Facts vs Feelings with Ryan Detrick & Sonu Varghese

Sonu Varghese warns of AI bubble fragility, negative hyperscaler cash flow, and risks in neo-cloud debt and market concentration.

1h 2m Listen to the original
AI BubbleHyperscaler Cash FlowNeo-Cloud DebtUS-China AI StrategyMarket ConcentrationCybersecurity RisksHealthcare AI

Summary

Sonu Varghese argues that the AI sector is experiencing a bubble characterized by excessive investment relative to economic substance, comparing current market weight to the historical railway bubble. He highlights that major hyperscalers are generating negative free cash flow and that neo-cloud providers like CoreWeave face distress due to drying debt financing. Varghese contrasts the US focus on frontier AGI with China’s strategy of rapid industrial diffusion, suggesting the US-China race is a 'Stag Hunt' where cooperation yields better outcomes. He also identifies Google’s AI search features as cannibalizing core ad revenue and advocates for diversification into low-volatility stocks and healthcare, which leverages proprietary data rather than open-source models.

Episode sections

  1. 0:36 – 7:34

    Intel Earnings and Market Reaction

    Discussion of Intel's stock drop following earnings, contrasting its decline with the stability of other tech giants like Amazon and Google.

  2. 7:34 – 13:05

    Hyperscaler Cash Flow and Neo-Cloud Debt Risks

    Concerns raised about negative free cash flow among major hyperscalers and the fragility of 'neo-cloud' companies like CoreWeave that rely heavily on debt.

  3. 13:05 – 14:10

    Anthropic’s Financials and Valuation Concerns

    Discussion of Anthropic’s downgrade to a near-junk credit rating and the discrepancy between reported revenues and actual cash flow.

  4. 14:10 – 18:05

    The AI Bubble: Beer vs. Soap Analogy

    Alvin Varghese introduces a 'beer bubble' analogy, arguing that while AI has real substance, the current pace of investment is creating excessive foam and fragility.

  5. 18:05 – 22:11

    US vs. China AI Strategy: Frontier vs. Diffusion

    Contrast between the US focus on frontier AGI development and China’s strategy of rapid deployment and diffusion of existing AI technologies.

  6. 22:11 – 27:02

    Game Theory: Stag Hunt vs. Prisoner’s Dilemma

    Discussion of game theory frameworks, with Varghese arguing that the US-China AI race is better described by the 'Stag Hunt' model.

  7. 27:02 – 34:29

    AI Cannibalization of Google Search Revenue

    Argument that Google’s AI search features are cannibalizing its core search revenue, leading to negative cash flow and missed earnings.

  8. 34:29 – 40:25

    US-China Chip Controls and Semiconductor Cyclicality

    Contention that US chip export controls are failing to slow China’s AI progress and are creating well-funded competitors like CXMT.

  9. 40:25 – 44:34

    Market Concentration and Low Volatility Diversification

    Discussion of the risks of extreme market concentration and the benefits of diversification into low volatility stocks and international markets.

  10. 44:34 – 48:26

    Healthcare as an Anti-AI Trade and AI in Drug Discovery

    Identification of healthcare as a strong diversifier and potential beneficiary of AI through specialized small models for drug discovery.

  11. 48:26 – 53:23

    Cybersecurity Vulnerabilities and Crypto Risks

    Discussion of the recent $6 billion Bitcoin theft and broader cybersecurity landscape, citing the PHP bug and high volume of critical CVEs.

  12. 53:23 – 56:01

    Liability and Accountability in AI

    Discussion of the lack of clear liability frameworks for AI agents causing harm, contrasting the Western approach with China's direct liability for executives.

  13. 56:01 – 57:36

    Risk Disclosures and Market Valuations

    Discussion on the magnitude of risk disclosures in AI filings, citing Anthropic's S-1 and comparing it to SpaceX's filings.

  14. 57:36 – 1:01:59

    AI as a Public Good vs. Weapon of Power

    Argument that AI is currently treated as a weapon for wealth and domination, leading to a destructive race, and proposals for reframing it as a global public good.

Assets discussed

AssetKindStanceContext
AlphabetGOOGLequitymixedYou know, one of the stars of this market has been Alphabet.
NVIDIANVDAequityneutralAnd so these companies that would have gone out of business if we let them buy NVIDIA chips and there would be no competitors there.
MicrosoftMSFTequityneutralUh, Microsoft had some pretty good news.

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

Claims worth verifying

Intel stock was down 30.88 percent on the day of the podcast.

Unidentified speaker ·5:00

How to verify: Check historical stock price data for INTC on the podcast release date (2026-11-03) or the trading day prior if the drop occurred then.

Intel's stock is up 61 percent year-to-date despite the daily drop.

Unidentified speaker ·7:08

How to verify: Calculate the percentage change in INTC stock price from the start of the calendar year to the date of the podcast.

Amazon stock was down 6 percent pre-market on the night before the podcast.

Unidentified speaker ·7:37

How to verify: Check pre-market trading data for AMZN on the evening of 2026-11-02.

Google stock was down 2 percent on the day of the podcast.

Unidentified speaker ·7:54

How to verify: Check historical stock price data for GOOGL on the podcast release date (2026-11-03).

CoreWeave has $50 to $60 billion of debt and is rated as distressed.

Unidentified speaker ·9:19

How to verify: Review CoreWeave's public debt filings, credit rating reports from agencies like S&P or Moody's, or news reports from the week prior to the podcast.

Anthropic had $16 billion in revenues for the first half of the year.

Unidentified speaker ·13:33

How to verify: Check Anthropic's S-1 filing or public financial disclosures for the first half of 2026.

Anthropic's Q3 growth was essentially flat.

Unidentified speaker ·11:25

How to verify: Review Anthropic's S-1 filing or subsequent financial updates for Q3 2026 data.

The AI sector represents around 45 percent of the total US stock market.

Unidentified speaker ·16:26

How to verify: Calculate the market capitalization of AI-focused companies (e.g., NVDA, MSFT, GOOGL, AMZN, META, etc.) as a percentage of the total S&P 500 or US equity market cap.

China's AI Plus Plan targets 70% of companies using AI by 2027 and 90% by 2030.

Unidentified speaker ·19:28

How to verify: Search for the official Chinese government 'AI Plus Plan' document or related policy announcements from 2026.

China is spending one-tenth as much on data centers as the US, with China spending under $100 billion and the US over $1 trillion.

Unidentified speaker ·20:00

How to verify: Compare public data on national AI infrastructure spending or data center capital expenditure reports for China and the US for the relevant year.

Alphabet missed revenue estimates for Google Services, reporting $39.2 billion versus an expected ~$41 billion.

Unidentified speaker ·

How to verify: Check Alphabet Inc. (GOOGL) Q3 2024 earnings release or SEC Form 10-Q for Google Services segment revenue.

Alphabet experienced two consecutive quarters of negative cash flow.

Unidentified speaker ·

How to verify: Check Alphabet Inc. (GOOGL) recent quarterly earnings reports for free cash flow figures.

Alphabet's market cap dropped approximately $500 billion in a single day.

Unidentified speaker ·

How to verify: Check historical market cap data for Alphabet Inc. (GOOGL) on the date of the 'big down day' mentioned (likely late October 2024).

US holds 45-47% of total global AI-related profits, all in hardware.

Unidentified speaker ·

How to verify: This is a specific statistical claim by the speaker; verify via industry reports from firms like Gartner, IDC, or McKinsey on global AI profit distribution.

Korea holds 35% of global AI hardware profits.

Unidentified speaker ·

How to verify: Verify via semiconductor industry financial reports detailing profit margins of SK Hynix, Samsung, and other Korean entities.

Taiwan holds 7% of global AI hardware profits.

Unidentified speaker ·

How to verify: Verify via semiconductor industry financial reports detailing profit margins of TSMC and other Taiwanese entities.

China holds 5% of global AI hardware profits.

Unidentified speaker ·

How to verify: Verify via semiconductor industry financial reports detailing profit margins of Chinese chipmakers.

Korea's annualized trade surplus is six times what it was a year ago.

Unidentified speaker ·

How to verify: Check Bank of Korea or UN Comtrade data for Korea's trade balance year-over-year comparison.

High bandwidth memory (HBM) prices increased by 300% or more this year.

Unidentified speaker ·

How to verify: Check pricing data from memory market analysts like TrendForce or SemiAnalysis for HBM price trends.

CXMT became the most valuable company in China on its first day of trading.

Unidentified speaker ·

How to verify: Check news reports and market data from CXMT's IPO date (2024) regarding its market capitalization relative to other Chinese firms.

China has direct liability for companies and executives if AI models create harm in society.

Unidentified speaker ·54:50

How to verify: Research Chinese regulations on AI governance, specifically the 'Interim Measures for the Management of Generative Artificial Intelligence Services' (2023) and related liability frameworks for AI developers.

Anthropic's S-1 filing contains 90 pages of risk disclosures.

Unidentified speaker ·56:38

How to verify: Locate and review the text of Anthropic's S-1 registration statement filed with the SEC, specifically the 'Risk Factors' section.

SpaceX's S-1 filing contained 68 pages of risk disclosures.

Unidentified speaker ·56:42

How to verify: Locate and review the text of SpaceX's S-1 registration statement filed with the SEC, specifically the 'Risk Factors' section.

93% of SpaceX's revenues were from non-space-related businesses, specifically AI businesses.

Unidentified speaker ·56:52

How to verify: Review SpaceX's financial disclosures in its SEC filings or reputable financial news reports detailing SpaceX's revenue breakdown between space launch services and Starlink/AI software services.

SpaceX's total available market (TAM) was valued at $40 trillion, attributed to software, AI, and application services.

Unidentified speaker ·56:52

How to verify: Search for SpaceX's investor presentations or SEC filings that define their Total Addressable Market (TAM) and the specific breakdown of that valuation.

