Rational Dissent

Anthropic's record IPO and AI vendor financing dynamics discussed

33 min Listen to the original Episode page
IPOArtificial IntelligenceNVIDIAVenture CapitalMarket DynamicsRegulation

Summary

Alex Melitas introduces Anthropic's historic IPO, highlighting revenue growth from $41 million in 2023 to an $81 billion annualized run rate by October 2026. Greg Baja analyzes the IPO mechanics, noting a potential 16 billion share lock-up cliff and citing Federal Reserve data on positive net equity issuance. Jim Bianco argues AI demand is in its infancy, comparing current skepticism to early internet doubts, while discussing risks from NVIDIA's leadership changes and local political opposition to data centers.

Key takeaways

  1. 1
    Jay Ritter's data indicates that while IPOs typically pop on day one, they often underperform over the ensuing three years, as founders/CEOs time offerings when valuations are highest.

    Unidentified speaker (speaker_2) · 9:27

Episode sections

  1. Speaker_0 discusses vendor financing as a historical norm for new technologies, noting that while risky, it is not new, though the scale of AI financing is unprecedented. The segment includes a sponsor read for WisdomTree.

  2. 1:30 – 4:40

    Anthropic’s Record IPO & Explosive Revenue

    Alex Melitas introduces the episode, noting Anthropic's IPO as the largest in history. He details Anthropic's revenue growth from $41 million in 2023 to an $81 billion annualized run rate by October 2026, comparing it favorably to OpenAI's revenue and customer spending.

  3. 4:40 – 11:49

    The 16-Billion-Share Lock-Up Cliff & IPO Market Risks

    Greg Baja analyzes the IPO mechanics, highlighting that while only ~1.1 billion shares were available on day one, up to 16 billion shares could become available for insider sale over the next year. He cites Federal Reserve data showing net equity issuance turning positive due to large IPOs/secondaries outweighing buybacks, and Jay Ritter's data showing IPOs often underperform over three years despite strong first-day pops.

  4. 11:49 – 16:15

    Why AI Demand Is Still in Its Infancy

    Jim Bianco argues AI is the most transformative technology in a lifetime, capable of radically changing workflows. He contends that current AI usage is primitive (like a search engine) compared to potential workflow automation, implying demand is orders of magnitude greater than current revenue suggests.

  5. 16:15 – 18:32

    Jensen Huang’s Exit & The 'Central Bank of AI'

    Bianco identifies Jensen Huang's resignation from NVIDIA as a potential risk to the 'circular financing' model that funded AI build-outs. He notes the board has rejected his return, but speculation remains if no legal wrongdoing is found. Baja counters that the financing model is replicable by other executives.

  6. 18:32 – 23:03

    Data Center Pushback & Neo Cloud Financing Crises

    The speakers discuss the 'blue wave' election results leading to local opposition against data centers. They note that 'neo-cloud' companies face financing crises as NVIDIA's vendor financing dries up with Huang's departure, and mention President Trump's stance against building data centers outside the US.

  7. 23:03 – 27:52

    How AI Labs Can Win Over Local Communities

    Bianco suggests AI companies must overcome public fear of job loss by offering tangible community benefits (e.g., new fire trucks, road repairs, teacher raises) instead of using NDAs. He compares the current backlash to early skepticism of the internet.

  8. 27:52 – 33:04

    What’s Next for NVIDIA & Vendor Financing

    Baja believes NVIDIA can replicate Huang's financing role. Bianco adds that Huang could potentially move to Elon Musk's new chip company, Terafab, to continue being the 'central bank of AI.' He reiterates that vendor financing is a standard, albeit risky, part of new technology cycles.

Assets discussed

AssetKindStanceContext
NVIDIANVDAequityneutralThe first one is general. Janssen Wong resigning from NVIDIA. NVIDIA is more than just the largest company in the world and the largest chipmaker in the world.
Meta PlatformsMETAequityneutralwhen you know SpaceX can only build so fast meta can only build um you know Colossus or um I I can't I can't exactly remember the the meta's uh version but you
TeslaTSLAequityneutralTerafab is basically a chipbuilding company for SpaceX and Tesla and Optimus Robots and XAI and the like.

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

Claims worth verifying

Anthropic's revenue was $41 million in 2023, $4.5 billion in 2025, and an $81 billion annualized run rate by October 2026.

Alex Melitas ·2:35

How to verify: Anthropic S-1 filing or Q3 2026 earnings report; check SEC EDGAR for filing date and revenue figures. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

Anthropic had ~$17 billion in Q3 revenue and $33 billion in the first nine months of 2026.

