Beyond the Noise: Markets, Investing, and the Bigger Picture

Anthropic IPO details: $1.4T valuation, $90B raise, and key financial metrics

53 min Listen to the original
AI IPOAnthropicVenture CapitalTech ValuationCloud ComputingMarket Risks

Summary

Speakers analyze Anthropic's NASDAQ IPO under ticker ANT, valued at approximately $1.4 trillion and raising $90 billion. They highlight a 14-fold revenue increase to an $81 billion annualized run rate, while noting risks such as heavy reliance on rented compute from competitors like SpaceX and significant lock-up periods for early investors. The discussion covers the company's 85.5% gross margins, the impact of recent price cuts on revenue, and the structural control retained by founders despite dilution.

Key takeaways

  1. 1
    Q3 revenue puts them at an annualized run rate of right at $68 billion.

    Unidentified speaker (speaker_1) · 0:12

Episode sections

  1. 0:05 – 0:12

    Anthropic IPO Overview and Valuation

    Anthropic announced its IPO with an expected valuation of approximately $1.4 trillion. The company is set to list on the NASDAQ under the ticker ANT on November 19, 2026, raising about $90 billion through new shares and private placements.

  2. 0:12 – 4:47

    Revenue Growth and Historical Context

    Q3 revenue placed Anthropic at an annualized run rate of $68 billion, accelerating to $81 billion by October. This represents a 14-fold (1400%) increase from the previous year, highlighting extraordinary growth compared to prior periods.

  3. 4:47 – 12:31

    IPO Structure and Lock-up Periods

    The IPO involves $64 billion in new shares and $26 billion in private placements from investors like Google and Amazon. Existing investors face lock-up periods ranging from 180 days to one year, with founders locked for a full year, creating potential downward selling pressure.

  4. 12:31 – 16:40

    Business Model: Frontier Labs and APIs

    Anthropic operates as a frontier lab, competing with OpenAI in providing top-tier large language models. Its business model includes developer tools and APIs, targeting the software development community and integrating with major enterprise software.

  5. 16:40 – 21:49

    Revenue Streams and Pricing Models

    Revenue comes from subscription models, consumption-based pricing (80% of revenue), and cloud-based system access. Consumption pricing aligns incentives with compute constraints, encouraging efficient use of resources.

  6. 21:49 – 24:03

    Customer Base and Concentration Risks

    The S1 filing reveals 11,900 customers spending over $100k annually, 1,900 over $1M, and 250 over $10M. The top five clients account for 25% of revenue, with no single client exceeding 10%, indicating moderate concentration.

  7. 24:03 – 25:15

    Total Addressable Market (TAM)

    Referencing David Frereberg, the speakers estimate a million companies globally could spend over $100k annually on AI, representing less than 1% of their revenue. This suggests significant untapped growth potential for Anthropic.

  8. 25:15 – 25:40

    Cash Burn and Financial Health

    Contrary to bearish narratives of massive cash burn, Anthropic has burned through only about $9 to $10 billion in cash as of the filing date, suggesting a stronger financial position than previously assumed by some investors.

  9. 25:40 – 28:46

    Compute Commitments and Liquidity

    Speaker_1 notes Anthropic has ~$140B cash and will raise ~$90B, creating a $250B war chest to cover $800-900B in compute commitments over 10 years. Speaker_0 highlights Anthropic rents compute rather than owning data centers, citing a $1.25B/month deal with SpaceX's Colossus.

  10. 28:46 – 31:46

    Revenue Growth and Accounting Losses

    Speaker_0 cites an $81B annualized revenue run rate (as of October) and notes a $600M revenue drop from price cuts. Speaker_1 explains a reported $104B accounting loss was largely due to $94B in bond-to-equity conversions, signaling debt holders' confidence.

  11. 31:46 – 36:59

    Recursive Self-Improvement and Margins

    Speaker_0 describes RSI where Claude writes 90% of its own code, 45x faster than in 2024. Speaker_1 highlights 85.5% gross margins, arguing this proves profitability despite expensive rented compute.

  12. 36:59 – 42:16

    Compute Supply Risks and Competition

    Speakers discuss risks of compute shortages and competitors like Google/Microsoft canceling contracts. Speaker_1 notes SpaceX contracts are quarterly and cancelable, posing a risk if Elon Musk prioritizes his own needs or raises prices.

  13. 42:16 – 46:01

    Customer Stickiness and Open Source Threats

    Speaker_1 notes 11,900 customers pay >$100k/year. Speaker_0 discusses the 'Iran war' analogy: using expensive frontier models for simple tasks is inefficient, potentially driving customers to cheaper open-source models.

  14. 46:01 – 53:11

    Founder Control and Product Liability

    Founders retain 50.1% voting power despite only 14% economic interest. Speakers discuss risks from founders' 'doomer' narratives and potential liability if Anthropic's AI is used for malicious purposes.

Assets discussed

AssetKindStanceContext
Anthropicprivate companymixedthat Claude and Anthropic are really leaning into something that we call uh RSI.
AlphabetGOOGLequityneutralYeah, pure play. You know, yes, you get it with Google, and yes, you get it with Meta and Microsoft, but it's a small, small piece of their business.
Meta PlatformsMETAequityneutralYeah, pure play. You know, yes, you get it with Google, and yes, you get it with Meta and Microsoft, but it's a small, small piece of their business.
MicrosoftMSFTequityneutralYeah, pure play. You know, yes, you get it with Google, and yes, you get it with Meta and Microsoft, but it's a small, small piece of their business.

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

Claims worth verifying

Valuation is supposed to be somewhere upwards of about $1.4 trillion.

Unidentified speaker (speaker_0) ·0:05

How to verify: SEC EDGAR S-1 filing for Anthropic; check 'Risk Factors' and 'Capitalization' sections for offering price and shares outstanding to calculate implied valuation. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Q3 revenue puts them at an annualized run rate of right at $68 billion.

Unidentified speaker (speaker_1) ·0:12

How to verify: SEC EDGAR S-1 filing; check 'Management's Discussion and Analysis' for Q3 revenue figures and annualization calculation. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

They announced in October that it had accelerated to $81 billion [annualized run rate].

Unidentified speaker (speaker_0) ·0:18

How to verify: SEC EDGAR S-1 filing; check 'Revenue' section for October or Q3/Q4 run-rate disclosures. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

That was already 14 times what the previous year had been.

Unidentified speaker (speaker_1) ·0:25

How to verify: SEC EDGAR S-1 filing; compare Q3 annualized revenue to prior year Q3 revenue or full year revenue. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic has been publicly reported as saying AI has a 10% or a 15% chance of ending humanity.

Unidentified speaker (speaker_1) ·0:44

How to verify: Search for Anthropic's 'Constitutional AI' or 'AI Safety' whitepapers and public statements by Dario Amodei regarding existential risk probabilities. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

They're set to go public on November 19th, which is Thursday of this week.

Unidentified speaker (speaker_0) ·2:01

How to verify: SEC EDGAR S-1 filing; check 'Plan of Distribution' or 'Underwriting' section for expected listing date. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

They're set to raise about $90 billion for the company.

Unidentified speaker (speaker_0) ·4:48

How to verify: SEC EDGAR S-1 filing; check 'Use of Proceeds' and 'Capitalization' sections for total offering size. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Founders are tied up for the full year.

Unidentified speaker (UNKNOWN) ·7:51

How to verify: SEC EDGAR S-1 filing; check 'Lock-up Agreements' section for details on insider and institutional lock-up periods. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Eleven thousand nine hundred customers paying anthropic over a hundred thousand dollars a year.

Unidentified speaker (speaker_1) ·21:49

How to verify: SEC EDGAR S-1 filing; check 'Risk Factors' or 'Business' section for customer concentration and revenue thresholds. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic is only burned through about nine to ten billion dollars of cash.

Unidentified speaker (UNKNOWN) ·25:35

How to verify: SEC EDGAR S-1 filing; check 'Balance Sheet' for cash and cash equivalents and 'Statement of Operations' for cumulative net loss since inception. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic has roughly $140 billion in cash on its books.

Unidentified speaker (speaker_1) ·25:41

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Cash and cash equivalents' line item. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic will raise another $90 billion, resulting in a $250 billion war chest.

Unidentified speaker (speaker_1) ·28:23

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Proceeds from sale of equity' or 'Cash from financing activities'. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic has commitments to spend $800 to $900 billion on compute over the next 10 years.

Unidentified speaker (speaker_1) ·27:42

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Commitments and contingencies' or 'Purchase obligations' footnote. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's annualized revenue run rate was $81 billion based on October figures.

Unidentified speaker (speaker_0) ·28:46

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Revenue' for the most recent quarter and annualize. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's price reduction reduced revenue by about $600 million in the quarter.

Unidentified speaker (speaker_0) ·29:05

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Revenue' and management discussion for price impact disclosures. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Approximately $94 billion of the $104 billion accounting loss was due to bond-to-equity conversions.

Unidentified speaker (UNKNOWN) ·30:28

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Statement of Operations' and 'Statement of Changes in Equity' or notes on convertible debt. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

About 90% of Claude's new code is written by itself.

Unidentified speaker (UNKNOWN) ·31:54

How to verify: Anthropic corporate blog or press release regarding 'Recursive Self-Improvement' or 'Claude Code' capabilities. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Anthropic's gross margins are 85.5%.

