billions of dollars of savings, not going to fund the entire next five years of build by itself, but you know, definitely tells you a lot about where kind of big, what you know, the biggest corporations in the world and are seeing from uh both the opportunity to invest in this, but also on the back end of how what it's enabling for them.
20:27
Unidentified speaker: So collectively, you know, these are not all like good data points.
20:30
Unidentified speaker: I acknowledge that it's especially the Microsoft headline is uh problematic for uh social stability if you extrapolate it out quickly enough and broadly enough.
20:40
Unidentified speaker: But in general, I think it's still telling you the the ROI uh at the corporate level is is kind of becoming more and more clear.
20:46
Unidentified speaker: And so that's gonna be very worth uh monitoring over the next six to twelve months as those CapEx numbers just get you know bigger and bigger.
20:54
Unidentified speaker: Yeah.
20:55
Unidentified speaker: You alluded to it, but the next slide talking about Red October, the uh the correction that was last week, looking at equity indices right now across the board, everything is up more than a percent except for the S P 500, which is up 98 bips right now.
21:10
Unidentified speaker: Looks like we have a bit of a recovery, whether or not that is a dead cat bounce or last week was was a scare that that was trying to shake people out is yet to be determined.
21:21
Unidentified speaker: But yeah, it was pretty frothy last week on multiple fronts.
21:25
Unidentified speaker: Yeah, no, for sure.
21:26
Unidentified speaker: Yeah, I think the the slide kind of speaks for itself.
21:29
Unidentified speaker: Uh I mean the the Intel print is is maybe the poster child for this this whole thing, right?
21:33
Unidentified speaker: We've had uh we we talked about it on a prior episode months ago, but just this this crazy parabolic rip, a lot of which was basically just multiple expansion.
21:40
Unidentified speaker: Um and this is like the flip side of that when you're you know generating huge gains primarily on kind of uh rapid re-rating of forward expectations, and there's a lot of error under the stock, and if things start to get a little hairy, and uh you can see you know uh day days like Friday where Intel had its you know worst, worst intraday move or worst single day move since uh 1974.
22:00
Unidentified speaker: Uh that's kind of that's kind of what you get uh from something like that.
22:03
Unidentified speaker: And um, I think interestingly too, there's data going around from Morgan Stanley on on the right showing that uh this is the biggest degrossing event uh for hedge funds taking down exposure uh on record was uh was over the course of October, which is is is also interesting given you know we're still relatively near all-time highs on the S ⁇ P in the NASDAQ and the underlying, like we just pointed out, there are a lot of data points from model progress to to ROI on AI becoming clearer and clearer.
22:31
Unidentified speaker: And yet you're seeing this huge puke and a lot of the the most AI lovered names.
22:35
Unidentified speaker: And I think it just goes back to you know, this line I have on the bottom that the market is getting antsy about the mystery box.
22:40
Unidentified speaker: If you're uh if you if you're 90s kids will remember uh Family Guy and the uh the the classic moment where Peter Griffin is is offered a uh either a boat or the mystery box.
22:52
Unidentified speaker: And he yeah, instead of taking the boat, he's he takes the mystery box on the logic that it could be anything, it could even be a boat.
22:58
Unidentified speaker: And I the the that's always stuck with me, and I think that's really what investors love more than anything else is like the uncapped, unquantifiable upside.
23:05
Unidentified speaker: You know, they when when the rubber meets the road and when actually things have to get executed against, you know, and expectations have moved up a lot.
23:12
Unidentified speaker: That's where things get really hairy and choppy.
23:14
Unidentified speaker: And you know, the as the details start to emerge on things and performance becomes clearer and clearer, that's can often be a very, very choppy time.
23:21
Unidentified speaker: And we are firmly past, I think this slide is showing you, firmly past, you know, dream the dream, infinite TAM mystery box territory here.
23:28
Unidentified speaker: I love that line.
23:30
Unidentified speaker: Then moving on, we we touched on this last week, but I think really diving into it earnestly is strong prints, thinner cushion, this idea of all these hyperscalers and these frontier labs going cash flow negative and for Google, Meta, Amazon, Microsoft going cash flow negative for the first time in many years.
23:48
Unidentified speaker: We talked about Meta and the offering that they may, the equity offering that they may tap in the future.
23:55
Unidentified speaker: I think Google really set the trend earlier this summer when they raised 85 billion in equity for the first time uh in quite a while.
24:02
Unidentified speaker: And I think that's another big sub-theme here is how much or how long can the hyperscalers, after years of uh incredible free cash flow, can they stomach negative cash flow and can they actually articulate the the narrative to the market that it's gonna be worth all this capex expenditure?
24:20
Unidentified speaker: Yeah, like we talked, we had a slide, uh, a chart like this one last week.
24:23
Unidentified speaker: It was fairly focused on Meta.
24:25
Unidentified speaker: I mean, this is all the the big four hyperscalers, and you can see, you know, red bar now above blue bar, uh basically free cash flow negative for the cohort um as of as of Q3.
24:34
Unidentified speaker: And you know, this is the the key is just gonna be when does blue bar get back above red bar, basically, um, for people to kind of believe that this is you know gonna be durable and sustainable and create sustainable equity value.