Unidentified speaker (UNKNOWN) ·3:11

How to verify: Anthropic Q3 2026 earnings report or S-1 filing; verify via SEC EDGAR. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

Anthropic's Q3 revenue was ~$17 billion vs OpenAI's ~$10 billion; Anthropic run rate ~$81B vs OpenAI ~$50B.

Unidentified speaker (UNKNOWN) ·3:37

How to verify: Anthropic S-1 filing; OpenAI financial disclosures or credible third-party estimates; compare revenue recognition methods. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

Over 1,900 companies paid Anthropic >$1M and ~250 paid >$10M in the last year.

Unidentified speaker (UNKNOWN) ·4:02

How to verify: Anthropic S-1 filing; check 'Customer Concentration' or 'Revenue' sections in SEC EDGAR. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

~1.1 billion shares were available for sale on IPO day; up to 16 billion shares available for insider sale over the next year.

Unidentified speaker (UNKNOWN) ·5:22

How to verify: Anthropic S-1 filing; check 'Lock-up agreements' and 'Underwriting' sections in SEC EDGAR. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

SpaceX IPO lockup was 6 months; Anthropic's is 1 year.

Unidentified speaker (speaker_2) ·6:21

How to verify: SpaceX and Anthropic S-1 filings; compare lock-up periods in SEC EDGAR. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

Federal Reserve Flow of Funds data showed positive net equity issuance in Q2 2026.

Unidentified speaker (UNKNOWN) ·7:01

How to verify: Federal Reserve H.15 report or Flow of Funds data for Q2 2026; verify via federalreserve.gov. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Jay Ritter's data shows IPOs have positive day-one returns but negative 3-year returns.

Unidentified speaker (speaker_2) ·8:54

How to verify: Jay Ritter's IPO database at University of Florida; verify via gbrw.com or academic publications. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Elon Musk started Terafab in March 2026 as a chip-building company.

Unidentified speaker (speaker_3) ·30:15

How to verify: Terafab press releases or SEC filings (if any); verify via company website or news archives. For recent events, verify 2026-08-20 through 2026-11-20. Use another period only when the speaker explicitly names it.

Primary source

Do your own research

  • Verify Anthropic's revenue figures and run rate calculations against their S-1 filing or latest earnings reports.
  • Check Federal Reserve Flow of Funds data for Q2 2026 net equity issuance trends.
  • Review Jay Ritter's IPO performance studies for historical underperformance metrics.
  • Assess the impact of local zoning laws and political opposition on data center construction timelines.
  • Monitor NVIDIA's vendor financing disclosures and leadership stability.
  • Compare Anthropic's enterprise customer count and spend against OpenAI's public disclosures.
  • Investigate the 'neo-cloud' financing landscape and alternative funding sources for AI labs.
  • Look for conflicts of interest regarding speakers' affiliations with AI companies or investors.
Full transcriptexpand

0:00Unidentified speaker (speaker_0): That's why you get vendor financing to the degree that you get when you have a new technology. You don't go knocking at Jamie Diamond's door. He doesn't understand what it is. Uh and you know eventually he will, and eventually the you will get to them. But in the early stages, the vendor financing thing is very normal. Now that I'm the I'm not saying it's not risky, it may not work out, but everybody's act acting like this is something new. It's not, it's as old as time. what we're doing. The scale obviously is larger than we've ever seen, but everything associated with AI is the largest scale we've ever seen. Today's episode is sponsored by Wisdom Tree. Core fixed income does not mean static fixed income. WTBN, the Wisdom Tree Bianco, Total. Return Fund, brings the fixed income research of me, Jim Bianco, and Bianco Research into a strategy that can reposition across duration, yield curve, credit, and

1:01other areas of the bond market as conditions evolve. For a more dynamic approach to core fixed income, visit wisdomtree.comslash Wisdomtree.com slash WTBN to learn more. Investing involves risk. Principal loss is possible. Before investing, carefully consider a fund's objective risks, charges, and expenses, along with other information contained in the perspective, available at wisdomtree.com/slash US. Read it carefully.

1:30Alex Melitas: Everyone, welcome back to the Rational Dec Podcast. Every week we try to counter consensus thinking while we discuss the world of markets and finance. I'm your host, Alex Melitas, joined by my colleagues, Greg Baja and Jim Bianco, founder and president of Bianco Research. Yesterday, Anthropic went public in the largest IPO in history.