Unidentified speaker (speaker_0) ·33:12

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Gross Profit' and 'Revenue' to calculate margin. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic pays $1.2 billion a month for the Colossus data center deal.

Unidentified speaker (speaker_1) ·34:01

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Commitments and contingencies' or 'Leases' footnote for SpaceX/Colossus details. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Post-IPO, Anthropic's seven founders will hold 50.1% voting power.

Unidentified speaker (speaker_0) ·46:21

How to verify: SEC EDGAR filing for Anthropic's S-1 registration statement; check 'Capitalization' or 'Voting rights' section. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's future hinges on securing sufficient compute power and infrastructure to deliver expected growth.

Unidentified speaker (speaker_0) ·51:13

How to verify: Review Anthropic's S-1 filing for disclosed capital expenditure plans, data center leases, and compute procurement contracts. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's margins could drop from 85% to significantly lower levels if compute costs become prohibitive.

Unidentified speaker (speaker_0) ·51:26

How to verify: Check Anthropic's S-1 filing for historical gross margin percentages and risk factors related to compute costs. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's gross margin is 85%.

Unidentified speaker (speaker_0) ·51:26

How to verify: Verify against Anthropic's S-1 filing for reported gross profit margins. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Investors can get AI exposure through Google, Meta, and Microsoft, but it represents a small portion of their business.

Unidentified speaker (speaker_1) ·52:12

How to verify: Review recent quarterly earnings reports for Google, Meta, and Microsoft to assess the revenue contribution of AI-related segments versus legacy businesses. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

80% of investment in Google, Meta, and Microsoft is in legacy, older businesses.

Unidentified speaker (speaker_1) ·52:12

How to verify: Analyze segment reporting in recent quarterly filings for Google, Meta, and Microsoft to determine the revenue split between legacy and emerging AI businesses. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.

Primary source

Do your own research

  • Review the S-1 filing for detailed risk factors, specifically regarding compute supply contracts and customer concentration.
  • Verify the $1.25B/month SpaceX compute agreement terms and cancellation clauses in public filings or press releases.
  • Check historical performance data for the 10 largest tech IPOs to assess median intra-year declines and post-IPO returns.
  • Analyze Anthropic's revenue breakdown between subscription and consumption-based pricing to understand margin sustainability.
  • Investigate the 'recursive self-improvement' claims and their impact on development costs versus model performance.
  • Compare Anthropic's valuation multiples against peers like OpenAI (if private data is available) or public AI-adjacent companies.
  • Assess the impact of recent price cuts on revenue growth rates and customer acquisition costs.
  • Review founder lock-up expiration dates and potential insider selling patterns post-IPO.
Full transcriptexpand

0:05Unidentified speaker (speaker_0): Anthropic has announced that they are going public. Valuation is supposed to be somewhere upwards of about $1.4 trillion.

0:12Unidentified speaker (speaker_1): Q3 revenue puts them at an annualized run rate of right at $68 billion.

0:18Unidentified speaker (speaker_0): To demonstrate how quickly this company is growing, they announced in October that it had accelerated to $81 billion. The

0:25Unidentified speaker (speaker_1): crazy part is like that was already 14 times what the previous year had been. Not 14%, 14 times, 1400% higher than what it had been the year before. But

0:37Unidentified speaker (speaker_0): the question we have to ask ourselves is is this reasonable? And what risks could we see to this going forward?

0:44Unidentified speaker (speaker_1): Anthropic has been publicly reported as saying AI has a 10% or a 15% chance of ending humanity. What do you do with that? This is one of those risks where it's like it's really hard to measure how significantly a person should take

0:59Unidentified speaker (UNKNOWN): it.

1:04Unidentified speaker (speaker_1): Welcome back to the Beyond the Noise podcast. So, Max, as we were prepping for this episode, So I had a ran run across and I had an interesting thought, which was,

1:12Unidentified speaker (UNKNOWN): and

1:13Unidentified speaker (speaker_1): I wanted to ask you this question. Where do you think the name anthropic came from?

1:19Unidentified speaker (UNKNOWN): Well,

1:19Unidentified speaker (speaker_0): that's a great question. I have no idea, honestly.

1:23Unidentified speaker (speaker_1): So this was an interesting one as I researched, you know, what was the what were some of the names that they were considering? And they had some really weird ones. Uh generative sloth, I think sparrows. Generative

1:35Unidentified speaker (UNKNOWN): sloth?

1:35Unidentified speaker (speaker_1): No, no, generative comma, sloth. And then sparrow systems. Like of all names, sparrow systems.

1:43Unidentified speaker (speaker_0): Anthropic is weird, but it's way better

1:45Unidentified speaker (speaker_1): than

1:45Unidentified speaker (UNKNOWN): all of those.

1:46Unidentified speaker (speaker_1): So the reason they ultimately chose to name it anthropic was because anthropic has means to deal with humans.

1:52Unidentified speaker (speaker_0): Okay.

1:53Unidentified speaker (speaker_1): And so they ultimately wanted to point back to the the their desire for human good with the with the name of the company. So anyway,

2:00Unidentified speaker (UNKNOWN): found it. That's interesting.

2:01Unidentified speaker (speaker_0): That's interesting. But that ties directly into the topic that we wanted to talk about today. Uh and the name Anthropic is is relevant because Anthropic has announced that they are going public. And we're recording this episode uh right before they're about to go public. They're set to go public on November 19th, which is Thursday of this week. And I mentioned Anthropic because uh their ticker is going to be ANT ant, which doesn't make me think at all about anthropic.

2:31Unidentified speaker (UNKNOWN): No, maybe more

2:31Unidentified speaker (speaker_1): of that sloth.

2:32Unidentified speaker (UNKNOWN): That's right.

2:33Unidentified speaker (speaker_0): Seems like it might have been more relevant there. But we want to get into the details today about the company and the IPO and what that's ultimately going to look like. Um, and kind of our thoughts on the future of that company.

2:46Unidentified speaker (UNKNOWN): Um,

2:46Unidentified speaker (speaker_0): we've done some previous episodes on Anthropic before and the foundation. fact that they have been honestly one of the most impressive companies when it comes to the amount of growth that they've experienced. But I think we should start out by just telling our listeners a little bit today about what Anthropic actually does. The company was founded in 2021 by seven founders initially. Obviously, the

3:10Unidentified speaker (UNKNOWN): the

3:11Unidentified speaker (speaker_0): the main kind of figurehead there is Dario Amade. And it all collectively, well, all of them collectively came out of OpenAI. And that was a very interesting situation because that came out of some disagreements that they had with Sam. Yeah, some some real drama. And it caused them that they left because they were concerned about some of the oversight

3:34Unidentified speaker (UNKNOWN): uh

3:34Unidentified speaker (speaker_0): or lack thereof associated with AI and you know, kind of and we've seen this. really play out consistently from a lot of their messaging there, hey, is this really going to be good for the people? And is there going to be something where AI ends up creating more issues than than good? Trevor

3:49Unidentified speaker (UNKNOWN): Burrus, Jr.

3:49Unidentified speaker (speaker_1): There may also be a personality issue as well.

3:52Unidentified speaker (speaker_0): There may

3:52Unidentified speaker (UNKNOWN): have been.

3:52Unidentified speaker (speaker_0): That wouldn't surprise me. Um but ultimately what they ended up developing and kind of their what they do uh at the foundational level is is claw, Claude, right? I mean, that's what a lot of people re uh understand them as. And so this company, Anthropic, which its primary product is Claude, uh, they're gonna be listing here on Thursday, uh November nineteenth. Um, and it's gonna be interesting to see how all that plays out because this is set to be the biggest IPO in history. And, you know, we had the biggest IPO in history earlier this year. as well with SpaceX. And I want to just reiterate some of the lessons and

4:33Unidentified speaker (UNKNOWN): uh

4:33Unidentified speaker (speaker_0): factors to consider for investors when it comes to IPOs. Because, you know, here at our company, we have historically been and continue to be very reluctant to participate in initial public offerings,

4:47Unidentified speaker (UNKNOWN): given their

4:48Unidentified speaker (speaker_0): significant volatility and just uncertain. uncertainty about what that's going to look like going forward. But the first thing to note about this listing or this IPO is they're set to raise about $90 billion for the company. Valuation is supposed to be somewhere upwards of about $1.4 trillion. And that $90 billion that they're raising is about $64 billion of new shares that are being issued. and traded on the NASDAQ.

5:17Unidentified speaker (UNKNOWN): And

5:18Unidentified speaker (speaker_0): then the other $26 billion is set to come from a private placement. You're going to have companies like Google and Amazon, an affiliate of Taiwan Semiconductor, several other companies are going to be involved in that private placement as well. But $90 billion is what they're expected to raise. Now, why are we reluctant to participate? in these initial public offerings of these IPOs?

5:42Unidentified speaker (speaker_1): Sure. And we're not saying necessarily anthropics specifically, but in general, there's a couple of structural things that people need to be aware of. So, and we talked about this in when SpaceX went public as well. When companies are going to go public, it's important to understand what's happened before this. Okay. Investors have put, you know, since 2021, in this case, anthropics not been around around nearly as long as SpaceX, SpaceX. was 25 plus years old, but Anthropic is, you know, four or five years old. The individuals have put money in and they've put in money for the long term. And especially, you know, investors call it more than three years ago. They are sitting on very substantial gains. Oh,

6:18Unidentified speaker (speaker_0): yeah.