Do your own research

  • Verify the free cash flow statements of major hyperscalers (e.g., Microsoft, Alphabet, Amazon, Meta) for recent quarters to confirm negative cash flow trends.
  • Review the S-1 filings and credit ratings of neo-cloud providers like CoreWeave to assess debt maturity schedules and reliance on chipmaker-backed financing.
  • Analyze Anthropic’s S-1 filing for risk disclosures and revenue recognition policies, specifically distinguishing between transfer pricing and direct cash receipts.
  • Check the latest S&P 500 sector weightings to confirm if technology and related sectors comprise over 50% of the index.
  • Review China’s 'AI Plus Plan' and official government targets for AI adoption rates by 2027 and 2030.
  • Examine Google’s (Alphabet) earnings reports for trends in search advertising revenue versus AI-driven query consumption.
  • Research the performance correlation of low-volatility ETFs (e.g., USMV, SPLV) against the S&P 500 over the last 12-24 months.
  • Investigate recent cybersecurity incidents in decentralized finance (DeFi) and the average patch times for critical CVEs in relevant software stacks.
Full transcriptexpand
0:00
Unidentified speaker: That's the advisor services are offered CWM LLC.
0:03
Unidentified speaker: SEC registered investment advisor is reporting its information purpose only as intended for an audience financial effect.
0:13
Unidentified speaker: Of CWM LLC, guest presenters over show are not affiliated with CWMLLC.
0:36
Unidentified speaker: I don't know.
0:37
Unidentified speaker: All right.
0:38
Unidentified speaker: I don't know who's cheering.
0:39
Unidentified speaker: There's not much to cheer about today.
0:42
Unidentified speaker: Yeah, although Alvin, expected.
0:46
Unidentified speaker: I think that's that's the thing.
0:48
Unidentified speaker: All right.
0:48
Unidentified speaker: Well, glad you could make it, Alvin.
0:50
Unidentified speaker: Sonu, Ryan here as always.
0:52
Unidentified speaker: Thanks to everyone who's out there listening to this live stream of the social hour.
0:55
Unidentified speaker: Uh we are so excited.
0:56
Unidentified speaker: And it's like, Alvin, we sent you a bunch of questions.
0:59
Unidentified speaker: And I was joking with Sony the last 24 hours.
1:01
Unidentified speaker: I think everything just changed.
1:02
Unidentified speaker: We've got we've got more we want to dive into.
1:04
Unidentified speaker: We have a lot to talk about.
1:05
Unidentified speaker: So uh yeah, we've got to be so bullish.
1:09
Unidentified speaker: I'm wearing my horns, but I will say, so here's my Dietrich whiskey.
1:14
Unidentified speaker: The way this is going, I might just chug this live on on air the way this the way this did.
1:18
Unidentified speaker: I just have a diet coke.
1:20
Unidentified speaker: Okay.
1:20
Unidentified speaker: I've got my coffee.
1:22
Unidentified speaker: I'm I'm ready.
1:22
Unidentified speaker: Now, Alvin, I want to we want to get into it.
1:24
Unidentified speaker: You're I'm heading to the airport like almost as soon as this is done in about 59 minutes.
1:27
Unidentified speaker: You're going to the airport too.
1:28
Unidentified speaker: Is that correct?
1:29
Unidentified speaker: Leaving at one o'clock to the airport, which is in an hour.
1:32
Unidentified speaker: So there you go.
1:32
Unidentified speaker: So we all have a hard hour.
1:35
Unidentified speaker: Sono can just talk by himself if he wants, but but Ryan and Alvin have to have to get rolling.
1:39
Unidentified speaker: Um, well, Alvin, we are so excited you could join us.
1:42
Unidentified speaker: And honestly, the timing of this is really nice because everything going on, all this uncertainty that's happening.
1:47
Unidentified speaker: Um, you know, we're gonna we're gonna dive into all of it.
1:50
Unidentified speaker: And I'm gonna ask Sonu some questions, I'll chime in.
1:52
Unidentified speaker: But I guess first things first, Alvin, I always like to start it with a kind of a layup.
1:55
Unidentified speaker: What do you do all day?
1:56
Unidentified speaker: Let the listeners know who don't quite know um what you're doing.
1:59
Unidentified speaker: We do like titles, we live in a world of compliance, whatever your official title is, just to keep the compliance officers happening.
2:05
Unidentified speaker: Sure.
2:05
Unidentified speaker: I I've got a lot of hats actually at the same time.
2:08
Unidentified speaker: So I am uh professor of AI policy at Union Washington.
2:11
Unidentified speaker: Uh I'm a digital fellow at the Stanford Uh Human Center AI Institute and a senior fellow at the Asia Society Policy Institute.
2:20
Unidentified speaker: So uh and an author and speaker.
2:23
Unidentified speaker: So how did you get involved in AI?
2:26
Unidentified speaker: And you know, I know we've chatted before.
2:28
Unidentified speaker: I'd love for the listeners to hear your background because you have a long background in technology and then policy advice and not to mention writing, right?
2:36
Unidentified speaker: Yeah, I I've been you know developing AI since the early 90s.
2:40
Unidentified speaker: So um about 35 years.
2:42
Unidentified speaker: I've I've you know we're working on uh natural language uh processing and neural networks back in the 90s, and then uh you know I was also in semiconductors.
2:51
Unidentified speaker: I actually designed the initial instruction set for Intel uh that was about 10 years ahead of uh NVIDIA and doing uh multimedia processing.
3:01
Unidentified speaker: It was called multimedia extension.
3:03
Unidentified speaker: And um, you know, uh Intel unfortunately didn't do much with it.
3:07
Unidentified speaker: And today is actually interesting.
3:09
Unidentified speaker: I was just going to say, yeah this is not a recommendation for anyone, by the way.
3:14
Unidentified speaker: You know, we mentioned some names and Alvin obviously I've actually worked in all five layers of the of the uh NVIDIA or the Jensen Huang AI cake.
3:24
Unidentified speaker: So everything from data centers to chips to AI models to applications to devices.
3:30
Unidentified speaker: So I I can and I've done it in uh you know the two sides of the Pacific.
3:34
Unidentified speaker: So half my career has been in China, half of it has been in the US.
3:37
Unidentified speaker: So just to give a little background.
3:40
Unidentified speaker: That's fascinating, especially that connection between China and the US.
3:44
Unidentified speaker: And we'll dig into you know how you think about that.
3:46
Unidentified speaker: I think you have a very unique perspective there.
3:50
Unidentified speaker: But just since you mentioned it, what was the big difference working on tech in China versus here?
3:58
Unidentified speaker: Um they they work at different speeds.
4:01
Unidentified speaker: I think both both, you know, if you're in Silicon Valley, things move very fast.
4:05
Unidentified speaker: The rest of America, actually, not that as much.
4:08
Unidentified speaker: Uh in China, everything is fast.
4:10
Unidentified speaker: And it's not just the software side, it's you know, it's uh infrastructure, target energy, it's robotics, it's you know, deployment.
4:18
Unidentified speaker: Um so uh it feels like when you're in Asia, everything's on fast forward, right?
4:24
Unidentified speaker: I think that that's the uh you know, I think there's a saying that uh Europe crawls, uh US walks, and China or Asia runs, right?
4:33
Unidentified speaker: So in some ways, I I feel like there is some truth in that.
4:37
Unidentified speaker: Parts of Asia, I should say.
4:38
Unidentified speaker: I originally from India, it's like there are parts of India that crawl, right?
4:42
Unidentified speaker: Yeah, no offense to anyone here, but yeah, maybe it's uh more East Asia or Southeast Asia a little bit.
4:49
Unidentified speaker: Sometimes I wish I could, you know, things in India would run faster, but you know, East Asia for sure.
4:55
Unidentified speaker: Let me let me just start with this.
4:57
Unidentified speaker: I'm just making sure I have the correct number.
4:58
Unidentified speaker: Intel, not a recommendation.
5:00
Unidentified speaker: That stock is down 30.88 percent.
5:04
Unidentified speaker: That is worse 87 crash, worse than anything financial crisis.
5:06
Unidentified speaker: This is like the worst day in the history in 50 years, I think.
5:09
Unidentified speaker: For now, you said you you you've got it some in you've got some insider, uh not inside, it's probably the wrong word to use, but you know about Intel because you've done a lot with them.
5:17
Unidentified speaker: Why in the world?
5:17
Unidentified speaker: I mean, yeah, they missed why at 31%?
5:20
Unidentified speaker: Do you have any thoughts on that, Alvin?
5:22
Unidentified speaker: Uh I mean, I I think the idea is that uh well they they've also gone up a couple hundred percent this year, right?
5:28
Unidentified speaker: So true.
5:29
Unidentified speaker: Um, so they're they're still significantly higher than they were the beginning year before the the various uh uh governments getting involved.
5:37
Unidentified speaker: So in in some ways, um you know, you know, and and they've actually come down almost 50% from the peak of just um maybe a couple months, a month ago, right?
5:48
Unidentified speaker: So uh today was it was a a uh a big drop, but I don't know if it's a um a sign of something as macro.
5:57
Unidentified speaker: Um clearly the the PC industry is is going through some some slumps and they are you know heavily PC centric in their revenue base.
6:06
Unidentified speaker: And you know, they they've also had a number of years of of uh I think poor management and poor decision making.
6:13
Unidentified speaker: I you know I was there during the boom boom years when when Andy Grove was was uh the CEO of we could do nothing wrong, right?
6:19
Unidentified speaker: We were the number the most valuable company in the world when I was there.
6:22
Unidentified speaker: So it was uh it was a glorious time for that company.
6:26
Unidentified speaker: Um the funny thing is I think it's it's actually worth less now than it was uh in the in around 2000.
6:33
Unidentified speaker: So wow, isn't that amazing?
6:36
Unidentified speaker: Wow, like I didn't realize that.
6:39
Unidentified speaker: Yeah, so or maybe maybe around because after the after this drop, it's probably uh around the same, but it you know, for for a very, very long time it was actually worth less than it was 20 years ago, which is kind of so I do wanna I'm gonna ask Sony a question here.
6:51
Unidentified speaker: This is let's see how good he is.
6:53
Unidentified speaker: So again, the stock's down 31 today.
6:56
Unidentified speaker: I just looked, so I know the answer.
6:57
Unidentified speaker: You know how much it's up year to date, Sony?
6:59
Unidentified speaker: You want to take a guess?
6:59
Unidentified speaker: I mean, I just said up, so now you know it's up, but how much do you think it's up year to date?
7:03
Unidentified speaker: Down 31 today.
7:04
Unidentified speaker: You want to take a guess, Sony?
7:06
Unidentified speaker: Let's say 70.
7:08
Unidentified speaker: Yeah, 61.
7:09
Unidentified speaker: The stock's up 60.
7:10
Unidentified speaker: No, I remember.
7:12
Unidentified speaker: I swear I didn't look this up.
7:13
Unidentified speaker: Down 30 today.
7:14
Unidentified speaker: So Alvin, that's a great point about the stretch.
7:16
Unidentified speaker: Part of it was Alvin just said, you know, maybe up 100% or something.
7:19
Unidentified speaker: I was like, okay, something that goes 100 and gets cut by 30, you get to 70.
7:23
Unidentified speaker: There you go.
7:24
Unidentified speaker: That's why I just get 70.
7:25
Unidentified speaker: There you go.
7:26
Unidentified speaker: No, could have been off there.
7:27
Unidentified speaker: But no, going back to Alvin's point, it's like obviously these reported earnings and weak guidance, PC weakness.
7:34
Unidentified speaker: What was interesting is the contrast to Amazon.
7:37
Unidentified speaker: Amazon was actually down a lot more pre-market last night.
7:40
Unidentified speaker: I think it was down six percent.
7:42
Unidentified speaker: This today doesn't look as bad, probably everything's relative, right?
7:46
Unidentified speaker: Well, everything doesn't look as bad when somebody's down 30.
7:49
Unidentified speaker: So Google's down only two percent.
7:54
Unidentified speaker: Right.
7:54
Unidentified speaker: What was interesting is you know, over there the concerns are more like AI spending concerns.
7:59
Unidentified speaker: Now they're trailing 12-month cash flow has turned negative, yeah.
8:04
Unidentified speaker: Right.
8:04
Unidentified speaker: And which gets it like this is the problem.
8:06
Unidentified speaker: Every single major hyperscaler right now has a a negative free cash flow, and that that was not the case for the last you know many years.
8:15
Unidentified speaker: And this this is the problem.
8:16
Unidentified speaker: And you know, we we spoke about this last week when we were having our more private chat, but you know, I I've been kind of ringing the alarm on on this type of a thing happening, and I think this is really just the beginning.
8:27
Unidentified speaker: So I want to ask about that.
8:29
Unidentified speaker: Do you think they made I mean it's easy to second guess you know the CFOs and CEOs of those companies, right?
8:35
Unidentified speaker: I'm not I'm not Sundar Pichai or Andy Jazzy, anyone like that.
8:39
Unidentified speaker: But still, do you think they made a mistake by putting all this enormous capex spending on their balance sheets rather than doing things like joint ventures, going to the banks, and maybe that's a good thing for the rest of the economy that they didn't involve the banks in this thing, otherwise, today financials may be down a lot more.
8:58
Unidentified speaker: So now they're sort of, I don't want to say insulated from this from you know, they were fortunate that they actually had cash flow, free cash flow.
9:07
Unidentified speaker: Most of the the uh neo clouds do not, and they completely relied on debt.
