Milk Road Stocks

Speakers debate Anthropic's IPO valuation, citing margin risks and competitive pressures despite rapid revenue growth

52 min Listen to the original Episode page
AI InfrastructureIPO ValuationEnterprise SoftwareCompute CostsMarket Competition

Summary

Speakers analyze Anthropic's IPO, noting its historic revenue ramp from $2 billion to $17 billion quarterly but questioning the ~$1 trillion valuation. Kyle argues the market underprices AI agent demand and token usage growth, while Martin presents a valuation model suggesting limited upside at current prices. The discussion covers margin pressure from rising compute costs, enterprise stickiness with clients like Walmart and Salesforce, and potential trading opportunities if the stock dips. Speakers also compare Anthropic's post-IPO trajectory to SpaceX and suggest indirect exposure via AI infrastructure.

Episode sections

  1. 0:00 – 3:40

    Intro and IPO Context

    Speakers introduce the episode, noting Anthropic's rapid revenue growth from $2 billion to $17 billion quarterly and its IPO valuation drop from $2 trillion to ~$1 trillion. They outline the goal to value the company and discuss buy/sell decisions.

  2. 3:40 – 5:49

    Bear Case: Competition and Model Commoditization

    Martin argues that Anthropic faces intense competition from open-source models and rivals like Grok, which may reduce Anthropic's market share as enterprises optimize costs. He suggests Anthropic needs to pivot to niche application markets like Microsoft 365 to survive long-term.

  3. 5:49 – 13:19

    Financials and Growth Metrics

    Kyle highlights Anthropic's historical revenue growth, citing Q2 2026 growth of 144% QoQ and Q3 2026 revenue of $17 billion. He notes Goldman Sachs forecasts $600 billion revenue by 2030, with $52 billion expected by end of 2026.

  4. 13:19 – 15:41

    Valuation Model and Upside Analysis

    Martin presents a valuation model showing that even if Anthropic reaches $440 billion in revenue by 2030, the stock price would be ~$72, roughly equal to the IPO price. He assigns a 'Milk Road score' of 47, indicating skepticism and no upside at current prices.

  5. 15:41 – 28:10

    Bull Case: Agents and Token Usage

    Kyle argues the market underprices the demand for AI agents, which will drive significantly higher token usage. He cites the removal of Opus 5.5 breaking Grok as evidence of the superiority of frontier models and predicts budgets will expand to support agent-driven workflows.

  6. 28:10 – 31:03

    Compute Costs and Margin Pressure

    The speakers discuss Anthropic's $900 billion in compute commitments through 2030. Kyle notes that Q3 revenue per dollar of compute dropped from 2.08 to 1.72 because Anthropic had to pay higher spot prices for compute from SpaceX, unlike OpenAI which locked in long-term deals earlier.

  7. 31:03 – 32:12

    Margin Pressure and Competitive Costs

    Speaker 2 argues that Anthropic faces a difficult business model where input costs (compute) are rising while output prices (AI services) are dropping due to token creation and price pressure. They note that maintaining a frontier advantage is costly and the gap with competitors like OpenAI is narrowing.

  8. 32:12 – 34:02

    Enterprise Stickiness and Client Base

    Speaker 0 highlights Anthropic's massive enterprise clients like Eli Lilly, Walmart, and Salesforce, noting that switching costs and privacy tools create a sticky customer base. They compare this to Palantir's strategy of deep integration and feature expansion within large organizations.

  9. 34:02 – 35:27

    IPO Pricing and SpaceX Comparison

    Speaker 0 discusses the IPO mechanics, noting 6-9% of shares will trade on day one. They compare this to SpaceX's volatile post-IPO price action, which spiked to a $2-3 trillion valuation before falling back near the IPO price, driven by new business developments like compute expansion.

  10. 35:27 – 37:27

    Investor Base and Short-Term Price Prediction

    Speaker 0 points out that Anthropic shares a similar investor base to SpaceX, including Google, NVIDIA, and specific funds. Speaker 2 predicts a short-term price trajectory similar to SpaceX: an initial enthusiasm-driven spike followed by a sell-off below the IPO price within weeks or months.

  11. 37:27 – 38:58

    Diversification and Future Product Roadmap

    Speaker 0 argues that for Anthropic to reach multi-trillion dollar valuations, it must launch products beyond the model layer, such as curing cancer or other breakthroughs. They express uncertainty about whether Anthropic has a clear roadmap for this diversification post-IPO.

  12. 38:58 – 41:44

    Alternative Investment Strategies via Infrastructure

    Speaker 0 suggests that investors can gain exposure to Anthropic indirectly by buying AI infrastructure assets (GPUs, memory, cloud) which are highly correlated. They recommend using PortfolioOS to identify these correlations and note that infrastructure is currently undervalued compared to the IPO pricing.

  13. 41:44 – 48:05

    AI Adoption vs. Labor Market Impact

    Speaker 2 believes AI adoption is inevitable and bullish for compute infrastructure, predicting significant layoffs and margin increases for companies replacing human workers. Speaker 0 disagrees, arguing that AI agents currently expand human capabilities rather than replace jobs, and that physical AI is needed for major labor market disruption.

  14. 48:05 – 52:12

    Valuation and Trading Opportunity

    Speaker 0 states they would not buy at the IPO price but would consider buying if the stock sold off 20-30% to a $700-800 billion valuation, citing strong growth (44% QoQ) and $80 billion ARR. Speaker 2 acknowledges this as a potential short-term trade given Anthropic's $240 billion cash position and low debt.

Assets discussed

AssetKindStanceContext
Anthropicprivate companymixedcouple of months ago, again, I would say, yeah, well, anthropic is you know the the best, the best model, and they have huge advantage there.
SpaceXprivate companyneutralAnd then they also own 100 billion of SpaceX. So, or close to 100 billion.
NVIDIANVDAequityneutralSpaceX has just said that they're not going to, they're gonna be specific to to NVIDIA, but who knows, that could change.
AlphabetGOOGLequityneutralthink it does get so there's one, it's a great investment, obviously. I think they were they were you know knew that from from very early on. Two, it kind of helps them in case they eat away their business, um, because Gemini has not been super successful yet. And three, I think it brings them closer to um either integrating products or selling them their own products like Google Cloud or their TPUs, etc. et cetera. Kind of like the Microsoft OpenAI deal, how they were able. to, you know, both be a win win situation there. So I I'm assuming that's why. Um and then I mean I don't know when they f their first round as they got in, but they're making a shit ton of money from it. So uh obviously it was a smart move on on Google's point and NVIDIA.
AmazonAMZNequityneutralLike all of them owns them, right? Like Amazon too and I guess like all of them has some stake in the in those frontier lives by now.

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

Claims worth verifying

Anthropic went from $2 billion in quarterly revenue to $17 billion in under a year.

John Gillen ·0:44

How to verify: Check Anthropic's S-1 filing or press releases for quarterly revenue reports for Q4 2026 and Q3 2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's IPO valuation was expected at around $2 trillion two months ago and is now coming in at roughly $1 trillion.

John Gillen ·0:44

How to verify: Review news archives from September 2026 for initial IPO valuation estimates and November 2026 for the final pricing announcement. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

In Q2 2026, Anthropic grew 144% quarter-over-quarter in revenue.

Unidentified speaker (speaker_0) ·6:27

How to verify: Calculate growth rate using Anthropic's reported Q1 2026 and Q2 2026 revenue figures from their financial statements. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Goldman Sachs forecasts Anthropic revenue of $600 billion by 2030, $52 billion by end of 2026, and $136 billion in 2027.

Unidentified speaker (speaker_0) ·7:47

How to verify: Locate the Goldman Sachs IPO prospectus or research report for Anthropic published in November 2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Anthropic's revenue per dollar of compute dropped from 2.08 to 1.72 in Q3 2026.

Unidentified speaker (speaker_0) ·29:12

How to verify: Review Anthropic's Q3 2026 earnings report or investor presentation for compute expenditure and revenue figures. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Anthropic has $900 billion in compute commitments through 2030.

Unidentified speaker (speaker_0) ·28:30

How to verify: Check the S-1 filing or press releases regarding Anthropic's long-term supply agreements with compute providers like SpaceX. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Elon Musk/SpaceX is building 13 gigawatts of compute by the end of next year (2027).

Unidentified speaker (speaker_0) ·29:52

How to verify: Verify SpaceX's infrastructure announcements or Elon Musk's statements regarding compute capacity expansion in 2026-2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

OpenAI has locked into long-term compute deals earlier than Anthropic.

Unidentified speaker (speaker_0) ·29:32

How to verify: Review public announcements or filings from OpenAI and its compute partners (e.g., CoreWeave, Azure) regarding contract timing. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's Opus 5.5 model was removed from Grok, causing it to malfunction.

Unidentified speaker (speaker_0) ·16:01

How to verify: Check X (Twitter) posts or technical updates from xAI regarding the integration and subsequent removal of Anthropic's models. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Anthropic was founded in 2021.