6:19Unidentified speaker (speaker_1): Very large gains. Um, even the, I think it was, what was it, the 2025 early valuation was set at what, $65 sixty five billion dollars roughly? So that means at a one point four trillion dollar valuation, people are looking at a ten to twelve X increase from that point. So

6:36Unidentified speaker (UNKNOWN): and

6:36Unidentified speaker (speaker_1): these are people that before this have largely not been able to sell their position.

6:41Unidentified speaker (speaker_0): Right.

6:41Unidentified speaker (speaker_1): So

6:42Unidentified speaker (UNKNOWN): and so

6:42Unidentified speaker (speaker_1): what does that mean? This is the first chance that they're going to get and they're gonna get to chance to sell it at a very large increase. Right. So there's a 10 and so literally that and that's happening all at once. Every sh not and to be clear, not all of those people will get unlocked at the same time. I think in this case we're looking at about a year.

6:59Unidentified speaker (speaker_0): Yeah, that that's very important to note because when company private companies go public, there's always some sort of uh restriction or liquidity uh illiquidity uh restriction put in place. And in the case of anthropic, basic you

7:14Unidentified speaker (UNKNOWN): know

7:14Unidentified speaker (speaker_0): is as best we can tell at this point, based on the data that's been released, is depending on what type of investor you are, you're locked up for anywhere from 180 days to a full year. So six months to a year. Now, if you are a non-founder, um, you know, just kind of an employee or a non-founder officer or something, you can sell generally

7:35Unidentified speaker (UNKNOWN): uh

7:35Unidentified speaker (speaker_0): at that six month mark. I think actually 10% ten percent of the shares of those types of investors are liquid day one. Sure. And then they get gradually released up until everything's fully released at uh six months after. Uh founders are tied up for

7:51Unidentified speaker (UNKNOWN): the

7:51Unidentified speaker (speaker_0): full year. Uh

7:52Unidentified speaker (UNKNOWN): so are

7:52Unidentified speaker (speaker_1): institutional investors.

7:53Unidentified speaker (UNKNOWN): So

7:53Unidentified speaker (speaker_0): are institutional investors. Now, founders also have the ability to liquidate some uh over the years, but that that's very important.

8:00Unidentified speaker (UNKNOWN): It's a

8:00Unidentified speaker (speaker_1): relatively small amount is what they intend to sell.

8:02Unidentified speaker (speaker_0): Right. Yeah, I think I heard the seven founders are expected to sell about $250 million each. Sounds like a lot. But when you realize the fact that after they've sold that $250 million, they still have a $20 billion stake each

8:14Unidentified speaker (speaker_1): pennies

8:15Unidentified speaker (speaker_0): in the company. We're not talking uh about a significant amount of sh uh of their shares. I think I read that it's about it's less than uh two 2% percent of their shares being sold. So we're we're talking about very small. amount. But why do we bring that up? We bring that up because when you have these investors that have owned these this company for a while and they're looking to liquidate their positions, what does that create?

8:38Unidentified speaker (speaker_1): Downward selling pressure.

8:40Unidentified speaker (speaker_0): Downward selling pressure. It's that classic supply and demand, right? If we have increasing supply where people are wanting to sell, that's going to drive the uh I won't say it's going to, it has the potential to drive. the the share price down.

8:52Unidentified speaker (UNKNOWN): Yeah,

8:52Unidentified speaker (speaker_1): it has an increased likelihood. I

8:53Unidentified speaker (UNKNOWN): mean,

8:53Unidentified speaker (speaker_1): that's the reality is once again, you know, from investors that were in since day one, they haven't had liquidity in four to five years. Right. Now they in, you know, a year from now, they will have full liquidity. And the question you have to ask is what will they do with it? For Or in, you know, historically, if we look at this, most of the time you see a large amount of people looking to exit. In most cases, because people have been tied up for a long time. It's like, okay, we've been tied up, we've been patient, we've made our money. It's now time to sell and move on to the next opportunity. When you and when you literally open that gate basically all at the same time, it tends to result in a lot of people running for the exits. Right.

9:31Unidentified speaker (UNKNOWN): And

9:31Unidentified speaker (speaker_1): so And

9:32Unidentified speaker (UNKNOWN): we've

9:32Unidentified speaker (speaker_0): seen this. I'm sorry, Interrupt,

9:33Unidentified speaker (UNKNOWN): but if

9:34Unidentified speaker (speaker_0): we've seen this historically.

9:35Unidentified speaker (UNKNOWN): Um,

9:35Unidentified speaker (speaker_0): if you look at the actual data, we we shared this when we talked about SpaceX. before. But if you look at the 10 largest tech IPOs in US history,

9:43Unidentified speaker (UNKNOWN): um,

9:43Unidentified speaker (speaker_0): and you look at their results during that first year, and this is almost entirely tied to this factor that we're talking about where people are looking to sell, you've seen

9:54Unidentified speaker (UNKNOWN): uh

9:54Unidentified speaker (speaker_0): a median intra year decline of fifty three percent. Yep. I mean, that's a massive amount of volatility in that first year.

10:02Unidentified speaker (UNKNOWN): Not

10:02Unidentified speaker (speaker_1): a fun ride.

10:03Unidentified speaker (UNKNOWN): No,

10:03Unidentified speaker (speaker_0): not a fun ride at all. And if you look at one year post-IPO, the median return was 0.2%. We're talking about effectively nothing, right? So um that is driven almost exclusively by this fact that you have historically private investors that are now looking to sell because they've gotten that liquidity.

10:25Unidentified speaker (speaker_1): Exactly. So that's a reason to be cautious. I do think there are a few things that are different. We'll be talking about some of them as we go throughout, but a couple just to mention now. One, um, anthropic has not been around for a long time. Right. This is not SpaceX after 25 years. This is not Google after whatever, seven or eight years. This is a a shorter period of time.

10:46Unidentified speaker (UNKNOWN): So the

10:46Unidentified speaker (speaker_1): investors have not had to necessarily be in the boat as long. Um, and so I think. that is an important factor to consider here. Two, we are dealing with one of the leading major frontier labs, the only one that's available to be invested on the market right now. And so there is a question how many of those investors will want to get off. And so I think one just a little side tangent to take here some investors will be forced to get out. You have as an example private equity companies, or even some venture capital companies where they have certain provisions where they're not allowed to hold the

11:21Unidentified speaker (UNKNOWN): public shares

11:22Unidentified speaker (speaker_1): for an extensive period of time.

11:23Unidentified speaker (UNKNOWN): They're

11:24Unidentified speaker (speaker_1): allowed to for a period of time, maybe a year or a couple years. And so you'll have some groups like that. You'll have others that their their um prospectus allows them to hold it for much longer. And so some of those may choose to hold it for a longer period of time. But so it's just important to be aware of. that dynamic.

11:40Unidentified speaker (speaker_0): Yeah. Yeah. It's going to be very interesting to see how it plays out over this first year because I I think you make a very important point, which is we're seeing a company that is growing rapidly. And if you have exposure to that company, the question is, do you really want to get out? I mean, we've seen them grow, and we'll talk more about this in a little bit, but we've seen them grow like 10 ten times each year from the previous year. Exactly.

12:06Unidentified speaker (UNKNOWN): Over their last two or three

12:08Unidentified speaker (speaker_0): years. So we're seeing significant growth. And and I agree with you. I think this may be a little bit different, but it's still a risk that we need to be

12:14Unidentified speaker (UNKNOWN): aware of.

12:15Unidentified speaker (speaker_1): Absolutely.

12:16Unidentified speaker (speaker_0): So let's let's get into a little bit more specifics about what they do. And uh there's some different layers. I mean, obviously Claude is kind of their

12:23Unidentified speaker (UNKNOWN): their

12:24Unidentified speaker (speaker_0): um

12:24Unidentified speaker (speaker_1): It's what people most know them most for.

12:27Unidentified speaker (UNKNOWN): Right.

12:27Unidentified speaker (speaker_0): But let's get into some more specifics about exactly what they do and kind of what their customer base is.

12:31Unidentified speaker (speaker_1): Yeah. So let's first talk about the the f the first major piece, which is they are a frontier lab. What does that mean? So when we talk about large language models, most most, you know, average Americans probably are familiar with Claude, with um Chat GPT, with Grok, now with Muse from Meta.

12:49Unidentified speaker (speaker_0): Yep.

12:50Unidentified speaker (speaker_1): Um would which and so these are or Muse is a a personal assistant, but the large language model Llama ultimately is is backing them. So the point is this you have multiple companies that have large language models that ultimately is the firepower behind, you know, these chat bots that people interact with. Okay. So there are varying levels of those large language models. To be on the frontier means you are in the top of the category. You're

13:16Unidentified speaker (speaker_0): the cutting edge.

13:17Unidentified speaker (UNKNOWN): Yeah,

13:17Unidentified speaker (speaker_1): exactly. You're the cutting edge of what of what we're dealing with there. And so Anthropic is on the cutting edge in multiple categories when it comes to large language models, which means that right now their predominant business model is if you need the most sophisticated, complicated large language model out there that can deal with the most difficult and compelling questions, this is one of the ones. you want to use. So if you're a if you're a life sciences company trying to do drug research, or if you are an engineering firm or a law firm and you're wanting to deal with very, and but in specifically you're wanting to deal with very complex questions, you're probably going to want to use and pay for the best that's out there. Right. Because the difference between the best and you know, the next level down, which might be 80% of the best, it's not worth it. You're dealing with things where you can bill a lot of money for your work, and therefore as a result, you're willing to pay more for it.