9:12
Unidentified speaker: And this is why if you look at what happened to um Corweav last week, you know, and you look at their they're essentially distressed debt now.
9:19
Unidentified speaker: They're rated at distress, which you know, and when you have 50 or 60 billion dollars of debt that you know you have a you don't have ability to repay right now, and your interest rates are going up, you know, up a wall, uh, the company's at risk, right?
9:33
Unidentified speaker: They're there, you know.
9:34
Unidentified speaker: I think you you probably saw the um the tweet from Elon about a week ago where he said uh all debt has dried up for these neo clouds, and you know, there's going to be uh guppy level death, I think was what his his uh his quoted term was too hard to describe what's coming.
9:52
Unidentified speaker: And that's that's a little scary when when there's that many companies, and these companies are all multi-billion dollar companies just a few weeks ago, you know, with enormous moats as well.
10:03
Unidentified speaker: A lot of them.
10:04
Unidentified speaker: Well, and the problem is a lot of them were also backstoped by NVIDIA.
10:08
Unidentified speaker: So this is why I think we're we're just at the beginning, because the the insulary funding, circular funding issues are are going to become exposed a lot more when we you know start to have uh you know companies not be able to pay back their debt, uh, then that that backstop has to be executed.
10:28
Unidentified speaker: And uh all of the uh you know, 77 70, 80% of NVIDIA's profits and revenues are coming from you know data centers, right?
10:36
Unidentified speaker: So if that purchase slows down, stops delivery, doesn't get happen, doesn't get taken, um, there's gonna be a chain of events that's gonna happen.
10:45
Unidentified speaker: And Broadcom just did a big deal with anthropic similar structures, like, oh, we'll backstop a little bit, then you buy our chips and then TPUs and things like that as well.
10:54
Unidentified speaker: Yeah, I I think the the the time for that type of arrangement is um is coming to a close.
11:02
Unidentified speaker: You know, the the the the the frenzy that came, a lot of it was because in the first half of this year you had um you had anthropic showing you know like 10x growth over two quarters or something with some crazy numbers.
11:16
Unidentified speaker: Uh, but if you notice in their S1, there is no data about Q3.
11:21
Unidentified speaker: We're in October, but there's no data on Q3.
11:24
Unidentified speaker: Uh why?
11:25
Unidentified speaker: Because Q3 was essentially flat.
11:28
Unidentified speaker: And and that's a very bad sign when when everybody is expecting uh that 10x to continue, and if this slows down, you know, even if it doesn't stop, even if it doesn't go flat, even if it slows down, yeah, it's going you know 2x instead of 10x.
11:44
Unidentified speaker: And that's that second derivative where it's coming falling, right?
11:48
Unidentified speaker: First quarter of 140 plus.
11:50
Unidentified speaker: I remember there was there was a podcast in like you know, maybe June or July, people say, well, it's gonna come out 10x this year, next year they're gonna be a trillion dollars in revenue, the year after they're gonna be 10 trillion dollars, and then they're gonna own the world, right?
12:03
Unidentified speaker: I mean, this and people just expect that these kind of rates are going to continue forever, and we live in a real world, a physical world that has physical constraints.
12:12
Unidentified speaker: Um, and you know, what happened in Q2 with the token maxing was a temporary blip.
12:18
Unidentified speaker: Um, I'm not saying that you know there's no demand for AI.
12:22
Unidentified speaker: I think there's an amazing demand for AI.
12:23
Unidentified speaker: I think AI is going to do great things for businesses, for consumers, but for any technology, it takes time for it to diffuse into industry, for companies to change their workflows, for you know, uh for employees to be trained up on how to take advantage of what's coming and not be afraid of losing their jobs and and sometimes you know sabotaging these these projects because they're they don't want it to be successful.
12:50
Unidentified speaker: Right.
12:50
Unidentified speaker: So there's a lot of concerns out there uh in the real world that is different than a startup you know that is going starting from scratch with with you know a small number of employees and and no existing infrastructure and processes.
13:05
Unidentified speaker: Right.
13:05
Unidentified speaker: Actually, on the topic of anthropic and debt, I just saw you know the news that Blackstone had told investors that Anthropic will probably get an investment grade credit rating after IPO.
13:18
Unidentified speaker: Now they're saying they're going to be like double B, which is close to junk.
13:23
Unidentified speaker: Yeah.
13:23
Unidentified speaker: Right?
13:24
Unidentified speaker: I mean, goes to what you were just saying, but this just came out over the last hour or so.
13:28
Unidentified speaker: Anthropic has something like five or six hundred billion dollars, billion dollars of committed uh capital.
13:33
Unidentified speaker: I think 80% of that is non-cancelable, right?
13:36
Unidentified speaker: Um, you know, and their revenues for the first half was 16 billion.
13:40
Unidentified speaker: Um, so we'll we'll see.
13:42
Unidentified speaker: And and also their 16 billion is based on transfer of actual revenues of their partners who are selling it at a at a premium, uh, not the price, not the money they're getting, but the money that their customers are getting.
13:54
Unidentified speaker: So you gotta discount that by, I don't know, whatever their arrangements are with their distributors.
14:01
Unidentified speaker: So we need to be a little bit careful in terms of how the books are actually being um reported or cooked.
14:08
Unidentified speaker: So I get to go ahead, Ryan.
14:10
Unidentified speaker: Uh you know, I was gonna say, I just want to circle back on something.
14:13
Unidentified speaker: Twice I think you said, you know, it's just the beginning.
14:15
Unidentified speaker: I want to make sure I understand for the listeners, Alvin, when you say just the beginning, are you saying like these drops is volatility?
14:22
Unidentified speaker: What exactly does that mean?
14:24
Unidentified speaker: Just to make sure everyone understands.
14:25
Unidentified speaker: Um so I I wrote a paper about two months ago called the the reckoning, the great reckoning and the reconnecting.
14:30
Unidentified speaker: And it essentially talks about um the current bubble that we have in the AI space.
14:39
Unidentified speaker: Uh, and when the market realizes the you know, the valuations are not consistent with the actual revenue potential near term.
14:49
Unidentified speaker: Again, I I'm a big believer in AI, but you know, it just like the internet was an amazing thing, but it took 10 years for it to you know come into its own weight.
14:58
Unidentified speaker: And you know, the same with the railroads were great, but it was one of the biggest busts in history, you know.
15:03
Unidentified speaker: So um, and I also want to make sure we understand there are two types of of bubbles.
15:08
Unidentified speaker: There's the soap bubble, uh, like uh you know, tulip uh minia or or NFTs.
15:14
Unidentified speaker: And when they pop, there's nothing there.
15:16
Unidentified speaker: All right, but then there are the the beer bubbles, the the you know, the the railways, the the uh internet, and now AI, where uh when it pops, there is beer, there's real substance there.
15:28
Unidentified speaker: Um but the problem with with beer is that I'm sure you guys have all poured beer.
15:32
Unidentified speaker: When you pour it quickly, it's mostly foam and a little bit of liquid.
15:37
Unidentified speaker: And when you pour it slowly and you tilt the cup and you let it you know go at it at the right pace, you could get you know 95% liquid.
15:44
Unidentified speaker: And and uh if you go too quickly, it actually flows over the cup.
15:48
Unidentified speaker: And that's what we're doing right now.
15:49
Unidentified speaker: We we are overallocating to AI uh resources uh before it's able to bring value into the economy.
15:59
Unidentified speaker: And and that's that's the real problem.
16:01
Unidentified speaker: In fact, if you look at the the the biggest uh in terms of percentage of GDP, the biggest um bubble in history was actually uh was actually the railway bubble.
16:12
Unidentified speaker: It's around 65 percent uh of the of the GDP of America was this one segment.
16:18
Unidentified speaker: Right now, um the AI sector is around 45 percent of the total uh total stock market.
16:26
Unidentified speaker: Okay, actually, I think that 65% of the stock market was railroads back back in the the 1800s, okay.
16:33
Unidentified speaker: Um now around 45, and that 45% is worth more than the entire GDP of America.
16:39
Unidentified speaker: Okay, and if you take it on a per year basis of spending, we are actually now spending more per year in terms of uh percentage of GDP than the height of the the railway bubbles.
16:52
Unidentified speaker: So even though the the the railway bubble is bigger from a percentage of of market, uh right now, in terms of how much we're we're allocating, it's uh higher than than any industry ever.
17:03
Unidentified speaker: Now that's yeah, I mean we've been talking about the amazing amount of spending, right?
17:08
Unidentified speaker: And I mean, this was in our outlooks.
17:11
Unidentified speaker: And speaking of, I've never heard anyone use that beer analogy.
17:14
Unidentified speaker: I have some thoughts about you know our outlook now and what you know, some images in mind, right?
17:23
Unidentified speaker: It automatically teaches you, you know what, going at the right speed is actually important.
17:27
Unidentified speaker: So it's not just how much foam and how much beer, it's that you have to pour it at the right speed because we have you know when you have useful things, if you pour at the right speed and it's absorbed as you're making it, you have an amazing technology that is going to uplift society.
17:43
Unidentified speaker: But right now, we're just making the market more fragile, we're gonna make the job market more fragile, and we're also making geopolitics much higher tension, right?
17:53
Unidentified speaker: All of these things are are bad for overall you know global flourishing or human flourishing.
17:59
Unidentified speaker: So let's uh I'll pull on that and something you said, basically, and this refers to something you wrote.
18:05
Unidentified speaker: You wrote that the first to scale wins, not the first to invent.
18:10
Unidentified speaker: Yes.
18:11
Unidentified speaker: I want you to explain that within the concept.
18:13
Unidentified speaker: You kind of were talking about that.
18:15
Unidentified speaker: Are American companies spending too much on pushing the frontier, getting the latest and greatest model rather than deploying existing technology?
18:25
Unidentified speaker: And I want you to contrast that with what China is doing as well.
18:29
Unidentified speaker: Sure.
18:29
Unidentified speaker: Um, I mean, if you look at every technology in history, um, the countries that get the most value from that technology is the ones that are able to deploy it in their society, right?
18:41
Unidentified speaker: And in fact, you know, you you look at like the steam engine, we didn't, the Americans didn't invented electricity, right?
18:46
Unidentified speaker: I mean, we didn't invent it, but we we deployed it, we we just put it out to market and it just made it great.
18:53
Unidentified speaker: And you know, look at China, you know, EVs, they didn't invent it, but they freaking deployed the heck out of it, you know, solar panels, they didn't invent it, it was invented here, right?
19:02
Unidentified speaker: Right and they now own 90% of the market for it.
19:06
Unidentified speaker: So whoever is able to deploy it, scale it, and then you know, create it uh at a at a pace where their economy is able to utilize it, that's where the value comes.
19:16
Unidentified speaker: And and if you look at the strategy for China, their their strategy is uh is diffusion.
19:21
Unidentified speaker: Diffusion means diffusing it to industry.
19:24
Unidentified speaker: And uh they in fact have this thing called the AI Plus Plan.
19:28
Unidentified speaker: And it says by 2027, they want 70% of their companies to be using AI actively, and by 2030, they want 90% of their companies to be uh utilizing, and that's a national target, right?
19:41
Unidentified speaker: So essentially, at national level, province level, state level, all their governments trying to get people to use it.
19:47
Unidentified speaker: What's what's the the the the national AI action plan in America?
19:52
Unidentified speaker: Be the frontier, that's it.
19:55
Unidentified speaker: And spend it seems like here we are spending whatever it takes to get to the frontier.
20:00
Unidentified speaker: You know, China's spending one-tenth as much on data centers as we are.
20:03
Unidentified speaker: They are two percent two percent below us in terms of of all the benchmarks.
20:09
Unidentified speaker: So to there's they're spending less than a hundred billion dollars this year, we're spending over a trillion dollars this year.
20:16
Unidentified speaker: I'm I'm all on that.
20:17
Unidentified speaker: I'm curious what you're saying because they people talk about AI and the different degrees of like let's call intelligence, right?
20:24
Unidentified speaker: And then sentience and things like that.
20:27
Unidentified speaker: It seems like the American companies are like racing to figure out we need to get to you know, human type intelligence, and that's the race, rather than to your point, you know, let's use what we have and deploy it.
20:43
Unidentified speaker: Right.
20:44
Unidentified speaker: And it's it seems from what you're saying, China's not playing that game.