Unidentified speaker (speaker_2) ·3:47

How to verify: Check Anthropic's corporate history or Wikipedia page for founding date. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Anthropic's input costs (compute) are rising while output prices are dropping due to token creation and price pressure.

Unidentified speaker (speaker_2) ·31:03

How to verify: Review Anthropic's S-1 filing for revenue and cost of revenue trends; analyze AI service pricing trends from industry reports (e.g., Synergy Research Group). For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's clients include Eli Lilly, Walmart, and Salesforce.

Unidentified speaker (speaker_0) ·32:12

How to verify: Check Anthropic's S-1 filing for top customer disclosures; verify enterprise partnerships via company press releases or news archives. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

SpaceX released 5% of shares on day one of its IPO.

Unidentified speaker (UNKNOWN) ·34:16

How to verify: Review SpaceX's IPO prospectus or financial news reports from June 2026 regarding share float percentages. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

SpaceX's valuation spiked to $2-3 trillion post-IPO before falling back near the IPO price.

Unidentified speaker (speaker_0) ·34:36

How to verify: Check secondary market valuation reports from platforms like Forge Global or EquityZen for SpaceX's valuation history in 2026-2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Google and NVIDIA are major investors in Anthropic.

Unidentified speaker (UNKNOWN) ·35:27

How to verify: Review Anthropic's S-1 filing for investor disclosures; check SEC Form 13F filings for Google and NVIDIA holdings in Anthropic. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Anthropic has $80 billion in ARR and is growing 44% quarter over quarter.

Unidentified speaker (UNKNOWN) ·48:48

How to verify: Verify against Anthropic's S-1 filing for annual recurring revenue (ARR) and growth rates; cross-reference with financial news reports. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Google owns approximately 17-18% of Anthropic and is investing $10 billion in the IPO.

Unidentified speaker (UNKNOWN) ·46:07

How to verify: Check Anthropic's S-1 filing for Google's pre-IPO stake and IPO investment amount; verify via SEC filings. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Jordi Visor predicts significant layoffs in the market next year due to AI replacement.

Unidentified speaker (speaker_2) ·42:19

How to verify: Search for Jordi Visor's public statements or reports from One Bank (likely referring to a specific analyst or firm) regarding AI-driven layoffs in 2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Anthropic's valuation dropped from $2 trillion to near IPO price after a spike.

Unidentified speaker (speaker_0) ·34:36

How to verify: Review secondary market valuation data for SpaceX (not Anthropic, as the speaker conflates the two in the comparison) from 2026-2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Do your own research