1:50Unidentified speaker (UNKNOWN): That

1:50Alex Melitas: sounds like a record that ought to stand around for a while, but SpaceX held it all the way back in June and it didn't last too long.

1:58Unidentified speaker (UNKNOWN): So

1:58Alex Melitas: when we say this thing is big, what do we mean? Are we talking about the business? Anthropics built? What investors think it's worth, or how much stock actually came to the market yesterday? Those three, those are three different things, but the IPO headlines tend to roll them all into one.

2:15Unidentified speaker (UNKNOWN): So

2:15Alex Melitas: I'm gonna start with the company itself, then Greg's gonna put us put this IPO into context, and that should give us a good base for Jim to comment on what this all means for markets and the AI build-out. So start with the business. I want to set the valuation uh aside for the minute for a minute because the more useful question is just how much money this company actually brings in.

2:38Unidentified speaker (UNKNOWN): And

2:38Alex Melitas: the growth here is pretty hard to wrap your head around. Back in 2023, Anthropic did about 41 million dollars in revenue for the entire year. Million. By 2025, that was up to roughly four and a half billion. Billion with a B. And by the end of this October, just one month ago, the company said it was running at an annualized pace of $81 billion. Now that's a run rate, so it's not money they've actually collected yet, but it gives you the sense of how fast this thing has really accelerated.

3:11Unidentified speaker (UNKNOWN): And

3:11Alex Melitas: if you look at what they've actually booked, it's still enormous. They brought in about $17 billion in the third quarter and $33 billion over the first nine months of this year. And the run rate went up about $10 billion in just the month of October.

3:30Unidentified speaker (UNKNOWN): So

3:30Alex Melitas: that's how much they're bringing in. The next question is: how does it stack up against everybody else? And the obvious place to start is OpenAI.

3:37Unidentified speaker (UNKNOWN): On the

3:38Alex Melitas: headline number, numbers, Anthropic's now the bigger company. It did about 17 in the third quarter compared to roughly 10 for and for OpenAI, and its run rate, like I said, 81 versus something closer to 50 for OpenAI. Now I take that comparison with a grain of salt because they they report revenue a little bit differently.

3:56Unidentified speaker (UNKNOWN): But

3:57Alex Melitas: what I find more telling is how much individual customers are spending.

4:02Unidentified speaker (UNKNOWN): The

4:02Alex Melitas: filing says that more than 1900 companies paid Anthropic over a million dollars in the last year, and about 250 paid them more than 10 million dollars. When that many businesses are writing checks that big, they've clearly moved past just trying it

4:20Unidentified speaker (UNKNOWN): out. So

4:21Alex Melitas: when we say the business is big, that's what we mean. A few years ago, it was barely a business at all. Today, it's out-earning a large number of SP 500 companies.

4:32Unidentified speaker (UNKNOWN): Now,

4:32Alex Melitas: Anthropic didn't put anywhere near the whole company up for sale yesterday, and that matters for how the stock trades in the future. So, Greg, how did the IPO actually shake

4:42Unidentified speaker (UNKNOWN): out? Well,

4:44Unidentified speaker (speaker_2): I mean, if you're just looking at it from you know day one poopla and the PR that a company tries to get, probably mission accomplished. The stock was up what 17% on the day. Um, but I'm I'm kind of going to look a little bit more at what this means like moving forward in terms of you know, you just brought up a small part of their company was actually you know available for sale yesterday during the IPO. So let's look at you know their uh lockup schedule and see exactly how many shares. over the next year is how long this uh run rate goes that they're going to be uh allowing shares to be sold by insiders.

5:22Unidentified speaker (UNKNOWN): So the

5:23Unidentified speaker (speaker_2): chart I'm gonna look at here is if you look at the far left little bar, that was the 800, a little bit over 800 million shares that were available for IPO day. There were also a few insider shares that were available for sale. So in total, it's about 1.1 billion shares that could. have been sold on that day.

5:41Unidentified speaker (UNKNOWN): Now,

5:41Unidentified speaker (speaker_2): if you look at the subsequent bars, you know, you see this when fiscal year full full fiscal year results for this year are released. Up to 4 billion shares are going to be available for the insiders to sell. Uh going all the way up to a year from now, over 16 billion shares will be available for sale. Now that's you know, available for sale doesn't mean that everyone's going to actually gonna actually sell their shares. The founders are, you know, unlikely to sell their entire, you know, stake. And so it just kind of gives you an idea. But the point here being that yesterday was 800 million shares, a little bit over that, available in the IPO, whereas we're gonna see 16 billion shares coming on market throughout the next year.