24:45
Unidentified speaker: Um, and I think it's it's important too for everything else, because as we uh show on the left here, GDP growth was uh for Q3 was above expectations.
24:55
Unidentified speaker: You know, still we're still not at like 5% plus, but 3.4%, a lot of which was driven by the this capital goods spending, and then manufacturing PMIs back up to uh highest since May 2022.
25:06
Unidentified speaker: You know, if you look at that chart, it's it's pretty much up into the right for the last uh 18 months or so.
25:11
Unidentified speaker: And you really need that growth.
25:13
Unidentified speaker: This is like these two things are kind of like linked at the hip.
25:16
Unidentified speaker: This is really what's driving a lot of the growth story.
25:18
Unidentified speaker: You've had kind of this relatively you know easy funding from the best, the best uh businesses in the world, some of the highest margin businesses in the world that have fantastic returns on capital and can grow to these insane levels without all that much investment.
25:31
Unidentified speaker: Just, you know, these uh for for a long time where these buyback machines, you know, they they spit off so much free cash flow that the only thing to do with it was return it to shareholders aggressively through through buybacks.
25:40
Unidentified speaker: And, you know, for a couple of years, you could just uh flip the switch from instead of doing buybacks and and dividends, you know, it's just going to become AI capex and you know, the underlying businesses keep keep churning and you know, Instagram and Google, et cetera, all still work very well in Throw off Cash.
25:55
Unidentified speaker: So you just kind of you know siphon it out from that.
25:58
Unidentified speaker: But now you're at the point where the mystery box is is kind of evaporating because the you you've gotten to the the zero level or the negative level on that engine.
26:05
Unidentified speaker: Like we're spending now all that that engine can kind of easily provide.
26:09
Unidentified speaker: Um and we didn't talk about it on uh the last slide, we don't have to go back to it.
26:13
Unidentified speaker: But as it relates to that engine, interesting too that you know Google pointed out on their call this week that AI mode through search is not monetizing nearly as well, which is something that you know various analysts have been talking about for years that this was gonna be ultimately a threat to search.
26:27
Unidentified speaker: Um, you had similar comments from you know, as it relates to traffic from search on from from Reddit's CEO on their you know negatively received earnings call.
26:35
Unidentified speaker: So as it relates to like that engine that's been funding this stuff, like it's some some of those engines may be starting to get kind of a a little creakier, right?
26:42
Unidentified speaker: So that's that's not gonna help either.
26:44
Unidentified speaker: Um, but it's all very important for everything that we always talk about on the show to keep that you know manufacturing PMI moving, to keep GDP growth kind of up and to the right as much as possible through this, especially when we officially crossed over uh 100% of debt to GDP now held by the public for the first time in in basically 80 years.
27:02
Unidentified speaker: Um and, you know, I think the there's clearly a there are a lot of different ways that you could cut it up, but there's clearly some sort of debt reckoning like happening right now in global markets and especially the US.
27:14
Unidentified speaker: I think as we've discussed, the US is kind of the tallest of the seven dwarves and in a relatively advantaged and strong position.
27:20
Unidentified speaker: But you know, there's no magic button to make it go away.
27:23
Unidentified speaker: And so the this chart here basically and and you know, versions of it need to keep moving up and to the right, and blue bar eventually in the not too distant future needs to get above red bar such that the you know the denominator of Decha GDP, the growth engine, can start to work the magic that you know the the administration is hoping it will.
27:41
Unidentified speaker: Yeah, it's very interesting too.
27:42
Unidentified speaker: It'll be what I'm going to be paying attention to is a divergence between Meta's ad business and Google's, because I think Google search is very boring for lack of a better term.
27:52
Unidentified speaker: You search and yes, you get delivered ads.
27:54
Unidentified speaker: There's a thread going around in July of somebody who spends something like 500 grand a month on Google Ads alone, and he was highlighting how the quality of the ad search went down dramatically.
28:05
Unidentified speaker: And it it the case that he was putting forth is that Google was intentionally deprecating the quality of ads just to juice revenue to make it look like that that business was still doing well.
28:17
Unidentified speaker: But he was this gentleman was sounding an alarm bell in July that all was not well with Google search and the LLMs were definitely cannibalizing search revenue for Google.
28:26
Unidentified speaker: But when it comes to Meta, when you have Instagram, Facebook, and now Muse as well, but I think Instagram specifically, and I think that's very underappreciated and not well known by the market, but the first really AI implement implementation that the Meta did was with their ad product and targeting, and it's an incredible product.
28:44
Unidentified speaker: And I think the the sort of lock-in with Meta's ad platform that is driven by the engagement of the content that the ads are being uh served through is interesting.
28:55
Unidentified speaker: That'll be an interesting thing to watch is meta ads versus Google Ad Revenue moving forward, because I still think Meta has an advantage due to the engagement that their platform drives.
29:05
Unidentified speaker: So they have very engaged users.
29:07