  • Verify Anthropic's S-1 filing for actual revenue breakdowns (API vs. Subscription) and risk factors.
  • Check Goldman Sachs' full research report for the $600 billion 2030 revenue forecast assumptions.
  • Review Anthropic's compute contracts with SpaceX and other providers to assess margin pressure risks.
  • Compare Anthropic's enterprise client list (Eli Lilly, Walmart, Salesforce) against Palantir's for switching cost analysis.
  • Analyze historical post-IPO price action of SpaceX to model potential short-term volatility for Anthropic.
  • Assess AI infrastructure asset correlations using tools like PortfolioOS to evaluate indirect investment options.
  • Monitor Q3 2026 earnings call transcripts for details on compute spot pricing vs. long-term contract advantages.
  • Research the technical capabilities of Opus 5.5 vs. Grok to validate claims about frontier model superiority in agentic workflows.
Full transcriptexpand
0:00
Unidentified speaker (speaker_0): I feel like people have way bigger budgets to spend. They just need better use cases for them. And those better use cases come from two things: smarter models and agents, right? Agents that go and do things for you. Because that's way more token usage. And so I do think that my opinion, anyway, is the market is
0:19
Unidentified speaker (speaker_0): underpricing how much demand there's going to be for that.
0:24
John Gillen: What's up, everybody? It's John Gillen here. Welcome to Milk Road Stocks, the daily market show that swore it would never chase an IPO and has already checked the Anthropic pricing six times today. Today is Wednesday, November 18th. We are recording this on November 12th. Anthropic went from $2 billion in quarterly revenue to $17 billion in under a year.
0:44
John Gillen: Two months ago, it So was expected to IPO at around $2 trillion. Now it is coming in at roughly one. So is that a bargain or is it already? priced in today? Kyle, analyst and CEO here, and Martin, our lead researcher, are going to try to work all that out. Martin's valuation model lands right around the IPO price. Kyle, thanks the market, is underpricing agents here. Interesting take. And they both weigh the compute bill that
1:04
John Gillen: is squeezing anthropics margins. There's a lot to unpack in this, a lot of big numbers, and a lot of math. So this is gonna be a good episode. And stick around to the end of the call. They talk about what happens to the stock after it lists, their take on whether or not they would buy it and at what price, and they share other ways to trade the Anthropic IPO without actually owning it. Martin ran his whole analysis through something called
1:23
John Gillen: Portfolio OS, which is our new tool that analysts built that plugs into your portfolio and it learns your investing style. It brings that right into your AI tool. So it researches like an institutional research desk. It is live now with lifetime access on sale in the link below. Check it out, it's really powerful. It's really cool. I love it. I think you will too, especially if you're an investor. So that sounds good to you. Make sure you
1:43
John Gillen: like and subscribe. Share this episode with somebody who's going to enjoy it. Thank you so much for being here, and we hope you enjoy this episode.
1:49
Unidentified speaker (UNKNOWN): Hey
1:49
Unidentified speaker (speaker_0): everybody, welcome back to another episode on Milk Road Stocks. Today we are talking about. I got Martin and myself, and we're going to gonna talk about Anthropic. Enthropic. The IPO is coming. We are recording this a couple days before. we get the full details around the IPO. Um, but what we're gonna do today is kind of run through how we would value this company, um,
2:09
Unidentified speaker (speaker_0): how we think just about this company, and I guess the sector and the industry in general uh on where things are going and try to make a decision: would we buy this IPO or would we not buy this IPO? We've got a bunch of data that Enthropic Topic has is already released, so we're gonna just use what we have um and and we'll go from there. I will say with a bit. of context, IPOs in the short term. If you're listening to this and
2:29
Unidentified speaker (speaker_0): wondering should you buy today and you want to make money, you know, in a week from now, I don't know that we're gonna solve that for you. Uh who knows what happens when uh when a company of especially of this size uh goes um IPOs uh in the short term, but we'll talk about more, you know, does it make sense that this could be a company you hold in your portfolio uh for months, years,
2:49
Unidentified speaker (speaker_0): et cetera? So Martin, let's kick things off. I'll just share a bit of context. So it looks like Enthropic is going to IPO somewhere around $70 to $78,
2:58
Unidentified speaker (speaker_0): a valuation of around a trillion. We don't know the exact amount amount. Maybe by the time you guys are listening to this, you might know the exact amount, but let's say it's one to 1.25 trillion, I think. Um, which, you know, just like two months ago, this thing was gonna going to be around 2 two trillion. So it's come down quite a bit, even though its growth has continued. to uh, you know, its revenue has
3:18
Unidentified speaker (speaker_0): grown uh for you know the last few months. Um, October was a big month. It looks like we've hit maybe 80 billion-ish AR by the end of October. So, like overall, the business looks pretty good, I would say. I guess just first initial thought, Martin. How do you think about when a company's gonna IPO, especially of this size? How do you think through this process and how
3:38
Unidentified speaker (speaker_0): do you apply it? to your portfolio?
3:40
Unidentified speaker (speaker_2): Yeah, I think we are all very excited and heavy users of Cloud. Though
3:47
Unidentified speaker (speaker_2): let me just say the company was founded in 2021. So it's five years ago. And now, like, you know, next week they're gonna IPO at one trillion plus valuation. That's just super crazy. Um, but yeah, I mean, I'm using cloud. Claude pretty much every day. But if you would ask me a couple months ago,
4:07
Unidentified speaker (speaker_2): I would be like all in on Cloud. And now I'm actually using it less and less as I'm using more grog. So it kind of makes me think about this whole bull case for Cloud because I feel like retail is now having more options. So a lot of people are, you know, using GrogBot Grugbart or Muse or some other new age.
4:26
Unidentified speaker (speaker_2): agents platforms that exist today. And also enterprises that are looking to reduce their costs because they don't want to be just locked in on cloud. And so they build their own internal routers and use open source models on more stuff. And so I was much more bullish on Anthropic a couple of months ago. And now
4:46
Unidentified speaker (speaker_2): I really think that the future is gonna be pretty tough. And for me, I think they should really try to pivot into some niche markets, or as they just launch, you know,
4:59
Unidentified speaker (UNKNOWN): kind
4:59
Unidentified speaker (speaker_2): of Microsoft competitors. So they have like, you know, um Excel, Word, and kind of Microsoft 365. So I think that's their Damn niche. news. They need to f find some some markets where they can really dominate and kind of replace the existing software. But like I don't think that the
5:19
Unidentified speaker (speaker_2): the current model is gonna work for them in the long term as more companies, more enterprises, more options are available. And so there's gonna be pretty ch challenging.
5:29
Unidentified speaker (speaker_0): So you're thinking is the model layer is what's got them here. It's what got them to you know $17 billion in revenue and in Q3 uh and an $80 billion ARR. But there's a lot of competition on the model side. Um, and those kind of prices are getting squeezed as well. Uh, and so now they need to
5:49
Unidentified speaker (speaker_0): move to additional products, which is you're kind of saying, you know, the competition around um, you know, what they're doing with Microsoft. They need to build and go up the the ladder into the application layer is the way that I would say it. Or
5:59
Unidentified speaker (UNKNOWN): at least
6:00
Unidentified speaker (speaker_0): that's one of the things they can do to help. I will say you know, the numbers are pretty crazy. So in Q2, they grew 144% quarter over quarter in their revenue. Let me just actually bring up some charts so the listeners can kind of see what we're what we're talking about here. I'm going to remove this ad here. So here's their quarterly revenue,
6:20
Unidentified speaker (speaker_0): which has been, to say the least, like the most incredible thing we've ever seen for any company in history.
6:27
Unidentified speaker (speaker_0): Uh, literally the fastest revenue ramp in history. Uh, you know, in Q4, 2025, they generated 2 billion. And in Q3, 2026, they've hit 17 billion, right? So these numbers are absolutely crazy. 144% growth in Q2 from Q1. Now it's their growth is decelerating, right?
6:47
Unidentified speaker (speaker_0): It's slowing down. But when I say that, they still grew 44%. and a quarter, right? Uh, from 11 billion base. So, like these numbers are astronomical. So to say it's decelerating is a little bit ridiculous. Um, if you compare the number to OpenAI, right, which would be its main competitor, here's their growth. So 144% Q2,
7:07
Unidentified speaker (speaker_0): 47% in Q3. Open OpenAI AI was 17% in Q2, uh, and then jumped 50%. So actually grew more than Anthropic in Q3. So, but I mean it's just been consistent growth for Anthropic, you know, through this entire thing. Now, a lot of their revenue, I think it's about 80% of their
7:27
Unidentified speaker (speaker_0): revenue comes from kind of pay per usage, so the API fees. Uh, and then 20% comes from just the subscriptions. However, the subscriptions are increasing quarter over quarter. in terms of like its share. So I do think that's kind of important to understand as well. And then if you look at, and now you could take this with a grain of salt, but Goldman Sachs is the one that's running their IPO,
7:47
Unidentified speaker (speaker_0): they're forecasting 600 billion revenue by 2030, which is obviously an absolutely insane jump. They expect 52 billion by the end of 2026 and 136 billion in 2027.
8:07
John Gillen: What if the AI you already use could get an upgrade and turn into a personal investment analyst that knows your portfolio better than you do? Introducing Portfolio OS by Milk Road. Connect your investments, bring everything you own into one place, and immediately get a risk analysis on your portfolio. Then build your investor D. DNA.
8:27
John Gillen: Teach your AI how you invest and how much risk you're taking. Alpha Lab gives your AI the skills to find opportunities, research companies, challenge your assumptions, and evaluate what they're worth. And because it knows what you already own, Portfolio OS can help spot risks or opportunities you might otherwise miss. From your first investment idea to your
8:47
John Gillen: exit plan, Portfolio OS helps you follow a disciplined research process, just like the pros do, but without the big budgets and big teams behind it. This is the future of investing. For Black Friday, get lifetime access to Portfolio OS for just $1.99 plus three months of MilkRoad Pro free. Portfolio OS by
9:07
John Gillen: MilkRode. Your investing process upgraded.
9:14
Unidentified speaker (UNKNOWN): You
9:14
Unidentified speaker (speaker_0): agree with these numbers? Do you think they're actually gonna get there, or is this just absolutely ridiculous?
9:19
Unidentified speaker (speaker_2): I think that's the problem because it feels like this is everyone's base case right now, around 500 billion revenue by 2030. Which to me is quite aggressive, I think. And as I just said, let me actually share this. As you said, Kyle, uh 80% of their revenue is usage-based. usage
9:38
Unidentified speaker (speaker_2): based, so API calls, right? And then only 20% is uh from subscriptions like you know, retail users like VR. And so I think the few the future really for them is mainly in that uh in enterprises, the usage base. And again, I think the future there is
9:58
Unidentified speaker (speaker_2): everyone just wanted to explore and experiment with AI and start. started with cloud or OpenAI. But as you learn more about AI, you know how to use it, then at some point all these enterprises and companies they are gonna start thinking about optimizing the costs. And it will, yes, you might
10:18
Unidentified speaker (speaker_2): be you might be using AI more, but you will be less dependent on cloud specifically. Again, because you know open source models are reaching a very good level of intelligence. And so, again, I would argue that for 90% plus of everyday styles that people do, they don't need cloud at
10:38
Unidentified speaker (speaker_2): all. So you a lot of enterprises are gonna, you know, choose open source models for that. And so again, I keep asking myself what's the what's the reason that you know all the companies and all the people are gonna justify a lot of money being spent on cloud and not optimizing for some
10:57