14:09Unidentified speaker (UNKNOWN): So

14:09Unidentified speaker (speaker_1): that's the first thing is they are on they're dealing with the frontier side. OpenAI is the other major one that's on the frontier. And you've got others that are vying to try to

14:17Unidentified speaker (UNKNOWN): get on to

14:18Unidentified speaker (speaker_1): the frontier.

14:18Unidentified speaker (UNKNOWN): Yeah,

14:19Unidentified speaker (speaker_0): they're they're absolutely the two uh leaders there.

14:21Unidentified speaker (speaker_1): Yeah, exactly. So the the frontier models would be the the first major area. The second major area is developer tools and APIs. So specifically here, one of the things that early on Claude and Anthropic ultimately did well was they decided to target the developer community, specifically software development. This is the best use case. Um, you know, there's some speculation on exactly why they thought that, but it was it was a good call. And and coding, computer coding is an area where they do particularly well because it's very much a you have the right answer or you have the wrong answer. It's it's ones and zeros, it's a language which large language models, as the name suggests, do really well with. And so they began to really focus there and they at especially in the earlier years, crushed it compared to their competitors. And so they began to build a very defensive moat there. And so between their coding solutions that they ultimately propagate in the market, and then also the APIs, which what are APIs? APIs are think

15:25of them as kind of the digital key that you can use to connect a tool like a Claude or whatever anthropic product you're using to one of your existing software tools.

15:37Unidentified speaker (UNKNOWN): Another

15:37Unidentified speaker (speaker_0): piece of software that you're using. Yeah.

15:38Unidentified speaker (UNKNOWN): Yeah.

15:38Unidentified speaker (speaker_1): I mean, let's say as an example that I used, you know, most tools have APIs. So we in here at the Fox Alliance, we use Slack as our in-office communication. I'm not allowed to do this, but just from a compliance standpoint, but let's just say theoretically I was, I was able and allowed to connect um Claude to Slack. I could then use it to say, hey,

16:00Unidentified speaker (UNKNOWN): are

16:01Unidentified speaker (speaker_1): there any messages that I haven't answered? Yeah. Are there communications that I should be involved with that I'm not? You could use it in a similar way that you use it just to ask general questions. You could ask it to interface with your particular application. This is where a lot of the real value with AI gets unlocked when you can connect it to your applications. And so that's been another big part of their model is making sure that they have APIs with all the major softwares out there so that their users can get substantial. benefit from those tools.

16:29Unidentified speaker (speaker_0): Yeah. And I think there are

16:30Unidentified speaker (UNKNOWN): kind

16:30Unidentified speaker (speaker_0): of two other big areas that I can think of where they are receiving customers and revenue from. And I I think honestly, they were uh

16:40Unidentified speaker (UNKNOWN): kind

16:41Unidentified speaker (speaker_0): of first to the game here when it came to engaging enterprises.

16:45Unidentified speaker (speaker_1): Yes.

16:45Unidentified speaker (speaker_0): Right. So at least for me personally, and I think for a lot of us, I know a lot of these AI systems through their chatbots.

16:51Unidentified speaker (UNKNOWN): Sure. Right.

16:51Unidentified speaker (speaker_0): So

16:52Unidentified speaker (UNKNOWN): you sign up for

16:52Unidentified speaker (speaker_0): Cloud, you sign up for uh ChatGPT, Google Gemini, whatever it may be, you pay your 20, 40, 100 bucks a month or whatever, and you got your chatbot and you're pretty much there. Um, that is not the main source of revenue for a company like Anthropic.

17:06Unidentified speaker (speaker_1): Exactly.

17:07Unidentified speaker (speaker_0): Enterprise, where you're creating or you're you're you're uh winning contracts with these huge companies.

17:14Unidentified speaker (UNKNOWN): Uh think

17:14Unidentified speaker (speaker_0): I about

17:14Unidentified speaker (UNKNOWN): that.

17:15Unidentified speaker (speaker_0): half of the Fortune 500 companies have a relationship with Claude that is very um

17:21Unidentified speaker (UNKNOWN): uh

17:22Unidentified speaker (speaker_0): beneficial to Anthropic. Um but they are building in um relationships with these big enterprise companies. And I I think this would probably be a good point to to note that there are two different ways that Anthropic charges for their services. Yep.

17:38Unidentified speaker (speaker_1): Yeah.

17:38Unidentified speaker (speaker_0): Uh at a high level, right? So one, the one that we're probably most familiar with and a lot of our listeners are, would be that subscription based model, right? You pay twenty bucks, a hundred bucks a month. There's different levels that you can do, but you have varying uh you have a a set amount of compute power that you can use. A lot of us don't really know that.

17:56Unidentified speaker (UNKNOWN): Yeah.

17:56Unidentified speaker (speaker_1): It's so large that

17:57Unidentified speaker (UNKNOWN): most

17:57Unidentified speaker (speaker_1): people don't use it. It's

17:59Unidentified speaker (UNKNOWN): more

17:59Unidentified speaker (speaker_0): significant than we need and and far, far sufficient. Um the other, which is about 80% of Anthropics revenue and is the way most of these enterprise contracts are put in place, is what we call consumption-based pricing. So the more you use, the more you pay.

18:17Unidentified speaker (speaker_1): Yep.

18:17Unidentified speaker (speaker_0): And that makes a lot of sense in today's world because we're very compute constrained. We've talked about this before. I know we'll probably be talking about this ad nauseum for many

18:26Unidentified speaker (UNKNOWN): more.

18:26Unidentified speaker (speaker_0): many more years to come. But we don't have enough compute. We've we know that there's lots of data centers being built. The hyperscalers we talked about are investing significant amount of money in

18:35Unidentified speaker (UNKNOWN): the

18:36Unidentified speaker (speaker_1): spot

18:37Unidentified speaker (UNKNOWN): in

18:37Unidentified speaker (speaker_1): comparison to the need.

18:38Unidentified speaker (speaker_0): Right. It's not enough. And charging based on how much you use is a a very honestly prudent approach for Anthropic to take. Um It benefits them, but it also allows them to allocate the resources to where they are are needed. Aaron

18:52Unidentified speaker (UNKNOWN): Powell

18:53Unidentified speaker (speaker_1): Yeah. If you have a a a friend who's like a coder and they're using one of these, they probably are maxing because

18:58Unidentified speaker (UNKNOWN): they

18:58Unidentified speaker (speaker_1): they're using it to do coding and other things that are far more compute intensive. Maybe someone in um visual creation as an example would be another another common one where you could uh I mean I know how I've I've ultimately maxed out in in situations like that as well. Aaron

19:12Unidentified speaker (UNKNOWN): Powell

19:12Unidentified speaker (speaker_0): It probably puts you in a position where you're less motivated to do dumb things with AI. Oh

19:20Unidentified speaker (UNKNOWN): yeah.

19:20Unidentified speaker (speaker_0): Right? Because I know uh there's lots of people out there that'll make just crazy graphics and stuff just for fun because you're messing around with it. That's fine when you're on a subscription based model. That can get pretty expensive when you are on uh a consumption based price.

19:34Unidentified speaker (UNKNOWN): Yeah, you

19:34Unidentified speaker (speaker_1): actually have to ask yourself, is it worth what I'm about to spend?

19:37Unidentified speaker (speaker_0): Which is again kind of getting to aligning incentives for what Anthropic wants to do. We don't want to use compute that we don't have to use because we're so constrained. So then the last point which I'll make here from a customer-based standpoint is their their last source of customers comes through cloud-based systems. So if you're a customer of AWS, you know, Amazon's cloud-based system or Azure, Microsoft's cloud-based system, they they kind of have third-party access to cloud and it can kind of be a add-on benefit that they provide to some of their customers. But those are kind of the four layers of where Anthropic gets a lot of their customers and revenue. And that's relevant as we kind of shift into the next part of this conversation, which is the S1, which is the document that kind of describes hey, this is what a private this company used to be private. Here's all the information that public investors that are considering investing in the IPO would need to know

20:37Unidentified speaker (UNKNOWN): about the

20:38Unidentified speaker (speaker_0): company. Um, there's been some interesting stuff revealed. So let's dive into that a little bit.

20:42Unidentified speaker (speaker_1): There's definitely some interesting things that we saw in the S1. The first of which was their revenue. If we looked at Q3 Q three revenue, they were just shy of 17 seventeen billion, which put puts them at an annualized run rate of right at sixty eight billion dollars a year. Which

20:57Unidentified speaker (UNKNOWN): you know,

20:57Unidentified speaker (speaker_0): and it I mean that's just ridiculous. It's

20:59Unidentified speaker (UNKNOWN): huge. Right. I

20:59Unidentified speaker (speaker_0): mean, that's crazy. And it's a huge increase over where it was before. But here to to demonstrate how quickly this company is growing, they just announced, uh which is obviously in their benefit because they're looking to uh IPO here shortly, but they announced in October that that it had egg accelerated. to $81 billion on an annualized basis.