20:47
Unidentified speaker: Well, I mean, I I I think there's a couple of labs that want to their target is to go for AGI, but the government's saying, hey, while you're going for AGI, it's fine, but don't waste too much money and make sure whatever you make is available for everyone to use.
21:01
Unidentified speaker: And so this is kind of why they're also going after the open source model, because if it's if it's cheap, it's free, everybody's gonna use it.
21:08
Unidentified speaker: If I'm gonna charge you $50 per million tokens, then we have what we have in America, which is you know, probably 10 to 15 percent of companies are actively deploying it, and maybe you know, a few single, low single digital actually successfully deploying it to create ROI, right?
21:23
Unidentified speaker: Very few companies right now are coming out and saying, you know, I've made a bunch of money from it.
21:28
Unidentified speaker: Some startups have absolutely, but large companies, uh, it takes time.
21:32
Unidentified speaker: It takes time for for that to flow through.
21:34
Unidentified speaker: And they have to completely, you know, essentially redesign, re-engineer their workflow to make it AI ready before they can get value from it and retrain their workforce.
21:49
Unidentified speaker: Want to hear more from the Facts versus Feelings team?
21:51
Unidentified speaker: Sign up for the Carson Investor Research Newsletter.
21:54
Unidentified speaker: You'll get up-to-date market news, analysis, and insights from us and the rest of the team to help you serve as a better guide for your clients.
22:02
Unidentified speaker: Subscribe today at Carson Group.com slash investment newsletter.
22:11
Unidentified speaker: So interesting.
22:11
Unidentified speaker: And you've made actually what I think is a pretty contrarian argument.
22:17
Unidentified speaker: And you know, one thing we like for contrarian takes.
22:21
Unidentified speaker: I mean, just being in markets, we're always looking for something that's not price, right?
22:25
Unidentified speaker: By nature, it's I'm not saying we should be the sub contrarian for the sake of being contrarian, but I thought your contrarian argument is interesting because you're saying this AI race will actually bring the US and China closer together, where everything literally for the last decade, I feel like we've been going apart, right?
22:44
Unidentified speaker: Ever since the trans-Pacific partnership sort of fell apart.
22:48
Unidentified speaker: I mean, it seems like America's just separating from Asia, and not just China, for all of Asia, probably.
22:54
Unidentified speaker: So I want you also to talk about that.
22:56
Unidentified speaker: And you explained something called the stag hunt, you know, that framework, well, you know, which comes out of game theory, too.
23:03
Unidentified speaker: Yeah, yeah.
23:04
Unidentified speaker: So stag hunt is actually it's a very old game by Jean-Jacques Rousseau, who uh, you know, every when people think of game theory, everybody thinks about uh prisoners' dilemma, and and we share it a lot.
23:13
Unidentified speaker: Oh, you know, we can't trust them, it's prisoner, and the right thing to do in prisoner's middle is to defect and just to screw your your your your partner, your opponent, right?
23:21
Unidentified speaker: Right, um and that's like a zero-sum game.
23:24
Unidentified speaker: That is a zero sum game, and that conclusion of screw your partner only works if it's a single turn game.
23:30
Unidentified speaker: Okay, so we don't live in a single-turn game, we don't live in a zero-sum game, and and we also don't live in a single winner game, right?
23:39
Unidentified speaker: Um, and we don't have a clear finish line that it's done.
23:42
Unidentified speaker: This is this is the issue when people start to apply a theory to something that doesn't represent the playing field that you're you're working in, you know.
23:51
Unidentified speaker: And and the other thing about prisoners' dilemma, everybody, and you know, the key the key thesis is you can't talk to each other.
23:59
Unidentified speaker: You know what?
24:00
Unidentified speaker: We we have people that talk to each other every freaking day.
24:04
Unidentified speaker: Not only you know, between the last, but also between countries, between the governments in this, you know, side in your track two dialogues.
24:10
Unidentified speaker: Every day they're talking.
24:11
Unidentified speaker: And not only are you talking, you can see the actions of each country every day and each company.
24:16
Unidentified speaker: You know, how fast are they releasing product, you know, which which which new export controls that they put in, you know, what what what uh you know escapes have they had?
24:24
Unidentified speaker: You can see all of that, right?
24:26
Unidentified speaker: The the world is actually fairly transparent, which does not represent the the prisoner's dilemma.
24:32
Unidentified speaker: When much more accurate picture is actually uh this concept of stack kind of stack and the the story is uh two hunters go into a forest, they decide do I hunt for the big stag, the thing that's gonna feed my family for a month, or do I hunt for the rabbits and you know and feed my family for a day or two?
24:50
Unidentified speaker: Okay.
24:51
Unidentified speaker: And if I go for the stag, I need to go together with my partner who I'm going here with.
24:56
Unidentified speaker: If I'm hunting for the rabbits, I just everybody do their own thing, right?
24:59
Unidentified speaker: So you you you just essentially you compete for whatever you want.
25:02
Unidentified speaker: The worst thing to do is for um one of the the uh hunters to go for the stag by himself.
25:11
Unidentified speaker: First of all, he won't be able to get it, and even if he gets it, he can't take it home.
25:14
Unidentified speaker: Okay.
25:14
Unidentified speaker: Two, is that uh, you know, then he's wasted his whole time and his family stars, okay.
25:20
Unidentified speaker: And the other guy just gets for more rabbits.
25:22
Unidentified speaker: Okay.
25:23
Unidentified speaker: So this is what we're doing today is that we are going for this giant AGI stag, which we don't know how to control, which we have to spend 10 times as much to get, and we're not getting any value out of it yet, you know, and we're rushing to this thing that could potentially create a major, you know, um financial crisis, job crisis in America.
25:44
Unidentified speaker: At the same time, the Chinese are like, shit, I'm just gonna make open source, it's you know, 98% is good.
25:49
Unidentified speaker: I spent one tenth as much money.
25:52
Unidentified speaker: Um, and you know, everybody's using my things, you know, uh, it's not now I'm the most downloaded uh model in the world.
25:57
Unidentified speaker: Every country in the world is using mine to create their sovereign AI models.
26:01
Unidentified speaker: So freaking, you know, did America win?
26:04
Unidentified speaker: Are we winning?
26:05
Unidentified speaker: We're we're winning the benchmark, and then tomorrow some other uh lab will win that benchmark by one percent.
26:11
Unidentified speaker: And then the the day after that, and you know, but they didn't get any money from it, right?
26:16
Unidentified speaker: They didn't create any value from it.
26:17
Unidentified speaker: They got they got you know uh a star on the board for that day.
26:22
Unidentified speaker: That's it.
26:23
Unidentified speaker: The equivalent of the fields medal in mathematics for solving problems that haven't been solved before.
26:29
Unidentified speaker: I'm not saying that's not a good thing.
26:30
Unidentified speaker: I mean, I I fields medal is it's it's you know, you you don't not very many people get it.
26:34
Unidentified speaker: This no, absolutely, yeah.
26:36
Unidentified speaker: This thing, everybody every day gets a different one.
26:39
Unidentified speaker: So it's it's like you know, it's like getting getting the star at the at the kindergarten, right?
26:43
Unidentified speaker: This is that's why I say star.
26:44
Unidentified speaker: When you're at a kindergarten, you know, you you okay, I I put my shirt on, right?
26:48
Unidentified speaker: Oh, here's a star.
26:49
Unidentified speaker: Oh, you came on time.
26:50
Unidentified speaker: Here's a star.
26:51
Unidentified speaker: Okay, this this is what we're we're we're right now fighting for that little star.
26:55
Unidentified speaker: And you know, the the the Chinese are like, shit, I you know, I don't care about stars, I just want my economy to be growing five or ten or twenty percent.
27:02
Unidentified speaker: And you know, we're we're we're creating a a market in our GDP right now where pretty much all the growth in the last year or two data center uh investments.
27:14
Unidentified speaker: And we're we're stopping the we're not building offices, we're not building homes, we're not building electric plants, you know, we're and we're not building trains, you know, we're we're building one thing.
27:24
Unidentified speaker: You you essentially we're all in on one sector that has to pay off right away.
27:30
Unidentified speaker: And if it doesn't, we're we're gonna be in trouble.
27:34
Unidentified speaker: So, Alvin, you mentioned stars.
27:36
Unidentified speaker: You know, one of the stars of this market has been Alphabet.
27:39
Unidentified speaker: Um, so I know you might have some takes too.
27:41
Unidentified speaker: I'm gonna read some things here, and I just want everyone's take on this.
27:43
Unidentified speaker: So we mentioned big down days, literally down 12% yesterday.
27:47
Unidentified speaker: The worst day in the history of that stock since 2004 is IPO, two quarters in a row of negative cash flow, down approximately $500 billion market cap on only one day.
27:57
Unidentified speaker: Google, here's the here's the big question Google services, which is really YouTube and Google search, missed on revenue 39.2 billion versus over about 41 billion profit.
28:07
Unidentified speaker: And they said, Yeah, it turns out that some of the AI stuff, AI search is kind of almost can't our AI search is cannibalizing our revenue on the other side.
28:14
Unidentified speaker: Yeah, whoops.
28:16
Unidentified speaker: Um, that is I don't even know where to take this, but clearly the stock was hammered yesterday on that and some other concerns.
28:21
Unidentified speaker: But that one got me.
28:22
Unidentified speaker: Do you see that continuing, Alvin?
28:24
Unidentified speaker: That literally they created something that's hurting their revenue.
28:26
Unidentified speaker: Yeah, uh absolutely.
28:28
Unidentified speaker: I think it's only the beginning.
28:29
Unidentified speaker: I mean, I don't know how many man, I don't know how many uh you know how much uh Google searches you're doing this year versus last year.
28:37
Unidentified speaker: I can say I'm down.
28:38
Unidentified speaker: Oh I'm I'm almost completely out.
28:40
Unidentified speaker: First of all, that Google search is horrible because it was a whole bunch of ads.
28:44
Unidentified speaker: Yeah, yeah.
28:45
Unidentified speaker: Well, I mean that that this is the business model of Google and of all search companies, is that they're selling you ads, right?
28:52
Unidentified speaker: And then in fact, here's here's something that I that you didn't talk about, which is uh kind of big on Twitter in the last few days is uh all the companies that were buying ads said, you know what?
29:02
Unidentified speaker: I used to have um very specific ads, uh words I was buying and I was only gonna pay for.
29:09
Unidentified speaker: Now, automatically in the background, because they're in such dire straits to generate revenue, they've expanded my words to related to your words that are not really what I asked for, but I'm now paying this high price for a very specific word when they expand it to 10 or 50 times uh the the the the audience of the things that I weren't paying for.
29:31
Unidentified speaker: And so their costs are going up, but they're they're actually um you know results that they're getting in ROI is actually coming down.
29:38
Unidentified speaker: And so if that's the case, then people are going to continue to find alternative ways to reach customers.
29:43
Unidentified speaker: You know, this is a to me, this is a sign of desperation.
29:46
Unidentified speaker: Um they also changed the algorithm.
29:47
Unidentified speaker: They used to have a essentially the second best price was a price that they would charge their customer.
29:53
Unidentified speaker: Uh so if two people were pacing a bit uh were bidding on the same word, now what they're finding is that they're using the top price, is what they charge for the winner.
30:01
Unidentified speaker: So they they they've actually secretly in the back changed how they price uh ads to their customers because they need to make up revenues, right?
30:09
Unidentified speaker: That I mean that this this is why they didn't release uh a chat GPT equivalent when they had it a year before OpenIA had it because they they could see this happening, right?
30:26
Unidentified speaker: Yeah, I mean it's it's you can tell when somebody can give you the answer, or somebody's gonna give you a hundred links and you gotta find the answer.
30:33
Unidentified speaker: Which one do you choose?
30:34
Unidentified speaker: It it's it's kind of automatic.
30:36
Unidentified speaker: I'm surprised it took this long for it to show, but I think that that that trend is not going back the other way.
30:44
Unidentified speaker: People are not going back to to search.
30:46
Unidentified speaker: I mean, once in a while, I think you know, I probably I use it once a day now instead of I use it 10 times a day in the that's probably just muscle memory as well, right?
30:54
Unidentified speaker: Like you had been doing it, it's hard to change something, but slowly, but surely, yeah, yeah.
30:59
Unidentified speaker: And like my search takes place in AI bots now.
31:02
Unidentified speaker: Yeah, now the AI agents might be using the search, but they're not looking at the ads and they're not clicking on the ads, and they're not buying anything afterwards.