Unidentified speaker (speaker_2): cheaper alternatives? So that's my concern. And so right now they're gonna do IPO at uh let me share also this chart. As you said, something about one trillion. And right now, if you look at different um exchanges that have pre-IPO. um derivatives, they
11:17
Unidentified speaker (speaker_2): all think that you know enthropic uh value should be higher than 1.2 or trillion, uh, which is I think supposed uh IPO ad. And and so everyone is super bullish. I think the base case here, let me actually share something I have um on Milkload Milk Road Pro here. You do Carol kind of want wanna to talk about
11:37
Unidentified speaker (speaker_2): portfolio as S a bit?
11:38
Unidentified speaker (speaker_0): Portfolio S is the big new product that we've just launched at Milk Road. A lot of the research and things we're talking about here have come from Portfolio OS. Essentially, what this is, is um we've created a tool that becomes a new operating system for your portfolio. And it's essentially an upgrade for your AI. So whether you're using Cloud, GrockBot, you know, chat, whatever, right now, when you ask it, should
11:58
Unidentified speaker (speaker_0): I buy anthropic, you know, at its IPO, it's gonna going to give you some generic answer that. has nothing to do with the way you think about investing or, you know, what your portfolio is, et cetera. And it's also not going to give you a very well thought out kind of institutional grade, you know, research deep dive. It's just going to be very generic and basic. What we've done is create a portfolio S, which essentially
12:18
Unidentified speaker (speaker_0): connects to your portfolio. Um, it connects to your investor DNA, we call it, which is essentially how you think about investing, what are your risks, what are your guidelines, all that kind of stuff.
12:27
Unidentified speaker (speaker_0): which is what Martin's showing here on screen. And then finally, it it basically pulls a
12:33
Unidentified speaker (speaker_0): suite of different skills and tools for your AI to use so that it thinks and researches and provides outputs similar to an institutional grade research desk. But you don't need the millions of dollars to spend on that. You don't need the big team. It's literally all just done inside your AI by connecting portfolio. OS. So it'll do a full research for you. So all the data
12:53
Unidentified speaker (speaker_0): and stuff that we've been talking about, this came from our Portfolio OS um research system. Um, and you guys can all get it right now uh at a huge Black Friday sale. So I highly recommend you check that out. We're gonna show you guys some of this and how it works because what it'll also do is give you a score, it'll tell you if you if it thinks you should buy it based on what you hold in your portfolio and how you invest. So that's the scores that that Martin has here. But
13:13
Unidentified speaker (speaker_0): anyway, walk us through uh you know what what it's kind of helping you see for anthropic right now.
13:19
Unidentified speaker (speaker_2): Yeah, I ran the whole process for anthropic um using you know our skills as Kyle just described. And my issue is right now that if you scroll down all the way to assumptions for the model, so even if my base case, I assume that they are gonna have four hundred forty billion.
13:39
Unidentified speaker (speaker_2): uh revenue uh by twenty thirty, my the the base case right now is still about uh seventy two dollars, which is about IPO. So, you know, what is the upside for me? I'm quite skeptical, as as you can say by now, about the they will not reach all these numbers by
13:59
Unidentified speaker (speaker_2): twenty 2030 thirty unless they will really figure out something else. But the current model is definitely not gonna get them there. So
14:07
Unidentified speaker (speaker_0): why would you say that the price is still valued basically the same as it is today if its revenues go from, you know, let's say it's uh I don't know what it is today, but let's say it's gonna be 60 billion by the end of the year, all the way up to what you're saying is twenty thirty, it's it's four hundred and forty billion. Is that because of CapEx and expenses, or where do you get that from?
14:25
Unidentified speaker (speaker_2): Because it's all priced in already. So everyone is like we are pricing Anthropic at one point three two trillion because we think that by twenty thirty they will reach five hundred billion, which you know again, I don't I'm not so optimistic about that. So right now it's it's it really feels like it's priced quite fairly
14:44
Unidentified speaker (speaker_2): and I I struggle struggled to find some upside. So, you know, I was like my milk road score is below it's just 47. So it's just I might add this outside on my watch list, but I'm not buying at these prices for sure.
15:00
Unidentified speaker (speaker_0): And this score, by the way, is something that Portfolio S will give you. Um it looks at your investor DNA or framework and looks at kind of how you think about investments after you fill out this whole survey. Um and then it kind of uh and looks at your portfolio, then it gives you kind of a score on if you should, you know, look deeper into this, if you should buy it, et cetera. And I'll give you kind of all that, those insights.
15:21
Unidentified speaker (speaker_0): So what's the bull case then? There is a bull case here. What is that? I guess it's just the fact that it has to move off of just being a model layer into something bigger and having serious products and becoming like a SaaS tool essentially and competing with, I don't know, Salesforce and Microsoft. Like, is that the bookcase for you? Or is there something else that
15:41
Unidentified speaker (speaker_0): we're missing? The one thing that's interesting, I'll let you think about that. And I was gonna explain something is you know, a couple weeks ago, Grok turned really stupid and everyone was complaining on X. And we realized the reason it turned stupid and stopped working is because they removed Opus 5.5, which is one of Enthropics models. So we saw how powerful it is to have a frontier model. So
16:01
Unidentified speaker (speaker_0): there's, you know, everyone kind of said, all the models are just a, you know, a commodity. And that really proves that it's actually not true. There is a really big benefit on these frontier models. Um, being ahead by three months, four months, five months, whatever, six months seems to make a really big difference, especially when it comes to agentic use cases, right? Which is basically what broke when they removed Opus
16:21
Unidentified speaker (speaker_0): 5.5.
16:22
Unidentified speaker (UNKNOWN): I
16:22
Unidentified speaker (speaker_0): don't know that the market fully prices in agentic yet.
16:27
Unidentified speaker (speaker_0): because if you look at what the market says, it's a lot of like, you know, budgets are starting to shrink for AI, right? Like, sure, we're still using more tokens, but it's not going to be crazy. You know, you have the ramp data coming out saying that, you know, companies are using less tokens than they were, you know, a couple months ago. And so it feels like there's a lot of people, and a lot of the market thinks, okay, we've sort of peaked
16:47
Unidentified speaker (speaker_0): here in terms of this usage. And now everyone's like either. lowering their budgets or they're getting better at kind of routing these tokens. I really don't think that's the case. I think that's probably the bull case here is I feel like people have way bigger budgets to spend. They just need better use cases for them. And those better UK use cases come from two
17:07
Unidentified speaker (speaker_0): things smarter models and agents, right? Agents that go and do things for you. Because that's way more token use. usage. And so I do think the my opinion, anyway, is the market is underpricing how much demand there's going to be for that once we unlock the use cases for agents. And I mean, you know,
17:27
Unidentified speaker (speaker_0): we're early to it. We've used agents a ton with Crockbot. It now runs everything we do.
17:31
Unidentified speaker (UNKNOWN): And
17:31
Unidentified speaker (speaker_0): you know, Enthropic hasn't even launched that yet. OpenAI, I think, is like still in beta with dots. So, like, we're not even at Enthropic's Entropics also in beta for it. So we're barely even touching the surface of what's possible here. And I just don't know that the market realizes how much we're going to use this stuff and how just useful it really is. And the more useful it gets, the more it can
17:51
Unidentified speaker (speaker_0): replace, you know, people or expand what people can do, which means more budgets can go to it. Um, and so I do think there is that world where the API fees and the subscription uh fees. for these models, especially Frontiers, can go a lot bigger than it is today. Now, I still believe they need other products to go with it, but I do
18:11
Unidentified speaker (speaker_0): think they can get much, much bigger with just API usage and subscriptions. What do you think about that?
18:17
Unidentified speaker (UNKNOWN): A
18:17
Unidentified speaker (speaker_2): couple things. Right now, it really feels that the orchestration is probably the most important thing right now because whatever you do, you just need a very strong market. model that's gonna orchestrate, like you know, whatever you do. So I agree there. Now,
18:32
Unidentified speaker (UNKNOWN): a
18:32
Unidentified speaker (speaker_2): couple of months ago, again, I would say, yeah, well, anthropic is you know the the best, the best model, and they have huge advantage there. But now Musk just said on Twitter that Grok 5.0 is gonna be the same as Opus 5.5, which is what you were just talking about. So Groc will have
18:52
Unidentified speaker (speaker_2): their own orchestration model good enough, so they will not need to rely on cloud. And I also think that we've seen a pretty big comeback from Meta and also Google. So I think they will all realize that they need that model, the or orchestration level to be figured out on their
19:12
Unidentified speaker (speaker_2): own models. And so my main kind of if I should say one thing that will define how I feel about the the uh anthropic from now on is the percentage uh of money spent on AI used on anthropic only is gonna drop
19:32
Unidentified speaker (speaker_2): dramatically inside
19:34
Unidentified speaker (UNKNOWN): the
19:34
Unidentified speaker (speaker_2): companies. They might still use it, but like the bigger budgets will go to else elsewhere, but less less money, less money on the whole pie.
19:43
Unidentified speaker (UNKNOWN): So overall
19:43
Unidentified speaker (speaker_0): budgets will expand, but the share will probably shrink to
19:49
Unidentified speaker (speaker_0): is your guess. Yeah.
19:51
Unidentified speaker (UNKNOWN): I
19:51
Unidentified speaker (speaker_0): I guess I I mean that's exactly what's happened to me and what's happened to our companies in the last little
19:56
Unidentified speaker (UNKNOWN): while.
19:56
Unidentified speaker (speaker_0): Though I would say the pie is expanding so big that even though its share is less, it's still growing in terms of its uh at least in our companies, in terms of its uh budget and spend. So um, but I do agree with you that the competition is pretty strong here. And I also think the one thing that Grok um and Elon and SpaceX did really
20:16
Unidentified speaker (speaker_0): well is, you know, we've talked about this for a long time in the show is that the harness is so important. It's not just about using one model. It's like, can you create a good harness that routes to different models? Um, and and that's what Grokbot has done, which was super, super smart. Anthropic and and OpenAI kind of this, but only within their own subset of models, right? Uh and honestly, it doesn't do that great of a job at it right
20:36
Unidentified speaker (speaker_0): now. You gotta like manually select and stuff. I'm sure with agents, I don't have access to Anthropics agent yet, but I'm sure they will do that. Um, but Grokbot really, really nailed that ahead of them.
20:46
Unidentified speaker (speaker_2): Agree. Yeah. And the next part which makes some people really bullish on anthropic is that they will try to get more kind of software as a business market share. But I'm still not so sure about that either. Like they will obviously get some share, but like if I'm a huge company, enterprise, I don't know, more
21:06
Unidentified speaker (speaker_2): than 10,000 people, we probably use um Salesforce or we use you know Google, and they will probably have they will. also want that market share and they already are integrated very well in that in those companies and it's not just you know anthropic is gonna sell you uh like infrastructure and
21:26
Unidentified speaker (speaker_2): API calls but Salesforce or Google or maybe Google might not but Salesforce is gonna say hey we are gonna train our own AI models so we do the job and we will implement it in your company so it's not like, hey, here is API and just use it, you know, everywhere, or here's our anthropic document documents
21:46
Unidentified speaker (speaker_2): or uh Excel sheet or whatever, use that. No, they are gonna do it for you. So I feel like yes, there's something powerful that if you use an integrate cloud, but like to bring it and like really make sure that the the company on the on the you know broader level is free. really using it effectively. It's a different task and