6:21Unidentified speaker (UNKNOWN): If

6:22Unidentified speaker (speaker_2): you compare this to what SpaceX did back in June, uh they had a similarly kind of intricate, I guess, uh lockup schedule, but theirs, I believe, only goes six months out. And so this is a longer, you know, timeline for all the founder shares to be available for sale.

6:38Unidentified speaker (UNKNOWN): But

6:38Unidentified speaker (speaker_2): the reason this is important, as you alluded to Alex, is that in the past, you know, this isn't different from how other companies do it. You release a small portion of your company stock and then you have to kind of wait during this lockup period to release the rest of them. But now that SpaceX and Anthropic are so large, this kind of makes a bigger deal to the stock market in general.

7:01Unidentified speaker (UNKNOWN): So

7:01Unidentified speaker (speaker_2): I found this interesting that if you want to turn to the next chart, we can look at some Federal Reserve data, the flow of funds quarterly report, and it's through the second quarter as of right now. I believe we get third quarter data in the next couple weeks.

7:15Unidentified speaker (UNKNOWN): But

7:15Unidentified speaker (speaker_2): if you want to look at corporate corporate net issuance of equities on any given quarter,

7:21Unidentified speaker (UNKNOWN): and

7:21Unidentified speaker (speaker_2): that's IPOs and secondary issuance netted against buybacks and mergers. So if a company is, you know, has an IPO like yesterday, 800 million shares for anthropic, that's going to create positive net issuance. But as companies buy back their stock, that you know takes some of the supply off the market.

7:37Unidentified speaker (UNKNOWN): And

7:37Unidentified speaker (speaker_2): what you'll see in this chart. is that it's almost always negative. In other words, company buybacks largely outweigh any IPOs that are coming on board or secondary issuance. But in the second quarter, thanks to SpaceX, there was a huge positive number of net issuance. And if you project this forward, thinking about the chart that we just looked at of secondary issuance, that's going to continue here for SpaceX. and it's going to continue for the next year for anthropic. We're going to see some large positive bars in the next couple quarters here, whether it's IPO issuance or secondary issuance, probably outweighing, you

8:13Unidentified speaker (UNKNOWN): know,

8:13Unidentified speaker (speaker_2): buybacks. So, what that means is there's a lot more supply that's being being put out onto the market, and someone's got to be buying that. So it's kind of just going to be acting as a bit of a headwind to the market. So that doesn't mean stock market has to go down, but it's definitely something that's going to be working against you know stocks heading higher. So all that said, let's look at the next chart. Shows some data from Jay Ritter at the University of Florida. He looks more broadly at IPOs and he slices and dices IPO data a million different ways. And some of the more interesting data he has shows how IPOs do on day one, like Anthropic did yesterday, up 17%. I believe SpaceX was up about 20% on their first day of trading, versus how they do over the ensuing three years.

9:05Unidentified speaker (UNKNOWN): So

9:05Unidentified speaker (speaker_2): the blue bars here in this chart show how stocks do on average on their first day of trading, always positive. So in other words, if you bought every IPO in a given year. and sold it at the end of its first day of trading, you never would have lost money. Now, if you look at the orange bars, shows over the ensuing three years, not such a great picture. It's mostly negative bars.

9:27Unidentified speaker (UNKNOWN): So

9:27Unidentified speaker (speaker_2): Jay Ritter, you know, suggests this is probably because nobody's going to know a company's valuation better than founders or CEOs, and they're going gonna to try to go to market when their company is worth the most, when there's the most hype around the company.

9:41Unidentified speaker (UNKNOWN): So

9:41Unidentified speaker (speaker_2): if you're gonna be, you know, selling at what could be a pretty rich valuation, it's pretty hard to continue getting richer valuations as time goes on.

9:50Unidentified speaker (UNKNOWN): So

9:50Unidentified speaker (speaker_2): that's why he how he explains, you know, the orange bars or the ensuing three years kind of nut, maybe living up to the hoopla of the first day of trading. So all in all, these are some pretty negative headwinds in terms of you know IPOs and anthropic in particular. It's it presents a bearish picture, but Jim, I know you're pretty positive about all things AI. So talk me out of, you know, talk me off the ledge here. Why is this not, you know, a totally bearish picture and I should sell all my stocks based on those three charts that I just showed?