31:10
Unidentified speaker: So they're essentially giving money to do the traffic serving to these agents who are not generating revenue, so it's actually a negative revenue generation for them.
31:19
Unidentified speaker: Which begs the question do you think ads are coming to the chatbots?
31:23
Unidentified speaker: Because at the end of the day, a big source of revenue for all tech is ads, right?
31:27
Unidentified speaker: Yeah, you said it.
31:28
Unidentified speaker: Yeah, uh on the enterprise level, different.
31:30
Unidentified speaker: Okay, set that aside.
31:32
Unidentified speaker: And in Google, we can talk about YouTube.
31:34
Unidentified speaker: Maybe it should be it'll just be a YouTube company and uh turn them into that.
31:39
Unidentified speaker: But do you think ads are coming into the chat bots?
31:42
Unidentified speaker: I mean, I kind of fear that at some point, but I mean we we we we already talked about well, open AI already talked about this maybe four or five months ago.
31:50
Unidentified speaker: They said they were experimenting on ads on their low tier, low tier, I think like eight dollar, you know, or or free free versions, which I guess you know if it's free, then sure, I I get it.
32:00
Unidentified speaker: But if you're paying well, even if it's eight dollars, if you're paying for it, I don't think you should put ads because that completely um conflicts with the the trust you would have in that service.
32:12
Unidentified speaker: If I can't trust whether this answer is trying to sell me something or this chance, this this answer is trying to give me the truth.
32:18
Unidentified speaker: Uh I'm gonna I'm gonna use the one that doesn't give me ads.
32:22
Unidentified speaker: Like if I'm looking for the cheapest flight, I trust that it will give me the cheapest flight rather than you know, American paid them to show me their cheapest flight, and that's it.
32:33
Unidentified speaker: Yeah, so so this is this is the issue with chatbots going forward or agents in general, is that it will be the best way for any corporation to manipulate you right to buy whatever you want because they did they just bring it to you, it's there, all you gotta do is say okay, and you've bought it.
32:49
Unidentified speaker: In fact, you might have given it your credit card and just say buy me the cheapest one.
32:52
Unidentified speaker: Yeah, there is no okay, then right now, yeah, right.
32:56
Unidentified speaker: So, so if if you trust it to that level and it's not working for you, you are screwed, right?
33:03
Unidentified speaker: And as a consumer.
33:05
Unidentified speaker: Um, let me jump in.
33:06
Unidentified speaker: We're almost we're over halfway.
33:08
Unidentified speaker: It's amazing how fast this goes.
33:10
Unidentified speaker: The latest facts versus feeling social hour.
33:13
Unidentified speaker: Um, we're having a lot of fun here.
33:15
Unidentified speaker: Um, with Alvin Grayland here.
33:17
Unidentified speaker: Um, where do we want to take this?
33:20
Unidentified speaker: How let's go this origin stories, origin stories.
33:23
Unidentified speaker: How did you meet Mr.
33:24
Unidentified speaker: Omani Carson?
33:26
Unidentified speaker: Because Omani connected you with us, and I got a guy, I got a guy who's amazing, and obviously that connected this, and that's kind of why we're doing this.
33:34
Unidentified speaker: And by the way, you met with our investment committee last week.
33:37
Unidentified speaker: I I couldn't make it, I was traveling, but I heard it was an amazing discussion.
33:40
Unidentified speaker: So thank you for taking time to meet with uh our team twice here in about a week or so.
33:44
Unidentified speaker: But how'd you meet Mr.
33:45
Unidentified speaker: Carson?
33:46
Unidentified speaker: Um, so my brother is part of a group that he's organized, and uh so I I went to one of their events, and I think uh when I went there, uh uh he heard I was a writer and he said, Oh, well, what what do you write?
33:58
Unidentified speaker: And I sent him something, and and then he's like, Whoa, this stuff is really resonates with me.
34:02
Unidentified speaker: And then I think the next week he saw I was on uh Peter Diamandis's show, and uh and I guess he's a big fan of of um of the moon moonshots.
34:11
Unidentified speaker: Moonshots moonshots, moonshots, yeah.
34:13
Unidentified speaker: So I think I think that that kind of prompted him to say, Hey, I I want I want uh you to come to our show as well.
34:19
Unidentified speaker: So I think that's how it all kind of came out.
34:21
Unidentified speaker: I know we're we're we're thrilled, and I know um you know we all have to stop a hard stop top of the hour.
34:27
Unidentified speaker: So uh Sony, where else we want to take this?
34:29
Unidentified speaker: Uh I I was gonna ask about the same US China since we brought it up before, but continuing along that, do you think chip controls from the US side are working, however, we define working?
34:45
Unidentified speaker: Um, so so from from depends on how you measure working.
34:48
Unidentified speaker: Um, if you if you say is it is it making life difficult for the Chinese labs?
34:54
Unidentified speaker: Absolutely.
34:55
Unidentified speaker: I I talked to all those labs, and and one of the biggest things they complain about is we don't have enough chips.
35:00
Unidentified speaker: Okay.
35:00
Unidentified speaker: Is it slowing them from catching up?
35:03
Unidentified speaker: Not really.
35:04
Unidentified speaker: Is it is it uh you know creating more revenues for American companies?
35:09
Unidentified speaker: Absolutely not.
35:11
Unidentified speaker: Is it is it creating a uh potential future competitor uh that will be getting stronger and stronger over time that then competes with us on the chip market and on the AI market and the AI stack market?
35:25
Unidentified speaker: Absolutely.
35:26
Unidentified speaker: So um, you know, again, this is this is like the whole back to the whole uh winning the little star thing.
35:32
Unidentified speaker: So we're we're winning the the oh I've got a a rock in my shoe, a pebble in my shoe pain uh race, we're in the long run going to you know essentially lose the uh everybody in the world is using American Stack Race, right?
35:46
Unidentified speaker: That that's that's what's happening right now.
35:49
Unidentified speaker: Uh and you know the companies who are making chips in China, their chips are kind of crap.
35:54
Unidentified speaker: Okay, but because they the the Chinese labs can't buy any other chips, because the Chinese data centers cannot buy any chips, they have to buy crappy Chinese you know GPUs.
36:07
Unidentified speaker: And so these companies that would have gone out of business if we let them buy NVIDIA chips and there would be no competitors there.
36:14
Unidentified speaker: No, we've essentially created competitors that are now very well funded, you know, uh partially because they got customers, partially because the government said, shit, we cannot be relying on American chips because they can take it away anytime.
36:27
Unidentified speaker: Um, so they're putting you know hundreds of millions, and in some cases billions of dollars to try to prop up these companies who are now actually making semi-competitive products uh from a hardware perspective.
36:39
Unidentified speaker: And now they're even making you know uh high bandwidth memory.
36:42
Unidentified speaker: The the CM CXMT, I think uh uh uh they went public this this year, I think a few months ago, and on first day, they became the most valuable company in China on day one.
36:57
Unidentified speaker: So uh, and I don't know if you know, but for for the Koreans, the biggest uh benefactor of uh of the Korean stock market was around the the memory companies, right?
37:08
Unidentified speaker: So uh SK Heinex and and uh Samsung and those guys.
37:12
Unidentified speaker: And you know, that's where you know, in fact, here's a funny thing.
37:16
Unidentified speaker: I don't know if you know, but uh US has 45 or 47 of a total global AI-related profits, and all of it is hardware, none of it is models, right?
37:26
Unidentified speaker: Models all lose money around the world.
37:28
Unidentified speaker: All hardware was in America, 35% was in Korea.
37:33
Unidentified speaker: Wow, 7% was in Taiwan, and then like 5% was in China, right?
37:38
Unidentified speaker: So these essentially it's only hardware that's making money right now, and Korea, a country with like you know, not very many people, like 40-50 million people, is making almost as much total profits as America.
37:52
Unidentified speaker: So I I mean we see that in the trade surplus, right?
37:56
Unidentified speaker: The annualized trade surplus, just for everyone listening, exports minus imports, Korea always had a surplus.
38:04
Unidentified speaker: It's like six times what it was just a year ago.
38:08
Unidentified speaker: Yeah, yeah.
38:08
Unidentified speaker: So and it's it's all on memory, and memory have gone up, I think, 300%, maybe more this year, in terms of how it how much it costs to buy high bandwidth memory.
38:18
Unidentified speaker: So um, you know, so but but you you I think what what we'll also find is that very soon these corrections will all uh come back to normal.
38:26
Unidentified speaker: Uh as we all know, I mean, semiconductors, I you know, I like I said, I spent a lot of years in semiconductors, it's a very cyclical business.
38:34
Unidentified speaker: Uh and we are it seems like we've forgotten that.
38:37
Unidentified speaker: And I'll, you know, and who knows when it becomes cyclical, but the way people talk about AGI and all that, it's it seems like have we forgotten that semis are the classic cyclical business?
38:48
Unidentified speaker: I mean, Ryan's, you know, the technical guys have probably known this longer than anyone else.
38:55
Unidentified speaker: Well, the problem, you know, the problem is that the people who are pumping this up are you know 30-something CEOs that have lived a very sheltered life in Silver Valley, have only gone through one cycle, um, and haven't made anything that was physical.
39:11
Unidentified speaker: Uh, and if you're making software and you look at if you use software as an evidence of where the world's going, I can understand where they come from.
39:19
Unidentified speaker: Because you know, uh a year ago, you know, five or 10% of the code was written by AI.
39:24
Unidentified speaker: Now it's 95% of the code is written by AI.
39:27
Unidentified speaker: And so if you think about it that way, hey, next year it's gonna be 100%, and you know, we our revenue has gone up 10x in the last year, uh, it's gonna go up forever, and uh, you know, our our user traffic has gone up, you know, 10 or 20x this year.
39:40
Unidentified speaker: So that's gonna continue forever.
39:41
Unidentified speaker: And so we're gonna need uh 10 or 20x more chips, we're gonna need 10 to 20x more memory, we're gonna need, you know, whatever, right?
39:47
Unidentified speaker: And our our revenue is gonna be bigger than the GDP of America.
39:50
Unidentified speaker: And our TAM is like 40 trillion dollars.
39:54
Unidentified speaker: Well, yeah.
39:54
Unidentified speaker: So I mean this is the this is the issue with all of these companies, is that um they they live in a little sheltered uh echo chamber bubble of Silicon Valley.
40:05
Unidentified speaker: Um, and everybody there talks the same way, and when they talk the same way, uh, particularly in the software and AI circles, um you you don't realize that there are alternative uh perspectives.
40:18
Unidentified speaker: So let me jump in there because just two days ago, feels like a lifetime ago, big company called Oracle had their investor day.
40:25
Unidentified speaker: They said 35% compound annual rate on revenue growth through 2030.
40:30
Unidentified speaker: And then we mentioned you just said SK Heinrich, which remember it reminded me that stock was up 13%.
40:34
Unidentified speaker: I think it was the ADR was up 13%, but it was up a lot.
40:37
Unidentified speaker: They said demand is going to outstrip supply until 2030 again, record revenue, record profits.
40:42
Unidentified speaker: Uh, Microsoft had some pretty good news.
40:44
Unidentified speaker: Um, all in all, their stock was positive, but a lot of these other clearly, like we started this conversation, if your cash flow is negative and you miss by a little bit, you're getting absolutely annihilated.
40:54
Unidentified speaker: Maybe Sony, more for you.
40:56
Unidentified speaker: How do we think about this?
40:56
Unidentified speaker: So, Carson Group, we managed close to nine billion dollars for our financial advisors and their clients.
41:00
Unidentified speaker: We haven't overweight tech.
41:02
Unidentified speaker: Tech is still up on the year, still outperforming on the year, but my oh my, October has been an absolute just bludgeoning for tech and communication services.
41:09
Unidentified speaker: How are you thinking that?
41:10
Unidentified speaker: Because we could give a bullish argument, we can give a bearish argument just in the last three days.
41:14
Unidentified speaker: All the crazy stuff that's happened.
41:15
Unidentified speaker: What do you think, Sonu?
41:16
Unidentified speaker: I think I wrote a piece a couple of weeks.
41:18
Unidentified speaker: I want to give you one quick data point to think about.
41:20
Unidentified speaker: Yeah, yeah, yeah.
41:21
Unidentified speaker: Um, this in the last few months in China, their star market, which is like their Nasdaq, has gone down 40%.
41:30
Unidentified speaker: Wow.
41:30
Unidentified speaker: 40%.
41:31
Unidentified speaker: Okay, so so their their AI bubble has deflated.
41:36
Unidentified speaker: We have not.
41:37
Unidentified speaker: You know, we we from the peak, we're down like 10%, right?
41:40
Unidentified speaker: Which is which is like par for the course.
41:42
Unidentified speaker: Like Ryan's shown numbers, like you should expect that all pretty much every year, or sometimes even twice a year, right?