22:06
Unidentified speaker (speaker_2): it's quite important. And it's not that easy that, hey, you know, we just integrate cloud and that's it. So
22:12
Unidentified speaker (speaker_0): my only thinking is like, I don't know the number of users for Anthropic right now, but it's in the hundreds of millions. OpenAI is is you know a billion people that are using this thing. I think it's monthly active users, maybe it's weekly. I'm not sure. That doesn't just go, I mean it could go just go away, but I can't can imagine that just goes away. quickly. And I I saw a post, I think it was Andreas Steno that put this out. And he basically said
22:32
Unidentified speaker (speaker_0): he sees anthropic as like the the new operating system for the internet, right? So kind of like what Windows 95 was back in 1995, where all of a sudden, you know, there was other ways to use the computer, but then they made this operating system and it was like everyone just use that because it was so much easier. And then all apps came and integrated into that, right? Um,
22:51
Unidentified speaker (speaker_0): and kind of the same way here, right? Like,
22:54
Unidentified speaker (UNKNOWN): as you
22:54
Unidentified speaker (speaker_0): talked about, there's a bunch of apps out there, whether it's Salesforce or anything else. How many of us listening right now use your AI as your main sort of like like I got a screen right here on my left, and it's always just AI, right? That you used to be an internet browser, but now it's it's just my AI. And I actually have both Cloud Claude and GrockBot side by side. And that's how I like check my emails.
23:15
Unidentified speaker (speaker_0): That's how I see what's going on in Slack or in Discord. That's how I like get answers on whatever. And then so it's like all the apps are moving into there. We just launched Portfolio OS. What is that? It's an integration where sure there are beautiful charts and things that you use inside of the our Milk Road platform for Portfolio OS. So you can see all that, but the
23:34
Unidentified speaker (speaker_0): research and the conversation is all getting done inside your AI. So if people adopt Portfolio S, that's another sort of network effect added to these like models being sort of a
23:47
Unidentified speaker (speaker_0): not a browser, but an operating system, right? And so I just feel like whether that comes from more usage or you get revenue there from ads, or there's gotta be other things that they can find ways to monetize if they're gonna have hundreds of millions of users using this every single day. to use all their other applications. Like there's there's a lot of value there. And I don't know exactly how
24:07
Unidentified speaker (speaker_0): they're gonna tap into it. I know OpenAI is doing two things already. Anthropic, I don't think it's doing either of these, which is kind of odd. Um, but they're you know, OpenAI is already at like two billion dollars in uh AR for um for ads, makes sense, especially for consumers. And they're also doing um, they might have shut this down, but they're they're just trying to charge transaction fees for um.
24:27
Unidentified speaker (speaker_0): for anyone that's like buying things through their platform. And with agents, you could definitely see this playing out. So I do think there's gonna be other ways outside of just having to reproduce SaaS, right? Um, there's data that they have that no one else has that they can use or sell. For example, they could get healthcare data in here
24:47
Unidentified speaker (speaker_0): and then use that and sell it to Eli Lilly uh to produce, I don't know. some kind of drug, right? Or some kind of like pharmaceutical. So like there's there's there are things that they can do when you have hundreds of millions of users inside of one platform. And it appears Claude's already got that. So I do think there's some stickiness here. Um the only thing I would say is it's agents
25:07
Unidentified speaker (speaker_0): that are this real sticky part. Chat, I can move around to anyone, to any platform, and it's pretty simple. Um, the connectors. are, you know, do help. That makes it a little bit sticky, but it's it's the agents when you have scheduled runs and all that kind of stuff that make it a lot stickier. And Anthropic doesn't have that yet. So I think that's actually a potential downside for Enthropic.
25:25
Unidentified speaker (speaker_3): It's nine o'clock, the kids are finally asleep, the dishes are done, it's time to chill, or maybe start another project, or think about buying a house, or listen to an awesome podcast. Either way, you've got a few minutes to yourself, and researching something like life insurance is probably the last thing you ever want to spend. those minutes on. So it ends up on that list for another few months and you put it off again. Suriance, a partner of today's
25:45
Unidentified speaker (speaker_3): show, is actually built to fix that. At milkroad.soriance.com, you can request a free, no obligation quote for term life insurance, and the whole process is online. You can compare options from established carriers and talk to a licensed advocate if you'd like a person to walk you through it. And some policies don't even need a medical exam or blood test depending on eligibility. You've already done the hard part, which is building
26:04
Unidentified speaker (speaker_3): a life worth protecting. This is the step that makes sure it's covered. Get a free no obligation quote at milkroad.soriance.com.
26:12
Unidentified speaker (speaker_2): But you know what? It's uh agents are just, you know, very smart, capable employees that work 24-7.
26:19
Unidentified speaker (speaker_2): And if my employee is locked in into cloud or something, I'm not happy about it. And if I ask him to do some task and he's only gonna use cloud models that's gonna cost me two bucks instead of you know 50 cents, I'm not gonna be happy about it. So I I think I'm not sure how much
26:39
Unidentified speaker (speaker_2): welcome there is really. And right now it feels there is, but eventually
26:44
Unidentified speaker (UNKNOWN): the
26:44
Unidentified speaker (speaker_2): winner might be the one who will who's gonna just aggregate like all the models and like you know, bring me the grogbot feel and the performance in a like model agnostic way, and I'm gonna use it and I'm gonna stay there forever. So you know, it's it's not right now, people are
27:03
Unidentified speaker (UNKNOWN): kind
27:03
Unidentified speaker (speaker_2): of prefer, let's say, Microsoft offices because they're used to it, they like the brand, whatever. But now the more we the more we are gonna use more agents, like much more. Like I feel like I have you know a couple employees sitting just right here working all the time. That's just insane. And we are just getting started. I feel
27:23
Unidentified speaker (speaker_2): like we are very early adopters of AI agents. And so imagine what's gonna happen when companies are gonna do the same thing. Um we we can talk about the job market later because I think it's gonna be a big, big challenge in the future as well. But the point is that. agents they do not have, you know, preference or brands or
27:43
Unidentified speaker (speaker_2): relationship and stuff. They don't care about that. Like if I am managing my own agents, I want tasks to be done as as quickly and as cheaply as possible. That's it. And so you know that brings that stickiness a bit away, I think, from all these marvels. formals.
27:57
Unidentified speaker (speaker_0): Let's talk financials for a second here. One thing that you have to think about is in order so let's say let's say the demand is there, right? Because kind of everything we've discussed so far is like, is there going to be demand to continue to use these models? Let's just say it is.
28:10
Unidentified speaker (UNKNOWN): The
28:11
Unidentified speaker (speaker_0): one thing Anthropic needs to service that demand is a lot of compute. And I mean a lot of compute. And so that is one of the big fears in the market is you know, they've got 900, what is it, 900 billion or something in commitments through 2030, something like that. And really that those commitments are all around. bringing more compute in case there's more demand and more token usage, which we I I still
28:30
Unidentified speaker (speaker_0): assume it's gonna go higher. And it sounds like you do assume there is an there is demand. It might not be 144% growth demand every single quarter, but there's definitely growth. Um agreed?
28:42
Unidentified speaker (speaker_2): There is demand for AI. I'm not necessarily saying that
28:45
Unidentified speaker (UNKNOWN): there's Because you know
28:45
Unidentified speaker (speaker_2): as much demand for cloud specifically, but yeah, yeah.
28:48
Unidentified speaker (UNKNOWN): Well,
28:48
Unidentified speaker (speaker_2): they will grow, obviously, but yeah. maybe at a decelerated rate.
28:52
Unidentified speaker (speaker_0): But if we look at what happened here in Q3, their revenue per dollar of compute actually went down. Now it's still great. 1.72 is not a bad thing by any means, but it was 2.08 before that. And the reason it went down is because they had a bunch of demand and they didn't have the compute locked up. So what did they have to do? They had to pay SpaceX at spot price,
29:12
Unidentified speaker (speaker_0): right? Um, and that's a lot more expensive. It's like, you know, maybe 50. I don't know exactly what their number is. They didn't say, but let's say it's 50 million per megawatt uh instead of, you know, long-term deal might be 10, 20 million per megawatt, right? On a on a four or five-year-long contract. So this is one of the things that Enthropic did not do as well on
29:32
Unidentified speaker (speaker_0): OpenAI locked into a lot of deals way early on. They saw this way ahead of Enthropic. I flagged this a long time. ago. Somehow Enthropic ripped by OpenAI all year, anyway, and did well. But now they're kind of seeing shit. Maybe they were a little bit late to this. And this is why Elon, of course, being Elon was all over this. And the guy's gonna go build 13 gigawatts of
29:52
Unidentified speaker (speaker_0): compute by the end of next year. Um, and now he's also the one selling it to them. Is this going to be a big problem for Enthropic? Because the CapEx, my my guess is
30:02
Unidentified speaker (UNKNOWN): especially.
30:03
Unidentified speaker (speaker_0): if the Dems end up slowing down data center buildup, I don't really think they will slow it down much. But even if they do,
30:08
Unidentified speaker (UNKNOWN): even
30:09
Unidentified speaker (speaker_0): if they don't, the demand is so high that the spot price of compute is going to go up and continue to go up from here. And so Enthropic needs to either figure out a way to get some long-term deals or they're going to have to keep paying more and more to service this. How do you think through this?
30:23
Unidentified speaker (speaker_2): Yeah, I think the fact that they have commitments of 900. billion, it's it's just insane. We are talking like, you know, really huge numbers. And um yeah, I think as you said, OpenAI really saw that early and now they're fine. And it's also pretty nasty for
30:43
Unidentified speaker (speaker_2): Anthropic because now what they are doing is they are pretty much converting compute into tokens. And so you have supply and demand. And there is a lot of demand for AI. Like a lot of people we are gonna consume like orders of magnitude more demand more tokens in the future. So the demand is there. And the
31:03
Unidentified speaker (speaker_2): the problem is that you create tokens and there is a lot of price pressure on your tokens. So you probably need to sell your goods at lower prices. And at the same time, your inputs, which is computation costs, are up only. And so, like, you know, your inputs are
31:22
Unidentified speaker (speaker_2): up only, and probably it's gonna be that way for a maybe a couple more quarters. And then your outputs, the the the things you are selling, are actually like dropping prices down all the time. So again, it's pretty tricky business to be in, to be honest. And yes, the margins are good for now, but what
31:42
Unidentified speaker (speaker_2): about in a year or two years? That's gonna be pretty tricky. And don't forget that again, the biggest mode of Anthropic is that they are still frontier, they are ahead of everyone else. Well, OpenAI and Anthropic, they are still ahead. Well, that gap is narrowing, but it's pretty costly to stay above everyone.
32:03
Unidentified speaker (speaker_2): else. You know, you need more compute because the more computer you have, the better models you can produce. And so overall, it's pretty tough business to be honest.
32:12
Unidentified speaker (speaker_0): It's a tough business for sure. However, I would say I don't think that like if you looked at their S1, their clients are massive. Like they're talking, I'm talking about their clients are like Eli Lilly, Walmart, right? Salesforce. Like it's the biggest companies in the world. And they have like 40,000 to 60,000 employees that
32:32