21:18Unidentified speaker (speaker_1): And the crazy part is like that was already 14 times what the previous year had been. Fo not not not not 14%, 14 times, 1400% higher than what it had been the year before. And as a reminder, we we've you know we did an episode, it was actually our very first episode was on how anthropics what would anthropics growth that you've been. And even from the year before, they also had done a 10X in terms of their revenue. So this is a company that continues to grow very rapidly. So some other stats that we found from the S one. First, eleven thousand nine hundred customers paying anthropic over a hundred thousand dollars a year.

21:56Unidentified speaker (speaker_0): Wow.

21:57Unidentified speaker (speaker_1): Eleven 11,900. thousand nine hundred. Okay. We then have a total of nineteen 1,900 hundred companies. paying over a million dollars a year.

22:05Unidentified speaker (speaker_0): That's a significant number. I mean

22:06Unidentified speaker (UNKNOWN): when you

22:06Unidentified speaker (speaker_0): think about that. And this would obviously be based on that consumption pricing model. Exactly.

22:11Unidentified speaker (UNKNOWN): The more you use,

22:11Unidentified speaker (speaker_0): the more you pay.

22:12Unidentified speaker (speaker_1): Exactly. And then there are 250 customers paying more than $10 million a year to anthropic.

22:19Unidentified speaker (speaker_0): You know, th those are very interesting numbers. But I think that what that that does raise one concern. When you have customers paying that much is in annual revenue to you, um, and only 250 of them doing so, that raises some questions about concentration. And so do we know those numbers?

22:36Unidentified speaker (speaker_1): Yeah. So if you look at the top five clients do account for about 25% of the revenue. So there is some concentration there in the top. That's not overly concerning, especially when you pair that with the next number, which is there's not any one client that has uh more than 10%. of their revenue. So, yes, there's some bigger clients there at the top. And so, yes, from a concentration standpoint, that's a little bit more concentrated, but it's not honestly alarming for an enterprise solution. You know, there just are some bigger companies in the world that are going to demand more usage. Um, the other thing, though, I do think there's a opposite side of that, which is for a company to be willing to spend that much money,

23:13Unidentified speaker (UNKNOWN): that

23:14Unidentified speaker (speaker_1): means they at least for now have to believe that they are getting as much, if not more, value out of that spend

23:21Unidentified speaker (UNKNOWN): as

23:21Unidentified speaker (speaker_1): well, which I think is really key.

23:23Unidentified speaker (speaker_0): And that's been a question I think that has been asked for a long, long time. Are people getting benefit? Are we seeing a return from the amount of money that we're spending on AI? And I think based on those numbers, the the answer has to be. it sure seems so.

23:40Unidentified speaker (speaker_1): There's at least some companies that believe that's the case.

23:42Unidentified speaker (UNKNOWN): Exactly.

23:43Unidentified speaker (speaker_0): Now, when we when we kind of think about Anthropic's future growth and kind of where they're going from here, one of the questions that I think this raises is: all right, you've got companies that are spending a significant amount, right? 11,900 companies that are spending $100,000 or more.

23:57Unidentified speaker (UNKNOWN): I

23:57Unidentified speaker (speaker_0): think the question we need to ask ourselves is, is that sustainable and is there for

24:00Unidentified speaker (UNKNOWN): the same thing? a lot of things. Yeah, are we trapping this out or

24:02Unidentified speaker (speaker_1): is there a lot more room? to grow?

24:03Unidentified speaker (speaker_0): Right, right. So

24:04Unidentified speaker (UNKNOWN): to

24:05Unidentified speaker (speaker_0): answer that question, you kind of try to take a look at the total addressable market, the TAM. And to give credit where credit is due, um, David Freeberg from the All In podcast had provided some of this information here. But I think this is very relevant to when we think about anthropics' future growth. He estimated that there's about a million companies globally that could spend a hundred. thousand dollars a year or m on on AI, and that would be less or one percent or less of their revenue. So a very small amount of their revenue, but could they could easily spend that amount. That's a million companies. Yes.

24:41Unidentified speaker (speaker_1): And we're

24:42Unidentified speaker (speaker_0): only talking about eleven thousand nine hundred that are spending that or more

24:46Unidentified speaker (speaker_1): we're just scratching the surface of the potential.

24:48Unidentified speaker (speaker_0): There is a significant amount of growth potential. And I think this speaks to why we've seen such significant growth, right? I mean, if you're sitting there as a company and you're not spending much money on AI and you're seeing your competitors get real benefit from it, man, you better get on the train real quick. Sure.

25:04Unidentified speaker (UNKNOWN): Or

25:04Unidentified speaker (speaker_0): you're gonna be left behind.

25:05Unidentified speaker (speaker_1): Yeah, exactly. I think another key thing that's important that the S1 reveals is the amount of cash that Anthropic has on hand.

25:15Unidentified speaker (UNKNOWN): Yeah,

25:15Unidentified speaker (speaker_0): this was really interesting.

25:16Unidentified speaker (speaker_1): So this is a question that a lot of investors had speculated about for the last two or three years. It's like these these AI labs, they're burning cash like crazy. They're loss engines. It's like you're just dump throwing good money after bad. At least that's what the bears surrounding these labs ultimately were saying. But there's a couple of interesting things. First, as of right now,

25:35Unidentified speaker (UNKNOWN): uh

25:35Unidentified speaker (speaker_1): Anthropic is only burned through about nine to ten billion dollars. of casual cash.

25:40Unidentified speaker (speaker_0): Wow.

25:41Unidentified speaker (speaker_1): That is not a lot, especially considering the amount of money that they've raised, which is why they have a very large cash balance on their books. I think it was roughly 140 some odd billion dollars that they have on their books. Now, why is that important? Um, one of the major fears/slash Slash concerns that has been raised about anthropic anthropicus, is they've made all of these commitments to pay for compute. You know, they've said, hey, they several years back were like the biggest constraint is going to be compute back in a time where that wasn't a mainstream conviction. And so they began to work with multiple different groups and said, We will we agree to spend at least this much on compute. Now, some of it was officially contracted, some was more of a handshake arrangement, and some was a general, hey, we will try to. situation.

26:27Unidentified speaker (speaker_0): I think it's important to note here, Josh, just step back just a little bit. Um anthropic is different from the hyperscalers, you know, from the Googles, the Amazons, the Microsoft's of the world, because they don't actually own the compute, right? They're out there renting compute from people. I think a perfect example of this is kind of that uh blockbuster deal that Dario made with SpaceX, right?

26:50Unidentified speaker (UNKNOWN): And Elon Musk. Okay,

26:51Unidentified speaker (speaker_0): we're paying you one and a quarter billion dollars a month to be able to use your Colossus uh data center. But when we're talking about commitments, as these hyperscalers build out their data centers, they're not doing so just on the hopes that somebody's going to rent it from them. They want to have some commitments from their renters. Oh,

27:12Unidentified speaker (speaker_1): yeah,

27:12Unidentified speaker (UNKNOWN): because the

27:13Unidentified speaker (speaker_1): expenditures are massive.

27:14Unidentified speaker (UNKNOWN): Right.

27:14Unidentified speaker (speaker_0): And so they're they're investing a ton of money. They gotta know that somebody's on the other side saying, hey, we'll we'll take it once it's there. And so that's where all these commitments are coming from. Anthropic is not investing, uh at least uh in in the sense that they're building out their own data centers and and they're um uh you know actually physically owning these data centers. What they're doing is they're committing to say, hey, we will rent that from you. in the future once those uh data centers are up and running.

27:42Unidentified speaker (speaker_1): Exactly. And so, but the commit and the commitments they've made are sizable, depending on what how you want to classify it. And once again, it's not all in contractual, you have to spend this category, but it's on the order of roughly eight hundred to nine hundred billion dollars of commitments over the course of the next 10 years. Now, once again, not all those are illegal. legal. I have to do this. but there's a good chunk of them that are. And so one of the concerns was man, will they, I mean, this is a company that just started five years ago. Will they ever be able to actually do this? Part of the reason that they have this cash is to make sure they're in a position where they can do that. So you've got the hundred and forty billion, they're gonna be raising another ninety billion. Right. So this is a company that is going to have a quarter of a trillion dollar war chest. Yeah.

28:27Unidentified speaker (UNKNOWN): Let

28:27Unidentified speaker (speaker_1): me repeat that. They're going to have a quarter of a trillion dollar war chest. It's

28:32Unidentified speaker (speaker_0): a massive amount of liquidity.

28:33Unidentified speaker (speaker_1): It's a massive amount of liquidity. And when you combine that with the amount of growth that they're experiencing, I think this gives a pretty good reason to believe that these are commitments that they will be able Right. to keep. Right.

28:46Unidentified speaker (speaker_0): I mean, if you're, i you know, we talked about $81 billion run rate, meaning the revenue that they're receiving on an annual. annualized basis just based on October was $81 billion. It's very conceivable to think that just honestly in the next few months, if not sometime next year, we'll see them in a position where they're making a hundred billion dollars a year or more. And if you've got that combined with the $250 roughly billion dollars in liquidity available to you, that puts you in a position where you viably can uh follow through on these commitments that you've made.

29:17Unidentified speaker (speaker_1): Exactly. And I think it's also particularly that growth number is particularly compelling when you consider the fact that Anthropic lowered their prices recently. Um and they actually have said that it reduced their revenue by about six hundred million bucks in the quarter. So that's just one more reason to to look at it and say this is a company that is continuing to grow in a really strong and and meaningful way from that perspective.