41:48
Unidentified speaker: That's uh so so what that this is why I said in the beginning, we're just starting.
41:55
Unidentified speaker: And that gets to like Ryan, your question, you know, we've written about this in a full-year outlook, mid-year outlook, there's something big happening.
42:05
Unidentified speaker: And usually when you're recycling capex into earnings, I mean that's driving the market, right?
42:12
Unidentified speaker: Multiples have gone down.
42:14
Unidentified speaker: If you just look at that, you think, oh, the stock market's struggling, but then what's held it up?
42:18
Unidentified speaker: It's earnings growth, right?
42:19
Unidentified speaker: Now, where are earnings, where's earnings growth coming from?
42:21
Unidentified speaker: It's re CapEx being recycled, right?
42:24
Unidentified speaker: One company's spending is another company's revenue and source of profits.
42:28
Unidentified speaker: As long as that flywheel country continues, that's okay.
42:32
Unidentified speaker: But that's also why B wrote you want to avoid a wipeout.
42:37
Unidentified speaker: You want to be diversified, you want to combine things like momentum with, dare I say, one of the most unloved parts of the market, which is low volatility stocks.
42:47
Unidentified speaker: Yep.
42:48
Unidentified speaker: No, everyone hates that.
42:49
Unidentified speaker: You look at, you know, our friend Todd Sohn has some great charts on this showing you know cumulative flows into low volatility stocks.
42:57
Unidentified speaker: If that line just goes down, nobody wants it.
43:00
Unidentified speaker: And what's interesting, historically, low volatil stocks have a beta, which is you know, your exposure to the overall market, about 60, 70 percent thereabouts.
43:10
Unidentified speaker: Right now, it's almost zero.
43:13
Unidentified speaker: It's it's the correlation is negative.
43:15
Unidentified speaker: What's amazing, the correlation of a big part of the market, you know, certain sectors, sub-industries, to the overall market is negative.
43:23
Unidentified speaker: And that gets to you know the concentration issue, right?
43:28
Unidentified speaker: About 38, 40 percent of the SP 500 is tech stocks.
43:32
Unidentified speaker: You add meta or alphabet, which are in comm services, Amazon, Tesla, which are in consumer discretionary, that's 51% of the index.
43:41
Unidentified speaker: You have a concentration problem.
43:43
Unidentified speaker: But going to Ryan's point, what do what have we done about it?
43:47
Unidentified speaker: Well, you want to diversify.
43:48
Unidentified speaker: And not just low wall, I would argue even Europe, for example.
43:52
Unidentified speaker: Europe's doing just fine.
43:53
Unidentified speaker: You know, nobody's talking about Europe being an AI leader or anything like that, but we think that's a diversifier, right?
44:00
Unidentified speaker: So just to give you know everyone listening some context on how we're thinking about all of this and managing money at the same time.
44:06
Unidentified speaker: And I want to jump in, then I'll want to go to you.
44:08
Unidentified speaker: So just again, context around this.
44:09
Unidentified speaker: And I'm not gonna mention the specific ETFs, but there are a bunch of them you can find.
44:13
Unidentified speaker: I'm looking at a very large momentum ETF, down 11% for the month.
44:18
Unidentified speaker: Now you ready for this?
44:19
Unidentified speaker: The low, the biggest, biggest low vol ETF, up six percent for the month.
44:24
Unidentified speaker: I mean, you talk about just incredible what we've seen this month.
44:27
Unidentified speaker: I mean, it is uh an October massacre, and yet if you're a value investor, you're a low vol investor, you're a healthcare investor.
44:32
Unidentified speaker: That's where I wanted to go with this, Alvin.
44:34
Unidentified speaker: Literally, the time we're doing this XLV, not a recommendation.
44:36
Unidentified speaker: The large healthcare ETFs hitting the highest level ever.
44:39
Unidentified speaker: Healthcare is just ripping.
44:40
Unidentified speaker: Uh, you know, a lot of smart people, that's what your take on this think healthcare, I think it's more the anti-AI trade.
44:45
Unidentified speaker: At the same time, maybe healthcare is gonna benefit from AI.
44:47
Unidentified speaker: How do you think about investing with healthcare as a diversifier away from AI?
44:51
Unidentified speaker: Because it sure seems to be working as these AI stocks are going lower.
44:54
Unidentified speaker: Yeah, by the way, healthcare is up 10% for the month, thereabouts.
44:58
Unidentified speaker: And I actually think that healthcare will continue to go up.
45:01
Unidentified speaker: You know why?
45:02
Unidentified speaker: Because the cost of drug discovery is gonna come down, and they have something that the AI companies don't have, which is uh a, you know, the data sets of all their users and customers and patients and different drug combinations and molecule combinations, uh, and the pathway to clinical trials and all of these things that they've had, in fact, all of their prior data of prior clinical trials, they can now have AI look through it and see if there's things that they missed.
45:32
Unidentified speaker: Right.
45:33
Unidentified speaker: And so this, and and they can they don't have to have the biggest frontier models to do it.
45:39
Unidentified speaker: This this is the lie that the frontier AI companies tell everyone is we can't slow down because if we slow down, then we're not gonna come up with the cancer cures and people are gonna die because we went too slow.
45:53
Unidentified speaker: And it is a complete lie because not none of the the big drug companies are using frontier models to make their next drugs, they are all using specialized small models trained on specialized data, and and no smart AI uh uh drug company would give their entire data portfolio to these AI companies because we can see what happened.
46:16
Unidentified speaker: They're gonna go make their own thing and then they're gonna screw you, right?
46:20
Unidentified speaker: So there nobody's gonna give that away.
46:22
Unidentified speaker: So um, you know, when it normally it takes billions of dollars right now to make a big drug, and I think that that number is gonna come down dramatically, but they can still sell it for high numbers because there actually is a patent you know protection system in in medicine that is not the case in open source models.
46:41
Unidentified speaker: Um when when night when somebody creates a closed source model that costs them five billion dollars to do a training run, and then the next day somebody releases an open source model that is one-fifth the price, they've completely destroyed the value that you've had in that five billion dollars of investment.
47:00
Unidentified speaker: Right.
47:00
Unidentified speaker: You could put five billion dollars into drug discovery and you might get you know ten or fifteen billion dollars back over the next five years.
47:09
Unidentified speaker: Yeah, that's incredible.
47:10
Unidentified speaker: I'll just say this.
47:11
Unidentified speaker: I'm just trying to jump in a second.
47:12
Unidentified speaker: We have close to 3,000 people watching us at this very second.
47:16
Unidentified speaker: We've had many thousands over the past 45 minutes.
47:18
Unidentified speaker: So thanks to everyone who joined us on this latest social hour.
47:21
Unidentified speaker: Obviously, thanks, Alvin, for joining us.
47:23
Unidentified speaker: We have less than 15 minutes, Sonu.
47:25
Unidentified speaker: So, um, what else we want to talk about?
47:26
Unidentified speaker: Yeah, uh Alvin, maybe you can mention how people can find all their stuff and where they can, you know, feel free to name your books and where they can go to read everything you write.
47:37
Unidentified speaker: So I and I've read a big job of it, it's fantastic.
47:40
Unidentified speaker: Talks about kind of where things are going next few years.
47:42
Unidentified speaker: Um, I I would I guess that probably the place to go is on X, uh just a Grayland on X, uh, or the Abundanist uh on uh Substack is probably a good place to find me.
47:53
Unidentified speaker: And I I usually publish uh papers and articles and essays on there.
47:57
Unidentified speaker: Um yeah, so hopefully that'll give you a uh a few places.
48:02
Unidentified speaker: How many podcasts have you done?
48:04
Unidentified speaker: By the way, that Moonshots podcast was like three hours long or something.
48:08
Unidentified speaker: It was like two and a half hours, I think.
48:10
Unidentified speaker: Two and a half hours.
48:12
Unidentified speaker: Um I think this this week I did three podcasts.
48:14
Unidentified speaker: You're my third one this week.
48:16
Unidentified speaker: Wow, so thanks.
48:17
Unidentified speaker: I mean, it's it's a hot topic.
48:18
Unidentified speaker: I think everybody is a little concerned about what's happening with all the issues, and you know, one thing we didn't talk about is all of the uh cyber issues.
48:26
Unidentified speaker: You know, we just had I wanted to talk about the Bitcoin, the the the the six billion dollars of bitcoin.
48:31
Unidentified speaker: I wanted you to take there as well.
48:32
Unidentified speaker: Go ahead and build on this.
48:33
Unidentified speaker: In fact, uh a CZ is an old friend of mine.
48:36
Unidentified speaker: We used to play poker together when we were both in Shanghai when before he became rich and famous.
48:41
Unidentified speaker: We he when he was just an engineer, we used to play.
48:45
Unidentified speaker: You got some skeletons in the closet on him, I'm assuring drinking, hanging out and playing poker.
48:49
Unidentified speaker: I'm sure you can write a book on that, but but go ahead, yeah.
48:52
Unidentified speaker: Yeah, yeah.
48:53
Unidentified speaker: What so what's your question, Ryan?
48:55
Unidentified speaker: Well, I guess what's your take on the Bitcoin that was stolen?
48:57
Unidentified speaker: I mean, uh you know should we have expected that?
49:00
Unidentified speaker: Like, you know, when you take decentralized finance, and it's almost like, and I want to get to regulation and all of that.
49:07
Unidentified speaker: Part of it is like it's a free-for-all, right?
49:09
Unidentified speaker: I mean, I mean, very simply put, and a free-for-all means you know, anything can happen, and now it looks like and it's actually in some ways uh you're the the people who got affected were lucky that they were on Binance because I think Binance had agreed to make them whole, right?
49:24
Unidentified speaker: Which which is a very generous thing because that's you know six billion dollars or something.
49:30
Unidentified speaker: Um but if you were if you just had something in the in your on your computer that was a you know key to something, and right now every computer is vulnerable to being hacked.
49:43
Unidentified speaker: Uh, I mean the fact that 70% of the websites last week were essentially made available and open to you know the hackers to go and and essentially fully own with an external connection, that you know, that was what the PHP um bug essentially exposed.
49:58
Unidentified speaker: Uh Larryman.
49:59
Unidentified speaker: Luckily, they also put out some patches, and you know, it depends on how quickly people patch things up.
50:03
Unidentified speaker: But the reality is that you know, I used to work for a cybersecurity company, and the average patch time for it to be fully deployed is around two to three months.
50:10
Unidentified speaker: So you put out a patch, and it takes months for to get all these companies getting in all their systems and all the the end user consumers saying yes, right.
50:19
Unidentified speaker: And when that happens, every every moment that that's not patched up is an open door for bad guys to do something.
50:27
Unidentified speaker: And you know, we all you know, no matter how secure you are, you think your computer is pretty secure, you probably have a file that says, you know, a Bitcoin key or whatever you want to call it.
50:36
Unidentified speaker: You might have a secret name for it.
50:37
Unidentified speaker: But you know, somebody can just go and scan your drive.
50:40
Unidentified speaker: Uh they if they have you know, the cost of these small agents right now is essentially free.
50:45
Unidentified speaker: You can run them on you know uh uh a uh an edge device and just say scan everything on the on the internet, look for holes, go in there, steal whatever data you have, and let's figure it out later.
50:55
Unidentified speaker: So I'm actually surprised that there wasn't this kind of um essentially crypto losses before, or maybe some people have had it, they just haven't announced it.
51:06
Unidentified speaker: Yeah, we don't know yet, right?
51:07
Unidentified speaker: Yeah, so but but I I think what we will find is that there this issue uh will continue to expand.
51:14
Unidentified speaker: If you look at the the CVEs that the common uh vulnerabilities and exposures list, uh it has gone up, I think 700 uh no 300 for uh severe and 700 this year for critical bugs.
51:30
Unidentified speaker: So there are thousands and thousands of open vulnerabilities right now that if a bad guy wanted to and they were smart and they go and go just, I mean, it's all public.
51:40
Unidentified speaker: Those those vulnerability lists are public.
51:42
Unidentified speaker: If you want to use that, I mean it's it's meant for the the good guys to go patch.
51:46
Unidentified speaker: But if the bad guys went there and said, Where what are all the vulnerabilities?
51:49
Unidentified speaker: Let me go make some some hacking systems to go through all of these.
51:54
Unidentified speaker: Um there are hundreds of millions of devices right now that are open for attack.
51:59
Unidentified speaker: Well, we probably only have time for two more.
52:00
Unidentified speaker: Here's mine, and then Sandu feel free.
52:02
Unidentified speaker: Um, it's more of a playful one.
52:04