Unidentified speaker (speaker_0): are plugged into cloud, right? And they built all these like risk and privacy and like all these things around it. It's easy for you and I, and even Milk Road as a small business, to like pivot to the next better thing, you know? It is not easy for their clients to do that. So I I I and and they were first to sort of build the tools that
32:51
Unidentified speaker (speaker_0): enterprises need to like manage this across tens of thousands of employees. So I do think that customer base is very sticky and is not going anywhere. Like I don't think Salesforce is gonna go jump and use Groc Bot ever. I think they're gonna use Claude and they're gonna Cloud is gonna build things specific to what that company needs. Same with Walmart, same with the L, etc. I think
33:11
Unidentified speaker (speaker_0): they've got a bunch of these clients, they're just gonna going to be here and I think the whole goal is can they, and so far they have been, continue to build things that keep them around and spending more and more and more. Um, which, like, I don't know, if you look at Palantir earnings, that's exactly what they're doing with their clients. And Anthropic's kind of doing the same thing. They got to go into these big companies and like integrate
33:31
Unidentified speaker (speaker_0): their system in and just like keep building more and more features for them. Um, and I think, you know, they've got a pretty big enough base that they can. they can do that. I don't think as much as Grockbot is cool,
33:43
Unidentified speaker (UNKNOWN): I
33:43
Unidentified speaker (speaker_0): don't think it's going to take as much of I don't think it's as competitive as as we think it is to their their business at the moment. One thing I want to look at is let me just pull up this the the chart of spaces. I think this is really important for the listeners is IPOs are a little bit different. We're talking about the fundamentals and you know what it should be valued at. That doesn't mean that's how it's going to be priced at when we IPO. So when it I
34:02
Unidentified speaker (speaker_0): IPOs, they're releasing 9% of shares, I believe. Let me just double check on that. 6% on a fully diluted basis, 9% of the company can trade on day one. So
34:16
Unidentified speaker (UNKNOWN): this
34:16
Unidentified speaker (speaker_0): is actually kind of similar to SpaceX, right? If you remember, SpaceX released, I think it was 5% Uh that they released on on uh day one. And now the market was a bit different. back when SpaceX launched. I think it was a little bit more bullish. This was um when was it in June? So like the market was pumping in June. It's a whole different story today. But anyway,
34:36
Unidentified speaker (speaker_0): um, you know, it ripped from about what was it, 130 or something at IPO data, and it went all the way up to like 225 over a $2 trillion valuation. Uh, and then um almost $3 trillion, I think. And then yeah, it came flying down.
34:51
Unidentified speaker (speaker_0): uh for the next like two months, down to almost $100. And then now we're basically at right around the IPO price today. So it's kind of interesting. It's been a wild journey on SpaceX, but a lot has changed in that company in since it's IPO'd, right? Like when it IPO'd, no one really thought it was gonna Going be to a neo cloud and in a big compute business that was gonna have 13
35:11
Unidentified speaker (speaker_0): gigawatts in the next like five quarters, right? That wasn't even a conversation. Everyone just thought it was a rocket company that happened to have this like kind of niche AI on the side. And the game has really changed on SpaceX. So if it didn't have all that stuff
35:24
Unidentified speaker (UNKNOWN): in
35:25
Unidentified speaker (speaker_0): its back pocket post-IPO,
35:27
Unidentified speaker (UNKNOWN): I
35:27
Unidentified speaker (speaker_0): think SpaceX would end up being a lot lower, right? And so the question is what happens with On Enthropic? Maybe it runs up again because it kind of gets squeezed because there's only a short amount of shares, and so people. want to buy in on that. Um, I don't know. But it has a similar investor base, right? Like Google, who also bought SpaceX, is buying this NVIDIA as a big investor. Same with
35:47
Unidentified speaker (speaker_0): SpaceX. Uh, a lot of the same funds. Gavin Baker, Brad Gersner, all their funds are the big investors here too. So you kind of have a similar investor base. Um, if you had to speculate, Martin, what do you think happens in the short term for for Enthropic? anthropic? And I know you don't like short term price predictions, but too bad.
36:05
Unidentified speaker (speaker_2): I think it's gonna be similar to SpaceX chart actually. I think like there's gonna be a lot of enthusiasm, like excitement, uh even despite the current market, like probably like I don't know, a couple green candles and then you'll see some sell-off, I think. And so uh in in a couple of weeks or months, I think we'll be well
36:25
Unidentified speaker (speaker_2): below the IPO price.
36:27
Unidentified speaker (speaker_0): But then what really matters is what do they launch? And that's the thing we don't like. SpaceX knew they were gonna do this whole thing with you know, with Grokbot and with um uh with 13 gigawatt. I
36:39
Unidentified speaker (UNKNOWN): mean,
36:39
Unidentified speaker (speaker_0): I don't know if they knew it was gonna be 13 gigawatts, but they knew they were gonna go all in on building compute. They had already started, but that wasn't really in their like IPO, like that's not really the way people thought about it. They kind of saved it. for afterwards.
36:52
Unidentified speaker (UNKNOWN): I
36:52
Unidentified speaker (speaker_0): wonder if Anthropic's trying to do the same thing, right? Like, they still haven't even launched their own agents. I would imagine they've got a lineup of things they're prepared to launch right after this IPO to just pump the market. Um, if not, I I think they're gonna end up being lower a few months down the road from where their IPO is. Otherwise, they're
37:12
Unidentified speaker (speaker_0): gonna have to really start laying on the product launch. launches.
37:15
Unidentified speaker (speaker_2): Do you think that they will be able to uh you know bring some uh key breakthrough that is gonna be worth, I don't know, hundreds of billions or trillions, you know, like cancer treatment or something?
37:27
Unidentified speaker (UNKNOWN): So
37:27
Unidentified speaker (speaker_0): you've heard Dario talk about this, which is
37:30
Unidentified speaker (UNKNOWN): this
37:31
Unidentified speaker (speaker_0): really is the you have to believe that they're gonna do something outside of just the model layer if they're gonna really succeed and they're gonna be multiple trillions in the future, right? And they're gonna hit 600 billion in in in revenue. You have to believe that what Dario's saying is that they're gonna end up curing cancer with their model at some point, or they're gonna do, you know,
37:51
Unidentified speaker (speaker_0): whatever, other ideas he has. You have to believe those things are gonna come true for this company to really do well, you know? Not to not to go up a little bit. Like, sure, if the IPO wanted. at a trillion, can it go to 1.5 trillion, 2 trillion just on being a model layer? Like, sure, I think there's still plenty of room to grow there. But
38:11
Unidentified speaker (speaker_0): to really go from, you know, what are they at 80 billion AR to I don't know, 500 billion AR, they're gonna need different products here. I have no idea Yeah, is my is my uh answer to that. I don't know if they're gonna be able to do that. I I do feel confident they have things. up their sleeve for how they can diversify their revenues and create new
38:31
Unidentified speaker (speaker_0): products for sure. Um I just don't know what it is. I I don't really know how to predict that right now.
38:35
Unidentified speaker (speaker_2): I agree. I think the problem is that the future here is so unpredictable. That's why, you know, I'll probably skip this one and I'll just invest in some other assets where I'm more confident that I can predict their business with more accuracy.
38:50
Unidentified speaker (speaker_0): Is there another way to play anthropic? Is there another way to play the upside of anthropic without buying Anthropic?
38:58
Unidentified speaker (UNKNOWN): I
38:58
Unidentified speaker (speaker_0): mean like I would have said NVIDIA, but uh but like maybe it's just infrastructure. Well this is the other thing actually people should think about. Um when I ran Enthropic through my portfolio OS, um the big thing it kept talking about was hey a lot of your portfolio is already AI infrastructure and Anthropic Enthropic is basically uh a fully correlated asset to that, right? Like all
39:18
Unidentified speaker (speaker_0): of your so if you are someone that owns memory and GPUs and neo clouds, like you you essentially already own the anthropic trade. And Portfolio OS um really laid that out for me and showed me, hey, here's the assets you already hold that are correlated to anthropic, Enthropic, which is really, really cool. Again, if you guys are not using Portfolio OS, S, highly recommend you go and
39:38
Unidentified speaker (speaker_0): download. this. Um, there's a huge sale on it right now. So check it out. You just connect your portfolio right away. You can ask it about Enthropic, and it'll just give you your own personalized answer. It's insane. But it talks a lot about this because a lot of my assets are already in the infrastructure infrastructure trade. And this really is the thing that's been guiding the infrastructure trade anyway, right? So, like if you believe
39:58
Unidentified speaker (speaker_0): people are going to use more AI, you could just buy infra.
40:01
Unidentified speaker (UNKNOWN): A
40:01
Unidentified speaker (speaker_0): lot of that is super united. undervalued right now. And you you kind of have an idea of its fundamentals and its price valuations, a little bit different than an IPO where you're like, I have no idea if this thing's priced right or not. Um, or you could buy application layer stuff, companies that are actually using Anthropic, Salesforce, you know, maybe AppLovin, uh uh Palantir,
40:21
Unidentified speaker (speaker_0): like these kind of things is another way to sort of get exposure to AI, but it it's kind of model agnostic. The other thing you do is buy a basket of these things, Enthropic, Google, and SpaceX. And now you've got exposure to all the top models.
40:35
Unidentified speaker (speaker_2): You know, the kind of AI kind of pyramid where you have like different layers. I
40:40
Unidentified speaker (UNKNOWN): think
40:41
Unidentified speaker (speaker_2): the model layer is really the the most kind of challenging one. And especially if you are someone like Anthropic, who is starting pretty much from scratch, and you're your only kind of
40:52
Unidentified speaker (speaker_2): advantage is you know the the frontier you that you like the performance of your models, which like everyone else, like hyperscalers at some point will probably reach very, very similar performance, if you like. So yeah, I'm I'm very bullish on like all the layers below, which is mainly Only infrastructure because as I said, I'm super
41:11
Unidentified speaker (speaker_2): bullish on AI adoption. I I think it's really hard not to be. We are all using it like you know, we are just talking about Rogue Bot getting nerfed and using something else. And I was like, what am I what
41:23
Unidentified speaker (UNKNOWN): should I do today?
41:24
Unidentified speaker (speaker_2): Like I feel like my employees are drunk, or what what's going on? You know? Um so yeah, I I can see how dependent I am on on these technologies. And so I think like the more people are gonna be the same, you know, uh pretty soon. And so very bullish on AI compute
41:44
Unidentified speaker (speaker_2): overall and so infrastructure place are a pretty pretty safe place right now. Yeah.
41:49
Unidentified speaker (speaker_0): Because
41:49
Unidentified speaker (UNKNOWN): they are
41:49
Unidentified speaker (speaker_2): distressed a lot. Unless
41:51
Unidentified speaker (UNKNOWN): one thing I want you to talk about
41:52
Unidentified speaker (speaker_0): cheating on their wife.
41:55
Unidentified speaker (speaker_0): Yeah. It's
41:56
Unidentified speaker (UNKNOWN): pretty
41:56
Unidentified speaker (speaker_2): costly.
41:59
Unidentified speaker (UNKNOWN): I
41:59
Unidentified speaker (speaker_2): still think that the market is not talking enough about because you know they have commitments, 900 billion, and it's just one company. So where is all this money coming from? So at some point, I think Jordi Visor said he he thinks next year we are gonna hit the time when we are gonna see a lot of
42:19
Unidentified speaker (speaker_2): layoffs on the market because people are just gonna be replaced by AI agents. And
42:25
Unidentified speaker (UNKNOWN): I
42:26
Unidentified speaker (speaker_2): I cannot see a future where this is not happening. I'm not, I don't think that AI will create enough jobs to offset all these layoffs because I, you know, I can see companies with tens of thousands of people who are just like resending emails and do like really nothing. And so I think there's a ton
42:46
Unidentified speaker (speaker_2): of room for companies to like increase their margins by just using. AI and replacing
42:52