29:39Unidentified speaker (speaker_0): Another thing that they did, which did impact their earnings and and could potentially go forward, was they have made some real efforts to crack down on uh inappropriate uses, right, of their uh AI. So you had some companies that were out there trying to distill their uh their model, right? Where basically you ask a bunch of questions of their AI, you're trying

30:02Unidentified speaker (speaker_1): to create.

30:02Unidentified speaker (UNKNOWN): and you're trying to

30:03Unidentified speaker (speaker_0): train your own AI to do effectively the same thing. And so they've been able to crack down on some of that, which also impacted their revenue to some extent. But obviously that's hopefully more of a one time thing where going forward they're going to just have more pure numbers going forward.

30:19Unidentified speaker (UNKNOWN): Now

30:19Unidentified speaker (speaker_1): one number that I think for an average investor if they're just looking at it could have been like, man, this is really concerning was there was a roughly a negative 104 billion quote loss

30:28Unidentified speaker (UNKNOWN): on the books. Right.

30:29Unidentified speaker (speaker_1): And so, you know, at first glance, it's like, uh, Josh, you said there was like nine to ten billion dollars of cash burn. What's going on here? The a very important distinction is accounting and legitimate cash burn are two very different things. And so one of the things that is important to note here is about $94 billion of So that $104 billion number was a change of about of bonds being. converted into equity,

30:52Unidentified speaker (speaker_0): right?

30:53Unidentified speaker (speaker_1): Which from an accounting standpoint has to show up as a loss on the books. Am I concerned about people who owned the debt of a company wanting to convert to the equity?

31:04Unidentified speaker (speaker_0): Absolutely not.

31:05Unidentified speaker (speaker_1): What is that communicating?

31:05Unidentified speaker (UNKNOWN): That's a

31:06Unidentified speaker (speaker_0): great sign.

31:06Unidentified speaker (UNKNOWN): It

31:06Unidentified speaker (speaker_1): is. It's communicating that the people who own the debt would much or are don't care about safety. They're not concerned about the company going under. They like the upside of the company and they want to own the equity. instead. Right. So that's one big piece that I saw a lot of negative headlines saying, you know, Anthropic has $104 billion loss in fiscal year 2025 or in the previous twelve months. And I think that's a number that was very much misunderstood based on uh individuals not looking below the hood and seeing what's actually happening um inside the car in that situation. Yeah.

31:38Unidentified speaker (speaker_0): I think I think maybe one of the last points about growth and kind of the future of AI, at least that I would make, would be that Claude and Anthropic are really leaning into something that we call uh RSI.

31:54Unidentified speaker (UNKNOWN): What

31:54Unidentified speaker (speaker_0): is RSI? RSI stands for recursive self-improvement, Improvement, where basically you're allowing the AI to improve itself, right? This is not uh necessarily other

32:05Unidentified speaker (UNKNOWN): uh

32:06Unidentified speaker (speaker_0): engineers and all of them being involved and saying, hey, no, don't do this, do that, make things change or you know, make these changes. But you're allowing the AI to you know ask questions of itself and and make itself better. And currently, about ninety percent of Claude's new code is being written by itself, which is really impressive. And it allows uh anthropic to very rapidly iterate and bring new models to market a lot faster. And it's creating a a situation where they're, you know, the the best companies can get better faster than their than their competitors.

32:45Unidentified speaker (speaker_1): And technically it's not recursive self-improvement self improvement because yes, the code is written by them, but it's not like Claude is. actually directing what the code is going to be written. That's human directed. But they're

32:56Unidentified speaker (speaker_0): working that way.

32:56Unidentified speaker (speaker_1): It is working. They are working that direction. The other thing that I think that is important to uh come out of this from the S one is if you look at the the gross, the incremental gross margins of this business, they're at eighty 85%. five percent.

33:12Unidentified speaker (speaker_0): It's a massive number.

33:14Unidentified speaker (speaker_1): Not just for perspective, when the large language models first started, everybody's like, this is going to be commoditized. This is never going to be a business anybody wants to own. Their profit margins aren't going to be nearly good enough to make this work. That now 85% does not say that. So if you really think about that, that means for every, you know, dollar

33:37Unidentified speaker (UNKNOWN): that

33:37Unidentified speaker (speaker_1): ultimately comes in the door. 85 cents

33:41Unidentified speaker (UNKNOWN): is

33:41Unidentified speaker (speaker_1): ultimately what they get to keep. And what what that ultimately reflects is, and I think this is at a really important point, which is a lot of people question and ask with how expensive compute is and the shortages and the fact that Anthropic is having to rent these data centers, are they going to possibly be able to charge enough in order to make it work? Right. And I think this specific number shows. that that absolutely is not going to be a problem. Right. Because yes, compute is incredibly expensive. Literally with that Colossus data center deal, they're paying $1.2 billion a month. But for every, you know, one dollar of expense that they're paying, they're generating a dramatically higher amount of actual dollars and revenue coming in the door.

34:24Unidentified speaker (speaker_0): Right. Two factors there in terms of, you when, know kind of when we think about their future growth. One is the margins. Hey, you know, if we've got that much gross margin available when we bring on, you know, a new uh amount of revenue, um, we can see prices decline and still be very profitable. Right. But the other factor there is increasing your sales, right? So we talked about this earlier with the total addressable margin. market. If we have significant growth from new customer potential out there, even if we do see our margins come down, you can still see significant growth from a company that's growing from a revenue, you know, from a customer-based standpoint.

35:05Unidentified speaker (speaker_1): Exactly. I think one other thing I want to just come back to the point you made about how Claude is writing our or uh the Anthropics large language models

35:12Unidentified speaker (UNKNOWN): are

35:12Unidentified speaker (speaker_1): writing 90% of their code. Um, One of the things you mentioned to me when we were researching this was this was forty time forty-five times faster than what it was in 2024.

35:22Unidentified speaker (UNKNOWN): That's right.

35:22Unidentified speaker (speaker_1): 45 times faster.

35:24Unidentified speaker (speaker_0): In two years.

35:25Unidentified speaker (speaker_1): Two years. That's that's amazing.

35:28Unidentified speaker (UNKNOWN): That

35:28Unidentified speaker (speaker_0): is amazing.

35:29Unidentified speaker (speaker_1): That results in a massive amount of productivity gains that that company's ultimately able to generate. Now there are I th that now that doesn't mean what you have is what they have. I'm sure it's not, but still. that's an amazing feat.

35:40Unidentified speaker (speaker_0): Right. Yeah. And if they and if they can take what, like you said, I'm sure is even a better system than what we have access to, and they're able to apply that to their own system and make their system even better, faster, that allows them to further develop their moats as a country countr yeah as a company compared to competitors and put them in a very strong position going forward. Now this is all very positive for them.

36:05Unidentified speaker (speaker_1): Sure.

36:06Unidentified speaker (speaker_0): Um they're listing at you know that roughly that $1.4 trillion dollar valuation, which is roughly seventeen times gross revenue, something like that. Um

36:15Unidentified speaker (speaker_1): what was SpaceX at again? It was I think

36:17Unidentified speaker (UNKNOWN): it was in the

36:17Unidentified speaker (speaker_1): forty to fifty.

36:19Unidentified speaker (UNKNOWN): Yeah it was a

36:20Unidentified speaker (speaker_1): lot higher.

36:21Unidentified speaker (speaker_0): It honestly might have been closer to a hundred. I think it was I think

36:24Unidentified speaker (UNKNOWN): you're right actually it was I think it

36:25Unidentified speaker (speaker_1): was 90 ninety times three.

36:26Unidentified speaker (UNKNOWN): you're right.

36:26Unidentified speaker (speaker_1): Mm-hmm

36:27Unidentified speaker (speaker_0): So we're talking about a much more reasonable, I'll put quotes around reasonable valuation than what we saw with um SpaceX.

36:35Unidentified speaker (UNKNOWN): Hey

36:36Unidentified speaker (speaker_0): everyone, if you're enjoying the insights that we're bringing you on this channel, do us a huge favor and hit the subscribe button on Apple, Spotify, wherever it is that you listen to your podcast. It takes two seconds, is completely free, and helps us continue to bring you high quality content.

36:51Unidentified speaker (UNKNOWN): But

36:51Unidentified speaker (speaker_0): the question we have to ask ourselves is, is this reasonable and what risks could we see to this going forward? I you know, uh

36:59Unidentified speaker (UNKNOWN): uh

36:59Unidentified speaker (speaker_0): the the first one that comes to my mind is the constraint that we have with compute.

37:04Unidentified speaker (speaker_1): Yeah.

37:05Unidentified speaker (speaker_0): Right. And I think that has become um more and more prevalent and more and more relevant today. We mentioned this just earlier. Anthropic rents their compute from other uh companies.

37:20Unidentified speaker (speaker_1): Sure.

37:20Unidentified speaker (speaker_0): The Amazons of the world, the Facebooks of the or sorry, the uh Microsofts of the world, the Googles of the World, even the SpaceXs of the world, right?

37:28Unidentified speaker (UNKNOWN): Yeah.