Unidentified speaker: And then on top of the ad business, the AI implementation they've been working on most heavily is how to keep those users engaged.
29:14
Unidentified speaker: Uh time on screen, times you're opening up the app.
29:17
Unidentified speaker: And I think those metrics are up and to the right, too.
29:19
Unidentified speaker: So there could be a divergence there between Meta and Google shifting to geopolitical the geopolitical realm, big big elections across the world.
29:29
Unidentified speaker: Last week, particularly in Israel and Brazil, we were going to focus on the latter here in Brazil, where Lula surprisingly pulled out a win after some shenanigans at the Supreme Court level.
29:43
Unidentified speaker: Yeah, you know, I'll leave everyone else to draw their own conclusions on uh the various issues of election integrity, et cetera.
29:50
Unidentified speaker: Uh but uh in any case, we're you know in a position where a couple weeks ago, I think we talked about Bolsonaro pulling ahead unexpectedly, you know, EWZ, the Brazil ETF, and you know, Brazilian stocks generally ripped on that.
30:03
Unidentified speaker: Um, and then it suddenly kind of come to find out Monday or Sunday overnight that in fact Lua had won.
30:09
Unidentified speaker: Um Bolsonaro had to flee the country, um, which is always you know a great a great sign for the for the functioning of any country's government when both sides are effectively threatening to arrest and murder each other.
30:19
Unidentified speaker: Um, it's always a sustainable way to do business.
30:21
Unidentified speaker: But you know, EWZ sold off aggressively on that, and perhaps not surprisingly, but I think uh for our purposes, what's what's notable here is this is a you know significant divergence from the what we've been seeing over the last basically entire year in Latin America, which uh is is important for very important for the the so-called Don Roe doctrine.
30:41
Unidentified speaker: Um, Latin America, various Latin American countries had clearly gotten cosier with China in different ways and received more investment from them and gotten more kind of uh felt more asset or experienced more uh growth in Chinese asset ownership over the last decade or so.
30:56
Unidentified speaker: And in particular, Brazil is you know one of the largest economies in the hemisphere and most notably has large reserves of various rare earth minerals.
31:05
Unidentified speaker: And the US actually earlier this year, we highlighted you know, struck a deal with uh with Brazil or Brazilian company to have uh advantaged access to some of those.
31:13
Unidentified speaker: But I think that was that looked a lot different when it was under the uh the assumption that the whole kind of continent was was shifting one way in kind of a Trump friend friendly direction.
31:22
Unidentified speaker: And I think this is like a very obvious speed bump against that.
31:26
Unidentified speaker: You know, no idea ultimately how that's gonna play out.
31:29
Unidentified speaker: My my instinct is we haven't heard kind of the last of it, but especially when you consider um this this Trump uh this very Trump-esque post or a quote that he that he gave us at a at a rally over this past week that you know maybe maybe someday soon Venezuela will become a prosperous 51st state.
31:46
Unidentified speaker: He's floated that before.
31:47
Unidentified speaker: Um but clearly the there's there's remaining it remains a lot of interest from the the administration to continue cozying up to Latin America.
31:54
Unidentified speaker: I don't imagine this is going to be the end of the story, and we'll we'll we'll see what happens.
31:59
Unidentified speaker: It's it's no doubt a uh an interesting roadblock, though.
32:02
Unidentified speaker: Election integrity questions in Brazil, coup d'etat attempts in Poland.
32:07
Unidentified speaker: It's a crazy world out there.
32:08
Unidentified speaker: It's a crazy world out there.
32:09
Unidentified speaker: We'll be paying attention to this.
32:11
Unidentified speaker: And it would be remiss of us not to cover the big news in our corner of the internet, in our corner of the world, our domain, which is Bitcoin, the largest exchange in the world.
32:22
Unidentified speaker: They serve as Bitcoin and a number of other tokens, Binance, was hacked for the largest hack, I think, of any business, other than a crypto exchange.
32:32
Unidentified speaker: $6.3 billion, largest crypto theft on record.
32:36
Unidentified speaker: They paused withdrawals, set aside $6.34 billion to cover all losses.
32:41
Unidentified speaker: Bitcoin slipped below $77,000, yet still closed above its 50-week moving average.
32:46
Unidentified speaker: Uh Bitcoin is currently sitting at let's see here, $77,352.
32:53
Unidentified speaker: So still holding above the 50 day.
32:55
Unidentified speaker: We'll see if it maintains that.
32:57
Unidentified speaker: But yeah, this is a pretty insane story.
33:00
Unidentified speaker: It's I I think it's safe to say it's not exactly clear what's happened.
33:04
Unidentified speaker: There's uh we've got Binance's side of things where they're saying a DevOps engineer um was socially engineered by a North Korean spy who got access to his laptop and was able to initiate withdrawals from Binance's hot wallets across multiple chains, including Bitcoin.
33:21
Unidentified speaker: I believe 18,000 Bitcoin were swept, and then a number of other coins across the board, Ethereum, Solana, Tron, a lot of tether and USDC as well.
33:33
Unidentified speaker: The hackers were taking the funds, putting them in DeFi protocols with stable coins, taking using the stable coins to take out loans of ETH and Solana, I believe, at which point the stable coins were frozen, but they ran away with the ETH and Solana tokens, which are not as trivially frozen.
33:52
Unidentified speaker: Uh and yeah, a big black eye on the industry, very interesting timing in the week before midterm elections here in the United States and everything going on geopolitically with the war in Iran.