Unidentified speaker: But I I grew up in Springfield, Ohio, near Dayton, Ohio.
52:06
Unidentified speaker: They're Wright brothers, right?
52:07
Unidentified speaker: Kitty Hawk, the first plane to fly, way back a long time ago.
52:11
Unidentified speaker: I've heard one of the craziest stats I've ever heard is if you look at all the profits made from all the airplane companies since that fateful, not fateful, that first flight.
52:19
Unidentified speaker: They've literally made no money.
52:20
Unidentified speaker: Like literally, airplanes have never made any money, even in boom times.
52:23
Unidentified speaker: Airplanes don't make money.
52:24
Unidentified speaker: Yeah, my question to you that didn't make any money, and be honest, you're on a podcast, you're not allowed to you're not allowed to fib.
52:30
Unidentified speaker: You versus C um CZ in poker.
52:33
Unidentified speaker: Did you take money from him or did he take money from you?
52:35
Unidentified speaker: I I have to ask.
52:36
Unidentified speaker: I'm sorry, I just have to go down this path.
52:38
Unidentified speaker: I'm I'm very interested to know who's the better poker player.
52:41
Unidentified speaker: He's a very good poker player.
52:42
Unidentified speaker: Sure, yeah.
52:43
Unidentified speaker: Yeah, he he's very analytical and he doesn't he doesn't bluff as much, right?
52:48
Unidentified speaker: So he's he's very uh methodical in his play.
52:51
Unidentified speaker: I remember he probably won more than he lost.
52:55
Unidentified speaker: I'll just say that.
52:55
Unidentified speaker: Uh well yeah, okay.
52:57
Unidentified speaker: I gotta I I can see that.
52:58
Unidentified speaker: He yeah, okay, all right.
52:59
Unidentified speaker: You're you're pretty good too, though, right?
53:01
Unidentified speaker: You're good, I'm sure you're good poker players.
53:02
Unidentified speaker: I I held my own, but there you go.
53:06
Unidentified speaker: That's great.
53:07
Unidentified speaker: Go ahead, sorry.
53:08
Unidentified speaker: Alvin, you mentioned when you were talking about cybersecurity arrests and like mentioned the bad guys being able to do something now.
53:15
Unidentified speaker: That's in that event, well, if he catch them, put them in a jail.
53:20
Unidentified speaker: Throw the book at them, put them in jail.
53:23
Unidentified speaker: What if an agent is doing this?
53:27
Unidentified speaker: Who should be held liable if any of these agents have caused harm?
53:32
Unidentified speaker: Should it be the companies or shareholders, executives, or the people deploying them?
53:38
Unidentified speaker: And I want you to contrast that.
53:39
Unidentified speaker: I we don't have a lot of time, but you know, maybe in five minutes or something, three or four, three to five minutes.
53:44
Unidentified speaker: How does this differ from China's approach to accountability?
53:49
Unidentified speaker: Compare that to you know what we have, the Western world has really.
53:53
Unidentified speaker: Um, I I think this is actually one of the biggest problems we have as an industry, particularly in the AI space right now, is that that's not clear.
54:01
Unidentified speaker: Nobody can answer that question for you today.
54:04
Unidentified speaker: And if it's not clear, then that means that people are just going to do whatever um they think will give them the highest return on their investments or the highest value for their IPO.
54:16
Unidentified speaker: So it pushes people to take bigger risk.
54:20
Unidentified speaker: So if if we were playing poker and somebody says, uh, all your winnings you get to keep, and all your losses, uh, we'll just uh you know make it shift.
54:28
Unidentified speaker: Ryan D.
54:29
Unidentified speaker: Yeah, we'll socialize it.
54:31
Unidentified speaker: Um you want to play poker with me.
54:33
Unidentified speaker: I'm terrible.
54:33
Unidentified speaker: I'll big, I'll just drink a bunch, I'm fun, give me your money.
54:38
Unidentified speaker: Essentially, every play, every hand is an all-in hand.
54:41
Unidentified speaker: That's what happens, right?
54:42
Unidentified speaker: This this is the problem is that you only get the good side, there's no downside.
54:47
Unidentified speaker: And um, this is actually very different than what's happening in China.
54:50
Unidentified speaker: If you look at China, they they have uh direct liability to the companies and to the executives of those companies if the models create harm in society.
55:00
Unidentified speaker: Can they just make an executive disappear?
55:02
Unidentified speaker: I mean, I'm being very facetious there, but I mean, I I I think they they will definitely be called in for uh for some tea.
55:09
Unidentified speaker: So um, but but I mean, and and the the risk of that, I think, is what it is.
55:15
Unidentified speaker: The key is when you have that risk, you self-police because sure, because you have something to lose.
55:21
Unidentified speaker: Yeah, and right now these companies only have something to win, and they're all in a race to get to IPO because once they get the IPO, then they can pay off all their investors that's put their money in them for five to ten years who are closing out their funds and saying, get my money out before people wake up to this bubble.
55:38
Unidentified speaker: Um and let's let's socialize it to all the the retirement funds and and the pension funds out there, right?
55:46
Unidentified speaker: And so um I I I think this this needs to be clarified in America, and if it does, it'll actually make the world safer because then these companies will decide to move at a pace that is reasonable uh when they consider that there's there is upside and downside.
56:01
Unidentified speaker: In fact, I was in the in a uh uh in a AI kind of conference uh yesterday, yesterday, um, with uh one of the kind of big safety advocates, and they're saying, Oh, you know, what we need to do is get the insurance companies to to put higher fees on these companies.
56:16
Unidentified speaker: I'm like, how do you ensure the the destruction of all humanity?
56:21
Unidentified speaker: How do you price that?
56:22
Unidentified speaker: I don't know how you price that.
56:23
Unidentified speaker: I don't know.
56:24
Unidentified speaker: Well in that case, we don't have to worry about it.
56:26
Unidentified speaker: Well, no, but but this is the problem.
56:28
Unidentified speaker: And if you look at the anthropic s1, it's uh 90 pages of risk disclosures, and right, and it specifically says uh one of the risks is that we might kill all humanity.
56:38
Unidentified speaker: Um SpaceX was what 68 pages of risk?
56:42
Unidentified speaker: And they send space, they send rockets outer space, it was less pages, yeah.
56:46
Unidentified speaker: Yeah, well, but although I think a lot of the SpaceX was also AI-related uh risk as well.
56:51
Unidentified speaker: True, true.
56:52
Unidentified speaker: Um, and in fact, uh, SpaceX, um, I think 93% of their revenues were not space-related businesses, it was all AI businesses, and their $40 trillion you know, uh total available market was all about uh software and AI and and what they call application services, which essentially uh is uh replacing white-collar worker payroll.
57:14
Unidentified speaker: That was what the $35 trillion was, was uh global white-collar payroll uh uh amount.
57:21
Unidentified speaker: So well, Alvin, this was incredible.
57:25
Unidentified speaker: Uh this was the latest facts versus feeling social hour.
57:27
Unidentified speaker: Uh, we really appreciate this.
57:29
Unidentified speaker: And we've done six of them.
57:30
Unidentified speaker: I mean, this one I've I've learned more during this one than the rest of them because I didn't know too much coming in.
57:33
Unidentified speaker: Uh, we want to give you the floor for like the final two minutes or so.
57:36
Unidentified speaker: Maybe what makes you excited about this, about the industry?
57:39
Unidentified speaker: Well, what excites you about the future with AI and technology and all that stuff?
57:43
Unidentified speaker: And we'll kind of wrap it up with that.
57:44
Unidentified speaker: But again, thank you so much for joining us today.
57:46
Unidentified speaker: Yeah, I I I'm I'm both excited and and scared out of my mind at the same time right now.
57:52
Unidentified speaker: Because we what we do as as a society, as a species, the next three to five years is going to decide how you know our species turns out for essentially the the remainder of time.
58:08
Unidentified speaker: And um right now the decisions are being made by both our business leaders and our government leaders are pushing us towards a path where it does not end well.
58:19
Unidentified speaker: Um, and unless we change our ways, unless we change our our perception of how we deal with this technology, right now it's seen as a weapon for wealth or a weapon for power and domination.
58:32
Unidentified speaker: When you see it that way, then whatever it costs to invest to get it is worth it because at the end of that of that race, you own the world.
58:43
Unidentified speaker: And you've heard that in some of the leaders around saying things exactly like that.
58:47
Unidentified speaker: Um and whatever risk that you put yourself in for that is worth it because you're now leader of the world or something.
58:58
Unidentified speaker: So the the we we we need to change to a new way of looking at this.
59:03
Unidentified speaker: We need to change it to uh AI as a global public good.
59:07
Unidentified speaker: If we if we looked at it as you know, fresh air, fresh water, you know, just the the parks, right?
59:13
Unidentified speaker: Nobody's fighting to say who can make uh the freshest air.
59:16
Unidentified speaker: There aren't you know, people just like I we all want fresh air, let's just like not make bad air, right?
59:21
Unidentified speaker: That's kind of the the the type of thing, but nobody's racing to spend the most amount of money to to to clean up the climate right now.
59:27
Unidentified speaker: At least I I I wish there were more people doing it, but nobody's doing it, right?
59:30
Unidentified speaker: So if we looked at it as a public good, we want to make it available to everyone, we want to make it accessible and affordable.
59:38
Unidentified speaker: And right now we're we're doing the opposite.
59:40
Unidentified speaker: So uh, and if we if it was going to be something that everybody shares, then everybody should pay for it.
59:45
Unidentified speaker: So we should work on it together.
59:46
Unidentified speaker: We share the cost.
59:47
Unidentified speaker: We don't have to build another, you know, five or fifty trillion dollars of data centers.
59:51
Unidentified speaker: We could take the the amount of data centers we have today and just make one giant foundational model, which then everybody can fine-tune and and and you know do whatever they want with it afterwards and and distill the smaller models or whatever.
1:00:06
Unidentified speaker: And you know, right now 70% of the global compute for AI is used for training, not used for inference.
1:00:12
Unidentified speaker: Okay.
1:00:13
Unidentified speaker: So if we just and then it's split between you know, probably 20 or so labs.
1:00:17
Unidentified speaker: If if they all just made one, we immediately essentially, you know, 20x our compute capabilities without spending any money.
1:00:27
Unidentified speaker: Right.
1:00:28
Unidentified speaker: And then if we aren't racing, then we're gonna have more time to keep it safe.
1:00:32
Unidentified speaker: We're gonna put the right guardrails around it, you know, we're gonna have it come out in a time where uh the companies can actually utilize it and uh improve their productivity, improve their their innovation, right?
1:00:44
Unidentified speaker: Right now we're we're releasing models every week, you know, sometimes every day.
1:00:48
Unidentified speaker: Yep.
1:00:49
Unidentified speaker: How many people can actually keep up with that?
1:00:51
Unidentified speaker: How many companies, no, no companies can keep up with that?
1:00:53
Unidentified speaker: So why are we doing it?
1:00:55
Unidentified speaker: It is for this ego trip of saying I have the star for the week of I had the highest you know uh benchmark score.
1:01:01
Unidentified speaker: That that is worthless in the long run.
1:01:03
Unidentified speaker: Well, you get a star for joining us.
1:01:05
Unidentified speaker: I'll just say that.
1:01:06
Unidentified speaker: Uh Alvin, this this was amazing.
1:01:07
Unidentified speaker: Thank you to Sonu.
1:01:08
Unidentified speaker: Thank you to our producer, Pod Pony.
1:01:10
Unidentified speaker: Um, and thanks, Alvin, and thanks to everyone.
1:01:12
Unidentified speaker: I mean, I was looking, we had well over three 3,400 people or so listening just a few minutes ago.
1:01:16
Unidentified speaker: So throughout the I know both of you have flights to catch.
1:01:19
Unidentified speaker: So we're both a flight to catch.
1:01:21
Unidentified speaker: I see people down here doing stuff in the basement.
1:01:22
Unidentified speaker: I need to start back in and getting out of here.
1:01:24
Unidentified speaker: But Alvin, thank you so much.
1:01:25
Unidentified speaker: Sonu, I'm sure we'll chat soon, and thank you to everybody who paid attention.
1:01:30
Unidentified speaker: Look forward to coming in the future.
1:01:31
Unidentified speaker: Take care, guys.
1:01:32
Unidentified speaker: See you guys, and bye bye.
1:01:34
Unidentified speaker: Investment advisory services are offered to CWM LLC.
1:01:37
Unidentified speaker: That's the key registered investment advisor.
1:01:39
Unidentified speaker: This reporting is for informational purposes only and is intended for an audience of financial professionals who should not stop to reproduce double consent of CWM LLC.
1:01:49
Unidentified speaker: Guest presenters of our show are not affiliated with CWM LLC.
1:01:53
Unidentified speaker: Opinions expressed by these presenters may not be representative of CWM LLC.
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 Facts vs Feelings with Ryan Detrick & Sonu Varghese