10:25Unidentified speaker (speaker_3): Well, those those charts uh do present a a bearish picture. But before I jump into my thing, Greg, I want to ask you a question. Um, is that Jay Ritter chart that you're looking at, that the first day is positive, the next three days is negative. Is there a lot of people that are playing that as the strategy that I want to degen my value into an IPO so I can dump it on

10:49Unidentified speaker (UNKNOWN): the

10:49Unidentified speaker (speaker_0): first

10:49Unidentified speaker (UNKNOWN): day

10:49Unidentified speaker (speaker_0): at

10:49Unidentified speaker (speaker_3): that

10:50Unidentified speaker (speaker_0): big

10:50Unidentified speaker (speaker_3): positive? and not have to deal with the next three years. And is that what we've been seeing over and over again with SpaceX and maybe even with Anthropic as well?

11:02Unidentified speaker (UNKNOWN): Well,

11:02Unidentified speaker (speaker_2): I certainly think so. And I think, you know, maybe if I had to guess, a lot of times companies probably purposely underprice a little bit just to get that positive pop in the first day. I mean, that's great Greg PR. Everyone's saying, oh my gosh. anthropic's up 17%. Let's get in on this thing. But the reality, as the church shows, and as you point out, is I think a lot of people, especially retail investors, go in with this mindset of I'm just gonna take this pop, sell it, and let some other sucker, you know, take the bag for the next three years. And so, you know, the data does bear that out, and it it is what it is, it's the game we play here, but again, I don't I don't think that means that it's gotta be a bearish pitch for stocks here, but I am kind of curious what your take is on it, being a very bullish guy on the AI front.

11:49Unidentified speaker (speaker_3): So I'm a bullish guy on the AI front because I do think that it is the most transformative com technology that we've seen in our lifetime. Time. Um I'm just given giving the this feel that I always give, but I I I firmly believe it. I mean, modern life modern jobs, all our jobs, we sit in front of screens all day long, whether it's a phone, an iPad, or computer. And we have all these software programs that don't talk to each other. And we spend all day long being the conduit between getting data from one to the other, trying to discern data, hunting it on down on the internet, moving it into a spreadsheet, transforming it. into a chart, making it into a PDF, emailing it to a list. AI is going to reduce that to just one sentence that you could tell in your microphone to tell your computer to do it. And it is going to be enormously powerful to change the way that business works, to change the way that our jobs are functioning. There's no doubt about that. Now, I,

12:49you know, so that's the height. hype around AI. And obviously a lot of people believe it because that's why we're talking about CapEx in the hundreds of billions, and we're talking about in the trillions of dollars in valuations. So, you know, the Austin Gouldsbeek statement, he said this yesterday, he's the Chicago Fed president, worrying that AI could be a flash in the pan, is harking back to you know, uh Paul Krugman saying that the internet will be nothing more, you know, transformational than a fax machine in the late 1990s. Uh so let's reject that out of hand. This is the most important technology. Now, I think the good news is, and I have a table of that. I'll show the table here quickly for a quick second, because there's a lot of numbers on it. but it'll be in the show notes. So we could take it down then after a few seconds so it doesn't get in the way. That it just shows the amount of money that these companies raise, the cash that they have on hand, and their stated compute and capex numbers for next year and out to 2030. And all of them now have enough money, especially Enthropy. anthropic, to fund the next year to two years of model building, compute,

14:08and everything else. They've got the money. So the stock market go up, down sideways. We could blather about this or that, but they're all going to show up at work tomorrow or Monday, and they're going to continue to build the next greatest model, Model, the next the next Wiz Bank harness. Harness is just the way that you access the them model. out of Quad is a harness. You know, and the model is um uh you know um opus uh 5.5 or or or something like that. And so I think that they will continue to do that, whether or not the market goes up, down, or sideways. Um the so the transformational uh uh technology of AI is with us. We are going to continue to go forward. with it. There is no going backwards with it. Now, uh, you know, so the next question is can they meet these expectations? They've borrowed a lot of money, they've raised a lot of money through equities, and these people that have invested in this have great expectations of being paid back fully if you're an educator debt investor or making big gains if you're an equity investor. The answer I think. in the long term is yes, because the demand for AI is so low. Most of us that are watching this, we use AI as a sophisticated Google search. That's it. You have