34:04
Unidentified speaker: And there's many lessons to take from this.
34:06
Unidentified speaker: One of which is make sure that your preferred exchanges, hot wallet infrastructure, isn't susceptible to a single point of failure like Binance's was at this scale.
34:17
Unidentified speaker: It's pretty insane that they had a single point of failure of this of this scale.
34:22
Unidentified speaker: Yeah, look, I think uh, you know, it's it's notable that you know they're gonna allegedly cover all losses.
34:26
Unidentified speaker: I think they reopened withdrawals late last night, and you haven't seen, I think you saw the spike in outflows, but that's calmed down since then.
34:32
Unidentified speaker: So it looks like that there's not an obvious like um kind of insolvency angle here, fortunately, or kind of general kind of run on the industry.
34:40
Unidentified speaker: All I'll say here is just it's it's it's humorous how you know history doesn't repeat, but it often rhymes.
34:44
Unidentified speaker: Four years ago, we were having the same conversation, or not the same conversation, it's very different with FTX being, you know, a a fraud in a Ponzi scheme, but massive exchange downside event in midterms week essentially pretty much put in the Bitcoin bottom.
34:57
Unidentified speaker: Um, and you know, no, no guarantees that it'll that'll happen the same way this time around, but just important to zoom out and remember that we're five X off of those lows and ultimately got up to 8x off that level.
35:08
Unidentified speaker: And again, I think like there's a little parallelism here where we only, you know, we dropped below the 80k level, so it lost that level.
35:16
Unidentified speaker: But to this point on Alex Alex Thorne's chart here, you know, we've we've held in for these rel these technical levels that are important in the short term.
35:24
Unidentified speaker: And you know, end of the day, like being down a couple grand on this headline is you know, not exactly to me, it looks a lot like what you would see toward the tail end of a bottom, kind of like what you saw, or the the tail end of a bear, kind of like what you saw with with FTX, where you know it's kind of an extinction level event for all of crypto and Bitcoin was down uh you know uh 10% or something, or I can't remember exactly what it was, but um, it was not down nearly as much as I think people would have thought in isolation, you know, knowing none of the other facts.
35:51
Unidentified speaker: So similar, kind of potentially similar dynamic here that's just worth uh worth noting.
35:55
Unidentified speaker: Yeah.
35:56
Unidentified speaker: And last but not least, to highlight Bitcoin's relative advantage as we would describe it to other chains.
36:02
Unidentified speaker: You have an incredible title here, one of my favorite blog posts ever written about Bitcoin by the infamous Gwern, Bitcoin is worse is better.
36:12
Unidentified speaker: And I think one of the key details of this Binance attack was uh reportedly ran through EIP 7702, which is an Ethereum improvement proposal that was implemented, I believe, in May of last year.
36:26
Unidentified speaker: And so it's a delegation feature that Ethereum added to make ordinary accounts behave like smart contracts uh the same week.
36:35
Unidentified speaker: 16 AI agents spent a week on Bitcoin Core and came back with two minor items.
36:41
Unidentified speaker: So we're just juxtaposing this this EIP 7702 was a critical uh variable to the Binance Hack.
36:49
Unidentified speaker: I believe they were tunneling the the sort of uh what's the word I'm looking for?
36:55
Unidentified speaker: The go ahead to sign multiple private keys across multiple chains through this delegation feature that Ethereum added last year to make accounts like smart contracts.
37:05
Unidentified speaker: And it just highlights one of the risks that we've been talking about 1031.
37:11
Unidentified speaker: I'm personally when it when it comes to comparing Bitcoin to other chains like Ethereum and Solana, is that Bitcoin is worse, is better, it's simple, it's stupid, it's very easy, simple protocol with a limited, a relatively limited attack surface compared to chains like Ethereum and Solana, which have robust smart contracting and virtual machines built into what they do, and that increases the the attack surface, which Ethereum did that last year, and that attack surface was exploited last week with the Binance hack.
37:43
Unidentified speaker: Yeah, co-sign all that for something to be you know the it with everything we talk about on the show and the world kind of moving into you know this this potential phase shift into a totally new kind of financial architecture and substrate, you know, if if uh an asset is going to be a meaningful part of that substrate, you know, minimizing attack service and complexity is I think uh a must-have.
38:03
Unidentified speaker: And I think this this juxtaposition is is you know the the latest and and one of many that we've talked about uh over the over the years, kind of pointing in that direction for Bitcoin.
38:12
Unidentified speaker: Yeah.
38:12
Unidentified speaker: Bitcoin core, all the frontier models pointed at it.
38:16
Unidentified speaker: This is a Coinbase experiment or an exercise by the chief security officer.
38:22
Unidentified speaker: At Coinbase, Jeff Longhofer, and he he reported last week, hey, we we pointed these frontier models at Bitcoin Core for weeks and only came back with two minor bugs that have since been patched.
38:32
Unidentified speaker: So Bitcoin Core, one of the most reviewed core code bases in the world, even before AI frontier models, and I have a feeling moving forward the the review will will only be better.
38:43
Unidentified speaker: See you guys next week.
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3 and interpreted by Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.