Unidentified speaker (UNKNOWN): is this a bull case
42:53
Unidentified speaker (speaker_0): philanthropic though or a bear case?
42:55
Unidentified speaker (speaker_2): It's bull case for AI, it's bear case for the economy.
42:59
Unidentified speaker (speaker_0): Right. But so is that a bull case for the price of of these models? Like it so all I'm saying is like if
43:07
Unidentified speaker (UNKNOWN): that
43:07
Unidentified speaker (speaker_0): happens, these companies have to be doing extremely well for that to happen, right? Um now I understand it means a lot of people are out of a job, so that that kind of makes the market a bit murky, but like
43:17
Unidentified speaker (UNKNOWN): the
43:17
Unidentified speaker (speaker_0): money still exists somewhere if if there's all like if they're they're letting go of people, it means these companies are making revenue enough that they can use more and more AI. So like it's not like the money all just disappears. It's gotta be somewhere. And it that in that case, it's gotta be going to these model layers and they're making a shit ton of money in that world.
43:35
Unidentified speaker (speaker_2): What do you think that the market direction is gonna be on when this starts? happening? And like we see uh, you know, pretty big playoffs and like
43:43
Unidentified speaker (UNKNOWN): so. This is where
43:43
Unidentified speaker (speaker_0): we sort of diverge. I don't I don't think this is happening next year by any means. Um, I don't think agents are going to be at that level. Agents are great and they can do a lot for you if you know how to use them and if you're good enough. But I think what I see right now at least is that agents just expand your capabilities as a human. Kind of like
44:03
Unidentified speaker (speaker_0): the internet expanded your capabilities as a human, and that unlocked a ton more jobs and a ton more prosperity and a ton more opportunities. I think that's going to happen here as well. I think we don't see a big labor shortage until we get physical AI into the world. That's when I think things become a big problem. Because the more that AI gets adopted, the
44:23
Unidentified speaker (speaker_0): more jobs there are in the physical world to do so many other things. And honestly, the more jobs there are in general, because I just think humans become way more valuable if they use agents. So I don't see a world where we go into high unemployment anytime soon.
44:36
Unidentified speaker (UNKNOWN): And
44:36
Unidentified speaker (speaker_0): even so far, like employment has been super strong as a result of AI. Um, I think that continues for multiple years, um, to be completely honest. So uh I'm I'm not a bear on the on the job market from from AI. Not today. Not today. I I am.
44:51
Unidentified speaker (UNKNOWN): I
44:51
Unidentified speaker (speaker_0): don't know at what point in the future, but it but but not today.
44:54
Unidentified speaker (speaker_2): One last question. I think twenty twenty six is gonna be a year where there is less people uh working. Like when you we will look at uh SNP employees count at the end of the year, it will be lower than a year ago. And so at that point
45:11
Unidentified speaker (UNKNOWN): by the end of twenty
45:11
Unidentified speaker (speaker_0): 2026?
45:12
Unidentified speaker (UNKNOWN): twenty six?
45:12
Unidentified speaker (speaker_2): Yeah, yeah. Are
45:13
Unidentified speaker (speaker_0): we tracking that right now? I don't think we are.
45:15
Unidentified speaker (speaker_2): No, I think um one bank does the survey like every every year. So we we are gonna get those numbers, I think, in in February next year.
45:24
John Gillen: Interesting.
45:25
Unidentified speaker (speaker_0): Okay.
45:25
Unidentified speaker (UNKNOWN): What are you
45:26
Unidentified speaker (speaker_0): doing with ZPO? You're just setting it up fully? Do you have like a potential um price that you would buy it at if it gets sold off like crazy? Kind of like you know, a lot of people are like, Oh, SpaceX, if it goes down to a hundred bucks, I'm buying or below a hundred bucks, I'm buying. No, a lot of people didn't get that opportunity. Do you have a number of mine?
45:43
Unidentified speaker (speaker_2): I do not because I don't have again the confidence that I'm I know
45:49
Unidentified speaker (UNKNOWN): my
45:49
Unidentified speaker (speaker_2): prediction about Anthropic's future are, you know, accurate. So like when I don't feel that confidence, I I just I'm not gonna touch it. And I think a lot of it is price priced already. And even though if the price drops like twenty 20%, percent, it's it still will not offer me enough offside for me to get in.
46:07
Unidentified speaker (UNKNOWN): You
46:07
Unidentified speaker (speaker_0): can also get exposure on Anthropic by just buying Google. They own 20%, almost 20% of Anthropic. So they own 17 and they're also putting 10 billion into this IPO. Uh so they're gonna own, like, I don't know, 18% or something of Anthropic. And then they also own 100 billion of SpaceX. So, or close to 100 billion.
46:25
Unidentified speaker (speaker_2): Why do you think they are doing all these things? investments? Is it just because they has to because like they have to because they have a lot of cash and they just need kind of to be diversified and they need stakes in in those potentially future details.
46:37
Unidentified speaker (UNKNOWN): I mean, it's very
46:38
Unidentified speaker (speaker_0): clear that hyperscalers know that AI is is everything. That's obvious. So they're spending trillions of dollars a year on CapEx. Uh and
46:47
Unidentified speaker (UNKNOWN): um,
46:47
Unidentified speaker (speaker_0): and then many of them are also trying to build their own models and compete in this world. So like we know that we know that AI is is everything in the future. And so I think it doesn't hurt for them to diversify, right? Um, I think it probably got them in the door and and you know that they get to sell them TPUs now to to Enthropic and to um potentially to
47:07
Unidentified speaker (speaker_0): SpaceX. I don't know if they will. Uh SpaceX has just said that they're not going to, they're gonna be specific to to NVIDIA, but who knows, that could change. Um and so
47:16
Unidentified speaker (UNKNOWN): I
47:16
Unidentified speaker (speaker_0): think it does get so there's one, it's a great investment, obviously. I think they were they were you know knew that from from very early on. Two, it kind of helps them in case they eat away their business, um, because Gemini has not been super successful yet. And three, I think it brings them closer to um either integrating products or selling them their own products like Google Cloud or their TPUs,
47:36
Unidentified speaker (speaker_0): etc. et cetera. Kind of like the Microsoft OpenAI deal, how they were able. to, you know, both be a win win situation there. So I I'm assuming that's why. Um and then I mean I don't know when they f their first round as they got in, but they're making a shit ton of money from it. So uh obviously it was a smart move on on Google's point and NVIDIA. They were early to it as well, I would
47:53
Unidentified speaker (UNKNOWN): say.
47:53
Unidentified speaker (speaker_0): So
47:54
Unidentified speaker (speaker_2): like all of them owns them, right? Like Amazon too and I guess like all of them has some stake in the in those frontier lives by now.
48:01
Unidentified speaker (speaker_0): Yeah. Yeah yeah a 100%. hundred percent. Are
48:03
Unidentified speaker (speaker_2): you a buyer at the IPO or?
48:05
Unidentified speaker (UNKNOWN): I
48:05
Unidentified speaker (speaker_0): don't think I'll buy at the day of the IPO. Um, I didn't with SpaceX either, even though I was super bullish. Um, I what I'd like to see is what does the market do right afterwards? If it sells it off irrationally, because we're kind of in a bad market right now, as we speak, like, yeah, sure, I would buy. I do think, you know, like if this thing
48:25
Unidentified speaker (speaker_0): sells off 20, 30 percent.
48:26
Unidentified speaker (UNKNOWN): I
48:27
Unidentified speaker (speaker_0): think it's pretty crazy to have anthropic below a trillion dollars. You know, like I was I was nowhere near buying it at two trillion. Nowhere near. That makes no sense, right? But at a trillion, it looks a lot better. And so if that does sell off,
48:40
Unidentified speaker (UNKNOWN): I
48:40
Unidentified speaker (speaker_0): I I even if it's just a short-term trade for the next year, um, you know, I think if this thing goes 800 billion, 700 billion,
48:48
Unidentified speaker (UNKNOWN): I
48:48
Unidentified speaker (speaker_0): don't know how you can't buy that thing when it's growing 44%.
48:51
Unidentified speaker (speaker_0): quarter over quarter and it's already at 80 billion in ARR. It's gonna hit 100 to 120 by the end of this year. Like that to me seems a little crazy, uh, and feels like you gotta buy that and just take the chance, you know. Only a few things need to go right for them to go and get to two trillion dollars instead.
49:08
Unidentified speaker (speaker_2): It's actually a good point as a trade, not as an investment. So when there's a huge sell-off. And also they have, you know, they have. something they will have something about 240 billion in cash so you know when it's company
49:21
Unidentified speaker (UNKNOWN): right
49:22
Unidentified speaker (speaker_2): they do not have pretty much any debt so if it's 1 trillion you you reduce the cash so you're buying company 700 eight eight hundred billion which makes sense yeah and like from that perspective I can see like some short term upside and like if there are some new things that they will regain accelerate their revenues,
49:42
Unidentified speaker (speaker_2): that's a a nice call option. So maybe yeah, it's interesting.
49:46
Unidentified speaker (speaker_0): So what I could see playing out here is I don't know if the spike happens like what happens in SpaceX. It usually happens on an IPO, you know, retail buys it up. Market's a bit ugly right now, though, so that might not happen. But let's say it does. Then I think it does sell off. Um and then they'll probably just start pumping out new products, new features, new things. And I think that's gonna drive things further into next year.
50:06
Unidentified speaker (speaker_0): And I and I think it catches a bid as like the market gets a little more bullish. Um, so I think there's a trade up play here. Um, but
50:14
Unidentified speaker (UNKNOWN): you know,
50:14
Unidentified speaker (speaker_0): it's a lot better that they're doing this at a trillion. If they were into this at two trillion, I was like, oh, this is a disaster. Um, so I'm I'm glad they pivoted to that anyway. Cool. Well, good good session, Martin. Thanks for the insights. That was a lot of fun. For everyone else out there, check out Portfolio OS. It's gone live today. Um, you can get lifetime access to it.
50:34
Unidentified speaker (speaker_0): And I'm telling you, this is gonna be the best tool that you have. It is literally like an institutional research desk in your back pocket. You're gonna upgrade your AI. Uh, it's gonna give you personalized insights on your portfolio. And you know, kind of like the conversation Martin just had, you're gonna be able to actually have that with your AI, but it's gonna know the context on what you hold. uh and how you think about investing. So it can give you real good answers instead
50:54
Unidentified speaker (speaker_0): of the generic crap you get right now when you use your AI. So check out Portfolio S. Uh, it was built by the analysts here at Milk Road. We use it every single day. Um, it's how we get our research, and it's been an absolute incredible unlock to save a shit ton of time, right? And honestly, for a lot of people, you probably don't know how to do a valuation framework and come up with a bare bull and base case
51:14
Unidentified speaker (speaker_0): for. assets that you're looking at. This will do it all for you. Right. So it it basically takes you no time and you get to do a full research on any asset you want. It's pretty cool.
51:24
Unidentified speaker (UNKNOWN): Um,
51:24
Unidentified speaker (speaker_0): so check it out. And uh yeah, let us know if you guys like it in the comments. Um and uh yeah, that's it. Good luck, everyone. And if you're gonna buy the IPO, I hope it works out for
51:35
Unidentified speaker (UNKNOWN): you.
51:37
Unidentified speaker (UNKNOWN): Yeah,
51:37
Unidentified speaker (speaker_2): me too. Thanks, guys.
51:39
Unidentified speaker (speaker_0): Thanks, everyone. Take care.
51:40
Unidentified speaker (speaker_3): Thanks for listening to Milk Road. If you enjoyed the show, make sure you like and subscribe. And if you're struggling to find winners in the market, that's exactly what Milk Road Pro is built for. Our analysts have called some of the biggest winners early, and Pro lets you see what they're buying next, every trade they make, and the research behind every position. Check out Milk Road Pro at the link below. Everything you hear on Milk Road is
52:00
Unidentified speaker (speaker_3): for informational purposes only. These are our personal opinions. not financial advice. And we may own some of the investments we talk about. Always do your own research and make the decisions that are right for you. See you next time.
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.