37:28Unidentified speaker (speaker_0): And what I think is interesting about that dynamic is in a world where can compute is constrained, there's there a very strange dynamic here. where you have a company like Anthropic renting data center power compute um from their competitors. Sure. Right? You've got Googles of the world that have Gemini and are trying to compete with Anthropic. Obviously if you pay them enough, that may be valuable. But does there come a point where either A, you just don't have enough compute to go around

38:02Unidentified speaker (speaker_1): and

38:02Unidentified speaker (speaker_0): there they're unable to see the growth that they expect to see because there's not enough to go around. Or B, we find ourselves in a position where the Googles of the world and the SpaceX of the world say, hey, sorry, we need this compute for our own uh purposes, and you're you know, in a lurch.

38:21Unidentified speaker (speaker_1): Let's unpack this a little more. So I think there's two elements. The first of which is imagine that you own a group grocery store in the middle of a hurricane. Okay. And everybody comes in like they always do. They buy as much as they can. And what happens?

38:33Unidentified speaker (speaker_0): Particularly toilet paper.

38:34Unidentified speaker (speaker_1): Exactly.

38:34Unidentified speaker (UNKNOWN): What is the most important?

38:35Unidentified speaker (speaker_1): Exactly. So they end up, everything's gone.

38:39Unidentified speaker (UNKNOWN): You

38:39Unidentified speaker (speaker_1): as a store owner have a line of customers out the door, but you literally have no merchandise to sell.

38:43Unidentified speaker (UNKNOWN): So

38:43Unidentified speaker (speaker_1): that's the first predicament that Anthropic can could find themselves in is um one, data centers are already in short supply. Um, you know, with some of the political pressure now from the coming out of the election, there's some risk that we don't get as many data centers built as we originally were hoping for. And so you could be in a position where Anthropic has no problem selling their service. There's plenty of customers willing to pay,

39:07Unidentified speaker (UNKNOWN): but

39:07Unidentified speaker (speaker_1): if you don't actually have the data centers to fulfill their desire, their need, the amount of demand that they generate, that's the first problem you have. The second problem is what you just described, which is in a model where you rent compute from individuals, which some of those um arrangements, I'm when we're not worried about the NeoClouds. Their

39:28Unidentified speaker (UNKNOWN): business

39:28Unidentified speaker (speaker_1): is to rent

39:28Unidentified speaker (UNKNOWN): compute.

39:29Unidentified speaker (speaker_1): They're never going to be a competitor to anthropic. I'm not worried about those contracts. What I am worried about though is probably most specifically is the SpaceX contract, because this that can literally be on a quarterly basis be canceled

39:40Unidentified speaker (speaker_0): every 90 days

39:41Unidentified speaker (speaker_1): by either side. And so, which I don't think at this point Anthropic has any interest in canceling.

39:46Unidentified speaker (speaker_0): Right.

39:47Unidentified speaker (speaker_1): Um, but one, Elon could decide, you know what? I want to hurt my competitor.

39:52Unidentified speaker (speaker_0): Right.

39:52Unidentified speaker (speaker_1): I'm going to, now that I have a need for this compute, I'm going to stop giving it to them. And suddenly Anthropic is not able, is has much more constraint when it comes to their compute that can't

40:02Unidentified speaker (UNKNOWN): be.

40:02Unidentified speaker (speaker_1): They can't serve as many customers. That becomes a big challenge. Another reason that Elon could ultimately end up doing that is we our belief is that as constrained as the compute situation is right now, it is going to get worse.

40:15Unidentified speaker (speaker_0): Yeah.

40:16Unidentified speaker (speaker_1): The the elections probably secured that being a high likelihood. Um and so as a result, we will likely see the cost that people have to pay for every um search quarry and everything else. continue to go up in price. So another possibility could be, hey, I am going to we either are renegotiating this contract to increase prices, or um I'm going to

40:38Unidentified speaker (UNKNOWN): found

40:38Unidentified speaker (speaker_1): someone else who's gonna

40:39Unidentified speaker (UNKNOWN): pay it and I'm

40:39Unidentified speaker (speaker_1): gonna go there instead. So I think that's another viable risk. And also, you know, going back to the hyperscalers, your your Googles, your Amazons, your Microsoft's, your Metas, these groups would love, they're not on most of them are not touching frontier. If they are, they're only in a couple of elements. Um

40:56Unidentified speaker (UNKNOWN): these

40:56Unidentified speaker (speaker_1): companies would love to be on the frontier though. And if they felt like taking away oxygen from the competitor would ultimately give them a net benefit, that is another viable strategy that they could ultimately choose to, you know, go after and execute.

41:08Unidentified speaker (speaker_0): And it wouldn't necessarily just be taking away oxygen from them. It could be then, you know, taking that and using it for their own model. to train it up and make it even stronger.

41:17Unidentified speaker (speaker_1): I'm less I am less worried about um some of the hyperscalers doing that just because they own substantial portions of anthropic. So they have a very vested interest in anthropic doing well, but um Elon does not,

41:31Unidentified speaker (UNKNOWN): or

41:31Unidentified speaker (speaker_1): at least not anything sizable that we know of. And there's a number there's a few of the hyperscalers that do not. Microsoft is an example. And so those are good.

41:38Unidentified speaker (UNKNOWN): they

41:39Unidentified speaker (speaker_0): actually own

41:40Unidentified speaker (UNKNOWN): a good chunk of

41:40Unidentified speaker (speaker_0): their competitors.

41:41Unidentified speaker (UNKNOWN): Exactly.

41:41Unidentified speaker (speaker_1): They own more of

41:42Unidentified speaker (UNKNOWN): Open

41:42Unidentified speaker (speaker_1): AI. Exactly.

41:43Unidentified speaker (speaker_0): Right. The other component here that is going to continue to be, and it it's all tied in with the data centers and the compute constraint that we have, comes back to electricity. I mean, we've talked about this many, many times in the past, but we are very constrained when it comes to power generation in this country. And I don't see that improving. anytime soon. Well,

42:04Unidentified speaker (speaker_1): it's improving, but just at a very slow rate.

42:06Unidentified speaker (speaker_0): Correct. Yeah. Speeding up, I

42:07Unidentified speaker (UNKNOWN): guess

42:07Unidentified speaker (speaker_0): I should say.

42:08Unidentified speaker (speaker_1): Sure.

42:08Unidentified speaker (speaker_0): Um, in in the near future. And so that's going to continue to be a constraint and certainly a risk that investors are going to need to pay attention to going forward.

42:16Unidentified speaker (speaker_1): Yeah. No, I absolutely agree. Another question that has come up in is how sticky will these customers be?

42:23Unidentified speaker (UNKNOWN): I mean,

42:23Unidentified speaker (speaker_1): once again, Anthropic has uh a lot of I mean, once again, level eleven thousand nine hundred customers paying over a hundred thousand dollars a year, two hundred and fifty paying over a million dollars a year, uh, and a smaller number paying over ten million dollars a year. So the question is, uh do they have viable alternatives? And so one of the things that's been talked about a lot is oh this these open source models. Yeah. Okay. They're not as good as the frontier, they're 80% as good as as the frontier labs. And so the question. is, could potentially some of these customers ultimately try to basically build out their own compute and then basically run these open weight models on local um or maybe outsourced data center compute and then try to do try to recreate enough of what they need that way at a lower cost point than what Anthropic does.

43:12Unidentified speaker (speaker_0): This has been a real interesting dynamic that I've seen. kind of present itself, especially over the last six months to a year. And it's kind of what they refer to as um on premise, you know,

43:23Unidentified speaker (UNKNOWN): um

43:23Unidentified speaker (speaker_0): build out of your own kind of effective data center or just renting it right from from somewhere else. But this concept of of using open source models to meet the needs that you have, right? Because open source, like you mentioned earlier, is there's there's not a cost for it. You can download the model effectively and use it

43:41Unidentified speaker (UNKNOWN): for

43:41Unidentified speaker (speaker_0): whatever reason. But you got to have some compute. And so uh, you know, companies like Dell are building these out for some comp uh, you know, for some larger companies that want to build that, you know, a data center out themselves, you know, kind of a local data center for lack of a better term. But that cost is pretty substantial to build that out yourself, but with the garage. rising cost of

44:04Unidentified speaker (speaker_1): and the rising usage.

44:05Unidentified speaker (speaker_0): Right. And the rising usage of,

44:07Unidentified speaker (UNKNOWN): you

44:08Unidentified speaker (speaker_0): know, a resource like Claude or ChatGPT or something that you would be using for your company. That's starting to make more and more sense. And so that ties in specifically to your question. Hey, are these companies that are currently spending a million dollars a year, are they going to continue to do that? Or is that maybe they'll use $100,000 or $200,000 a year for the real in-depth research that they need, but you know, kind of effectively move a lot of that compute to somewhere else?

44:34Unidentified speaker (speaker_1): Think about it this way. You know, in the um the Iran war, okay, we have seen um, and also with the war with the Houthis, we've seen small low-cost drones getting shot down by $10,000, $10, $15 million missiles. That is the equivalent of what using a Claude or a OpenAI's uh ChatGPT is like. You're using a $15 million ballistic missile for a common project that could be done with a much lower cost tool.

45:03Unidentified speaker (UNKNOWN): To

45:03Unidentified speaker (speaker_0): make my meme. Yeah.