More from TFTC: A Bitcoin Podcast

TFTC: A Bitcoin Podcast

#810: Bitcoin Is in the Discount Zone with James Check

Episode discussion: James Check argues Bitcoin is in a discount zone with institutional normalization and strong on-chain holder behavior

Speakers’ claims: James Check and speakers analyze Bitcoin's price action, identifying the current range as a 'discount zone' following a rally to $87K and subsequent consolidation. They highlight institutional normalization through new ETF products and bank recommendations, contrasting this with retail capitulation. On-chain data suggests long-term holders are not selling, while short-term holders are absorbing supply. The discussion also covers stablecoin dynamics, AI infrastructure resilience, and political implications for monetary policy.

Bitcoin Price AnalysisInstitutional AdoptionOn-Chain Metrics
1h 18m6 assets23 claims to verify
TFTC: A Bitcoin Podcast

Ten31 Timestamp: Everything's Bigger in Texas

Episode discussion: Speakers link midterm 'blue wave' to AI regulatory headwinds, treasury volatility, and Bitcoin's capital competition dynamics

Speakers’ claims: Speakers analyze the political impact of Democratic midterm gains, including a Texas governorship flip and local ballot measures, arguing these shifts create regulatory headwinds for the AI data center buildout. They discuss market volatility driven by hedge fund unwinds and weak Treasury bids, suggesting a correlation between AI trade fragility and bond market stress. The conversation covers geopolitical tensions, citing China's export strategies and European instability, while debating whether AI's capital intensity crowds out Bitcoin or if AI commoditization will eventually benefit crypto. Finally, they highlight accelerating AI capabilities at firms like Anthropic despite valuation corrections, concluding that Bitcoin has multiple value pathways regardless of AI's economic outcome.

Midterm ElectionsAI InfrastructureTreasury Markets
44 min3 assets19 claims to verify
All insights