15:28most of us have not yet gone to changing workflows. Flows. You know, read my email, tell me which emails I need to answer, draft a response. Show me the response. I read the response. Send the response. Go get this data. Make up transform this data into this. Make it into a chart, into a PDF, send it to that list. That's a workflow. That's better than just, you know, tell me, you know, how to buy tickets for the next upcoming concert or baseball game that I'm interested in, which is the way that most people are using it. You can use it for that too. So I still think that the demand for AI is still orders and orders and orders of magnitude greater than what we've seen. And they've got the spending to do it. So given all that positiveness, why is the market going down? I think there's two things. The build-out needs to continue. And we've got two things that have kind of thrown up a potential flag. The first one is general. Janssen Wong resigning from NVIDIA. NVIDIA is more than just the largest company in the world and the largest chipmaker in the world. They were also the Federal

16:38Reserve of AI. He was personally negotiating deals to fund a lot of these companies. You know, look, we'll take a stake in you, we'll give you some funding, you can use that. funding then to turn around and buy our chips. Yes, we all considered it to be circular financing. There's nothing wrong with circular financing. It has been done, vendor financing is what is another way of saying it. It has been done throughout history. The railroads did it, the internet did it, the television and radio uh build-out build out did it, even with the auto companies in the uh turn of the 20th twentieth century. we s we saw it as well too. And you also saw they have trying to vendor circular financing to refinance um Europe after World War Two. That's not new that we've what was what we're seeing happen. And maybe it works out or maybe it doesn't work out, but at least it would get the build out to go. Well, Judson's not there anymore. And the question is without him, can a uh Nvidia NVIDIA be the central bank of AI? That is an open

17:42concern right now. And we'll have to see. And by the way, the last word on him being out of NVIDIA is still an open question. The board has said, no, we're moving on. He's not coming back. But the board represents the interests of shareholders, and a number of large shareholders said, you ought to reconnect. reconsider this. That if it is found that he did not break the law, and we still don't know, maybe he did, maybe he didn't. If she was, Alice Wong was or wasn't a spy, we know he was unfaithful in his marriage, but we don't know whether she was a spy. She's been accused of being a spy. Maybe she was, maybe she wasn't. If there was no wrongdoing on Jensen's part, there will be a push for him to come back. Um, and so the second thing is the blue wave. And the blue wave um is very simple. These companies are flushed with cash. Um, in video uh excuse me, not NVIDIA, uh Anthropic has $240 billion of cash. They are ready, they they've have they are ready to spend it to get computers. compute. Well, compute comes from a data center, but that data center might be an empty parking lot with a bunch of picketers in front of it, blocking it from being built because they've

19:02invet they've elected a bunch of politicians that agree with them that it shouldn't be built. So they're got big wads of money at hand and they're looking for compute to spend it on, and it's not there. And they've got investors that have great expectations, Alex you pointed out the huge growth rate, that they want to continue to have that revenue growth rate go. Well, we're going to need more compute in order to have that revenue growth rate to go. Now, President Trump, literally, as we've been talking, President Trump has now weighed into this conversation. And he has said that he does not want any data center built outside of the United States. And that if any company in the United States builds a data center outside the U.S., he's going to tax it. Well, he can't do it alone. I mean, you know, I last I checked the Constitution says that all tax bills have to start in the House, be passed by the Senate, and sent to the desk.

20:02of the president as a bill. He can't wave his hands and just tax U.S. companies. But nevertheless, the sentiment is important. And the sentiment is he is not going to want to see data center construction being pushed away from the United States or slowed down at any point. So maybe that's an offset to it as well, too. But nevertheless, I still think the genie's out of the box. bottle with this technology. We are not going back the other way. The only way is to go through with this technology and to completely adopt it and completely assimilate it into the way that we live our lives and we operate ourselves and our jobs in front of screens. So yeah, I'm an AI bull. I have found this technology to be tremendously uh transformative in the way that I do my job. I know Alex and Greg, you have to varying degrees as well, too. And it is going to continue to move forward. But yes, there is always with new technologies, there is hype and there is volatility, and that's what we're seeing right now.

21:09Alex Melitas: So to you know, to round out a couple of these uh ideas,

21:14Unidentified speaker (UNKNOWN): uh

21:14Alex Melitas: even better. before the blue wave, even before Jensen, we we knew that we had a bit of a data center problem. I think the stat is around 40 or 50 percent of data centers uh that are announced never actually get started or built or or the approvals needed to even do anything with them. Uh then a couple weeks ago you had Elon essentially say that financing for the these smaller scale uh data. center companies, they're called neo clouds, essentially, think core weave and these Nebius and all these other companies, that the funding for them has just dried up. And like you said, a lot of that was coming from NVIDIA kind of doing their vendor financing and allowing these companies to expand compute capacity for these frontier AI labs. Well, like you said, the problem we have now is Jensen's gone, data centers are were on the ballot, and people rejected them. And we now have no funding for these uh these neo clouds who are showing extreme stress

22:21in the debt they've already issued. So how do we push through? this the AI build out when you know SpaceX can only build so fast meta can only build um you know Colossus or um I I can't I can't exactly remember the the meta's uh version but you

22:43Unidentified speaker (UNKNOWN): know

22:43Alex Melitas: how how do you push through that how does AI continue to grow especially for anthropic and for the coming IPO of OpenAI? How do they continue to move forward in this environment?