More from Milk Road Stocks

Milk Road Stocks

Nvidia Is Down 30%. How Much Was Jensen Huang Worth?

Episode discussion: Nvidia Stock Drops 30% After Jensen Huang's Departure Amid Political Pressure

Speakers’ claims: John Gillen and Vincent discuss Nvidia's 30% stock decline and $1 trillion market cap loss following Jensen Huang's departure as CEO after 33 years. Vincent attributes the exit to political pressure from both Republican and Democratic officials regarding Huang's ties to Alice Huang, framing the leadership change as collateral damage in pre-midterm politics. The speakers analyze the operational risks of losing a founder-led CEO, comparing Nvidia's structure to Tesla's dependency on Elon Musk, particularly regarding innovation speed and pricing power. While acknowledging the leadership vacuum, Vincent cites analyst Tom Lee's argument that Nvidia's deep engineering talent and asset-light model support a 'buy the dip' thesis. The episode also covers geopolitical risks, including the RACE Act's impact on Chinese chip innovation and the potential for a rival AI stack.

NvidiaJensen HuangAI Infrastructure
34 min1 assets8 claims to verify
Milk Road Stocks

One Earnings Report Just Flipped the Entire AI Stock Market

Episode discussion: CoreWeave earnings flip AI market sentiment from survival to growth

Speakers’ claims: Speaker 0 reports that CoreWeave's Q3 results, featuring a 3% revenue beat and significant margin improvements, shifted market narrative from cash burn concerns to growth potential. The speaker highlights CoreWeave's $132.5 billion backlog and doubled contract pricing, noting that switched-on power, not demand, is the primary bottleneck. Bearish concerns regarding $46.9 billion in debt and $7.5 billion quarterly cash burn are acknowledged but offset by management's 2028 profitability target and customer prepayments. The earnings triggered rallies in AI infrastructure stocks like Nebius, IREN, and Oracle, with the speaker projecting strong results for Nebius. The episode concludes with anticipation for Anthropic's IPO and a disclaimer that the content is not financial advice.

AI InfrastructureCloud ComputingSemiconductors
38 min2 assets14 claims to verify
All insights