Facts vs Feelings with Ryan Detrick & Sonu Varghese

Are You Not Entertained? (FvF Ep. 213)

Episode discussion: Carson Group analysts discuss midterm election impacts, data center power constraints, and AI monetization signals

Speakers’ claims: Ryan Detrick, Sonu Varghese, Blake Anderson, and Barry Gilbert analyze market volatility following midterm elections and recent economic data. They highlight historical S&P 500 performance post-midterms and discuss how voter sentiment on price levels influenced election outcomes. The conversation shifts to infrastructure challenges, noting ballot measures restricting data centers and potential regulatory changes in Texas affecting ROI. Finally, they examine AI monetization through Anthropic's revenue growth and upcoming congressional hearings on tech regulation.

Midterm ElectionsData CentersAI Monetization
58 min1 assets24 claims to verify
Facts vs Feelings with Ryan Detrick & Sonu Varghese

When Will the Madness End? (FvF Ep. 212)

Episode discussion: S&P 500 concentration peaks while hyperscaler CapEx surges to $1.15 trillion in 2027, raising capital cycle risks.

Speakers’ claims: Ryan Detrick and Sonu Varghese analyze a market environment characterized by extreme concentration in the S&P 500, where the top 10 stocks now represent 39% of the index, compared to 25% in 1999. They highlight a divergence in earnings growth, with the S&P 493 outperforming the Magnificent 7 for the first time since late 2022, while hyperscaler capital expenditures are projected to jump to $1.15 trillion in 2027. The hosts argue that while current valuations are lower than the dot-com era, the massive scale of AI infrastructure spending creates significant capital cycle risks and off-balance-sheet liabilities. Additionally, they discuss seasonal market trends favoring year-end rallies, the impact of rising real yields on Bitcoin and gold, and shifting political sentiment regarding data center construction.

Market ConcentrationHyperscaler CapExEarnings Growth
1h 5m11 assets25 claims to verify
All insights