22:58Unidentified speaker (speaker_3): So I'll I'll start,

23:00Unidentified speaker (UNKNOWN): and

23:00Unidentified speaker (speaker_3): Greg, you could jump in if you got something. AI data centers, look, here's the problem. They oversold this technology in the early days. Um, look, you have to invest in anthropic, you have to invest in overcome. OpenAI. Why? Because it's going to be such a transformative technology. It's going to destroy the world. And, you know, if we believe Gavin Baker, when he said that the people at Anthropic, this goes back, you know, a year or so ago, that they were actually arguing that there's only gonna be one company left in the world. One company, it's gonna going to be anthropic. Everything else is gonna going to be a uh government agency. Government agency to make cars, a government agency to make toothpaste, a government agency to give you your phone. But there will only be one company. And everybody's going to be unemployed because of that company. And then you got Bill Gates saying it's gonna, you know, kill a billion people. He's been watching too many Terminator movies on rerun. Okay, you you accomplished your task. You sufficiently will. whipped up enough hype that everybody invested in your companies and you got it to a $2 trillion valuation at yesterday's high with Entrop Anthropic. Congratulations for a company that did exist five years ago. You went from zero to two trillion dollars at a speed that no one could have even fathomed five years ago.

24:19But you've made the public very mad at you right now. They don't like the idea of losing their jobs. They don't like the idea of dying. And they want to push back on you. And the data center is, to the public's mind, the church of AI, and they want to burn it down. And I don't think it's really if it was as logical as water usage and power requirements, pushing up my electric bill and the like, we could solve that in five minutes. And they've already solved it. these technical things that you might, but it's more than it, it's a visceral thing. enjoying it. And I used to joke that the investment theme, and I'm saying this as a piece of sarcasm, not investment advice, is to buy Turex and to buy Oshkash. They make fire trucks. Because the first thing you're going to do if you're going to go into a community and you're going to say we're going to build a data center in your community is you're going to you're going to um come in

25:29with a a bunch of fire trucks and go, here's a bunch of new fire trucks for your community. You could just now I'd like to discuss building a data center with you and I just gave you fire trucks. Imagine what else I could do. Oh, I've got a machine here that can, you know, um repave all your roads, and I can give all your teachers uh in all of your public schools raises, and we've got new garbage trucks and everything else. That's what you have to do. People have to believe they're getting something out of it. The worst thing these companies did was have was was signed NDAs, non-disclosure agreements with the uh uh local governments about these uh data centers did enormous reputational damage for them. So maybe the public would be more accepting of these data centers. Why do we want a data center in our community? Because no one's raising my taxes, all my roads are plowed, they get Uh, you know in the winter, we've got no potholes, our garbage gets picked up in shiny new garbage trucks, we have new we have new fire trucks, new police trucks, uh, police cars, and the teachers all get raises and the schools have been refurbished. That's why we want a data

26:35center in our neighborhood. That's why we want a data center in our community because we don't have to pay for that stuff. That would be the first thing. The second thing on the demand side is maybe as people start to see what AI can do for. them with workflows and efficiencies and the like, they'll stop. This is my beliefs. They'll stop fearing that it's the end of the world and start seeing the opportunity. It'll be like the internet. Like I said earlier, Paul Krugman said, the internet, uh, it's just the fax machine in 1996. By 2000, everybody was sitting in front of their computer on the internet. all day long and going, wow, this is a fantastic tool. If we can get everybody to that point with AI, then they'll stop being angry about it. They'll stop pushing back on it and they'll want more of it. Now that's the challenge for the community of the uh AI community to try and convince people to let them build the data centers and try and convince the rest of us that we need and want this tool and we shouldn't fear this tool. It's easy for me to say it's hard to do. Greg, do you have a follow-up on that or a thought on that?

27:52Unidentified speaker (UNKNOWN): Well

27:52Unidentified speaker (speaker_2): I wanted to circle back to you know, we've talked previously about this Jensen And Wong issue and I not to belabor it too much because I know it has become a politically charged issue and if we could set that aside, we've kind of had disagreements on, you know, I'm with you that I think the the blue wave is going to be slightly problematic for data centers being built. I think eventually they will be built, but you know, at a slower pace, and it's going to be at the Democrats speed. Um as far as you know the NVIDIA drama goes, I'm, you know, I hear that the Constant. constant stant that, you know, now that Jensen's gone, he had all these handshake deals, now NVIDIA can't do, you know, what they used to do and they're not gonna be the central bank of AI. I'm not buying that necessarily. I think that you know, whoever ends up taking his spot, the next person in line can pretty easily, I think, replicate that and just say, Okay, we're gonna keep doing, you know, exactly what we were doing before. We're gonna more or less finance this. You know, I would see a much bigger issue if say Elon left any of his companies. You know, that's a brain drain right there. And not that it's not a knock on Jensen at all. He's a brilliant guy, I'm sure, but you know, what he is known for is, you know, more or less financing this whole AI build out. That to me seems like something

29:12that is replicable. And I mean where where do you stand? on that, Jim?

29:16Unidentified speaker (speaker_3): Yeah, you know, a couple of things. It's unclear whether or not the board of NVIDIA wants to continue to be the central bank of AI. You know, if they show noise or the ability to continue with that, that's one thing, but they haven't yet. Now, part of that might be because they've only got an interim C. CEO and they don't want to be making long term commitments until they get a new long term CEO in place. And then that'll probably be the first question. You know, the second thing is Jensen is still out there and he's a free agent. Now, if he gets embroiled in law breaking and scandals with spies, I mean he's got a different issue at home with his wife, but you know, if he gets embroiled in that, that might sideline him. But if it doesn't, and he's a free agent, you know, Elon has started a company called Terrafab. He started it in March, and Terafab is basically a chipbuilding company for SpaceX and Tesla and Optimus Robots and XAI and the like. And he just posted for CEO of Terrafab. And if you read the um the requirements to be the CEO of Terrafab, there's only one person in the world that's qualified for that job. And it's a and it's

30:34Jetson right now. And so if Jetson were to go over to Terafab with Elon, he could replicate being the central bank of AI in about one day. He's got the cachet to raise a lot of money, he's got all of the contacts, they all trust him. Um, so maybe the central bank AI just continues in a different form as well. But to some degree, vendor financing has been all those flow charts and Venn diagrams everybody shows. The reason that it happens over and over again is because when you get a new technology, the only person that's really qualified to finance it are the people that. are in the middle of it, like Jensen. Traditional bankers, traditional investment banks, Wall Street. They don't have the tools, they don't have the markets, they don't have the expertise to finance this kind of thing because they don't fully understand it like somebody like Jensen does. That's why you get vendor financing to the degree that you get when you have a new technology. You don't go knocking out on Jamie Diamond's door. He doesn't understand what it is. Uh, and you know, eventually he will, and eventually the you will get to them. But in the early stages, the vendor financing thing is very normal. Now that I'm I'm not saying it's not risky, it may

31:53not work out, but everybody's acting like this is something new. It's not, it's as old as time what we're doing. The scale obviously is larger than we've ever seen, but it's everything associated with AI is the largest scale we've ever seen. Alex, you want to throw in any ideas or thoughts there?

32:10Unidentified speaker (UNKNOWN): No,

32:10Alex Melitas: I mean I think you covered it really well. I mean,

32:13Unidentified speaker (UNKNOWN): I I

32:14Alex Melitas: I see where you're coming from, Greg. I would say as much as I'm not a not a Patriots fan,

32:20Unidentified speaker (UNKNOWN): you

32:20Alex Melitas: take Brady off the Patriots all those years ago. It wasn't the same team, right? Same logo, same coach. but you know, didn't work. So, and and look at and look what happened. Brady left and you know did the same thing with the Bucks. So, yeah, Jensen can do the same thing. I mean, there are certain people that define,

32:41Unidentified speaker (UNKNOWN): you

32:41Alex Melitas: know, the the define the situation of success um for their organization. And I think Jensen was that for NVIDIA.

32:50Unidentified speaker (UNKNOWN): Uh,

32:50Alex Melitas: but that's all I have. I think that does it for this week's episode. If you enjoyed our commentary, please like and subscribe. If you have any ideas, comments, or questions, be sure to comment them down below. Thank you for joining us and we'll see you next week.

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