45:05Unidentified speaker (UNKNOWN): Exactly. It's like

45:05Unidentified speaker (speaker_1): I if when you ch do a search for, you know, what's the weather today? It's like you are using a massively over equipped tool to answer that question. You could do it with a far look low lower cost. Average person doesn't care because they're on a subscription and they don't see a difference. But a company, um, you know, when they're spending a million bucks a year, and let's say maybe half a million of it is more simple rudimentary tasks that

45:28Unidentified speaker (UNKNOWN): doesn't

45:28Unidentified speaker (speaker_0): require.

45:29Unidentified speaker (speaker_1): Yeah. That a open source model could do versus, you know, maybe half of what you're doing is stuff that you really need the leading frontier level models to do. Right. And so we have there's there's no question you've seen a growth in open source models. But the thing is, even while you've seen that, you've still seen very large growth for these closed models like anthropic, um, claude, and open I OpenAI is chat GBT as

45:55Unidentified speaker (UNKNOWN): an

45:55Unidentified speaker (speaker_1): example as

45:56Unidentified speaker (UNKNOWN): well.

45:56Unidentified speaker (speaker_0): It'll be interesting to see how that plays out. Another factor that I think is relevant,

46:01Unidentified speaker (UNKNOWN): um,

46:01Unidentified speaker (speaker_0): and you'll see this a lot of times. with companies that go public and their founders are still heavily involved. But an interesting dynamic about anthropic here is those seven founders I mentioned, they still have a significant stake in the company. But post APOP IPO, they are going to have about 50.1% voting power, right? So for all intents and purposes, they're still in control of the company. That's you know, a lot of times when you look at a publicly traded company, the shareholders have more influence, right? And how that plays out. That's not going to be as true in this situation. So we could find ourselves potentially at some point where shareholders have one desired direction and the founders say, no, we want to go this direction, and there's some conflict there. Yeah,

46:48Unidentified speaker (speaker_1): and that's an interesting dynamic because I think between the some founders that each own they a about two percent of the company. Yeah.

46:53Unidentified speaker (speaker_0): So

46:53Unidentified speaker (speaker_1): they own about 14% of the economic interest, but they own 50 plus 0.1%

46:58Unidentified speaker (UNKNOWN): of

46:58Unidentified speaker (speaker_1): the actual voting interest. So yeah, it's a risk that people have to take into account. And this is this is a risk that you see more in founder led companies. You see this with Zuckerberg.

47:07Unidentified speaker (speaker_0): Yep.

47:08Unidentified speaker (speaker_1): Zuckerberg does not own a 51% fifty one percent of the company, but he has control, controlling votes. So like what we're seeing with Jensen. right now would not be able to happen unless you had one of the foot other founders at OpenAI Break ranks from the other seven.

47:22Unidentified speaker (speaker_0): Right.

47:23Unidentified speaker (speaker_1): Um that wouldn't that would just wouldn't happen.

47:25Unidentified speaker (speaker_0): I mean, that's been true with uh companies that Elon Musk runs too, right? SpaceX and and others, very similar. And that can be a good thing, but that can also be a risk as well, right?

47:35Unidentified speaker (UNKNOWN): Yeah,

47:35Unidentified speaker (speaker_1): it it can provide insulation so that the the founders can do the things that they really believe are the best interests of the company. They don't have to worry about a couple of bad quarters. But on the flip side, if they're if they go off the rails and start to do some things that are really problematic, you it it's really difficult to remove them.

47:49Unidentified speaker (speaker_0): And you know, if there's any company whose founders I think could do some pretty wild things, anthropic is one of those. It gives me uh some question. Just, you know, they have had some very interesting. news stories and press releases that

48:06Unidentified speaker (UNKNOWN): they

48:06Unidentified speaker (speaker_0): have made over the last year or two, all related to this uh

48:11Unidentified speaker (UNKNOWN): you know,

48:11Unidentified speaker (speaker_0): kind of Doomer kind of um subculture, right? Where hey, yes, we're developing AI and we're leaning heavily into it, but man, we better watch out. This could take over the world.

48:22Unidentified speaker (speaker_1): It's a really so honestly, and that kind of brings up the really the one of the last major risks, which is this product liability risk. Okay. So the challenge is, you know, anthropic has been publicly reported as saying AI has a 10% or a 15% chance of ending humanity.

48:41Unidentified speaker (UNKNOWN): Right.

48:42Unidentified speaker (speaker_1): What do you do with that? I mean, like, seriously, okay, hey, I'm building this thing and I'm accelerating in and building this thing, but I think that what I'm building actually. could literally destroy mankind. Right. So the change.

48:54Unidentified speaker (UNKNOWN): If you

48:54Unidentified speaker (speaker_0): really believe that, yes. Why are you continuing to accelerate

48:58Unidentified speaker (speaker_1): at the same at

48:58Unidentified speaker (UNKNOWN): the

48:59Unidentified speaker (speaker_1): same level? And I'm sure the answer would be something to the effect of, well, if someone's gonna be in charge of this,

49:03Unidentified speaker (UNKNOWN): we would rather do us than

49:04Unidentified speaker (speaker_1): than somebody else. But the but the bottom line is investors are taking a risk. Yep. Um you had um another um,

49:13Unidentified speaker (UNKNOWN): you know,

49:13Unidentified speaker (speaker_1): you had a quote, whistle. whistleblower, a guy that worked there for two

49:16Unidentified speaker (UNKNOWN): weeks

49:16Unidentified speaker (speaker_1): come out and and basically, you know, decide all these problems. And then you had some people internally from inside of Anthropic saying, Yeah, we agree with this guy. And it's just like with these kinds of admissions, if something bad does happen and um a bad actor uses anthropic to carry out a terrorist attack or build a uh you know a uh bomb

49:38Unidentified speaker (UNKNOWN): or a dirty

49:39Unidentified speaker (speaker_1): bomb or something else like that the question is like will in most cases you know you know like things with like guns hey you know if if someone uses a gun to do a shooting it's like hey we're not responsible for what someone does with with the gun. But in this case with an anthropic with the you know level levels of executives saying you know this could happen you question is do you really open up your yourself to a a massive amount of liability that ultimately could impact shareholders in the future? Um this is one of those risks where it's like it's really hard to measure how significantly a person should actually and how seriously they a person should take it.

50:17Unidentified speaker (speaker_0): And I think just kind of the takeaway there is if you're going to own this stock, recognize that that's a reality. This is just going to be more susceptible to some of those liabilities liability concerns associated with their

50:29Unidentified speaker (UNKNOWN): um

50:30Unidentified speaker (speaker_0): with their with their product.

50:32Unidentified speaker (UNKNOWN): I

50:32Unidentified speaker (speaker_1): do think one other thing, kind of just stepping back from the risks for a moment to take into account is AI stocks right now are not doing well. You know, we've gone through a three to a six month period of time where AI stocks have been on a downturn. And if you remember, Anthropic was originally rumored to be going public at two 2 trillion bucks.

50:50Unidentified speaker (UNKNOWN): Now we're

50:50Unidentified speaker (speaker_1): seeing sitting at a number that's closer to 1.4. And I think that's just a reflection of the fact that Anthropic realizes the multiples aren't the same as they were six months ago. So I do think that actually is another positive piece, which is I do think that the IPO price is lower than what it would

51:07Unidentified speaker (UNKNOWN): have

51:07Unidentified speaker (speaker_1): been six months ago. And therefore the valuation that investors are getting is

51:11Unidentified speaker (UNKNOWN): a

51:11Unidentified speaker (speaker_1): little bit more attractive than what it would have been six months ago. But

51:13Unidentified speaker (speaker_0): I think their future is really. gonna hinge on a couple things. One, are they going to be able to have the compute power and the infrastructure that they need to be able to deliver the continued growth that people are expecting?

51:25Unidentified speaker (speaker_1): Yes.

51:26Unidentified speaker (speaker_0): Two, are we going to see a cost structure of that compute not be prohibitive, right? Where it puts them in a position where their margins go from 85% down to something far less.

51:37Unidentified speaker (speaker_1): Sure.

51:38Unidentified speaker (speaker_0): Um three, I think, are they going to be able to safely and reliably continue to develop this product, right? And what type of risks are going to be associated with that? And so ultimately, we'll see how they answer each of those questions and how all that plays out. But it's going to be very interesting to see Um with the the listing how everyday investors respond to this, how institutional investors respond to this. Because like you said right at the beginning, this is really kind of the first opportunity that everyday investors are going to have to get exposure to one of the frontier models.

52:12Unidentified speaker (speaker_1): Yeah, pure play. You know, yes, you get it with Google, and yes, you get it with Meta and Microsoft, but it's a small, small piece of their business. You're also 80% of your investment is in legacy, old, older businesses, not truly pure play. large language models.

52:28Unidentified speaker (speaker_0): Right. I'm excited to see how this all plays out.

52:30Unidentified speaker (speaker_1): I am as

52:31Unidentified speaker (UNKNOWN): well. The

52:43Unidentified speaker (speaker_0): opinions voiced in this show are for general information only and are not intended to provide specific advice or recommendations for any individual. To determine which strategies or in investments may be suitable for you, consult the appropriate qualified professional prior to making a decision. Investing includes risks, including fluctuating prices and loss of principle. Securities and advisory services offered through LPL Financial, a registered investment advisor, member FINRA, and SIPC. The Fox Alliance Wealth Advisors is a separate entity from LPL Financial.

This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights