SpaceX-Tesla merger talks advance; Anthropic's $1.4T IPO signals AI demand amid market rout
Summary
Speakers discuss advancing merger talks between SpaceX and Tesla, which could create a $3.5 trillion entity, while noting national security and shareholder control challenges. They analyze Anthropic's six-times oversubscribed IPO, valued at $1.4 trillion, highlighting major tech investments and its role as a bellwether for the AI capital expenditure cycle. Hyperbolic CEO Jasper Zhang argues that inference demand growth outpaces supply, keeping GPU prices high despite recent sell-offs. Speakers contrast Hyperliquid's pre-IPO pricing, which suggests higher upside, with traditional IPO metrics, while warning of macro risks like gold selling off and geopolitical tensions.
Key takeaways
- 1 Anthropic's IPO is six times oversubscribed, potentially raising $90.4 billion at a $1.4 trillion valuation, making it the largest IPO in history.
Unidentified speaker (speaker_1) · 31:47
- 2 the market, a lot of the AI trade depends upon them honoring their compute commitments, which again, the IPO is not the only source of funds for that, but like it's a great bellwether for how much real demand there is in the market, both to own Frontier Labs and to spend and like to predict what the CapEx cycle looks like on the short term.
Unidentified speaker (speaker_0) · 38:41
Episode sections
- 0:00 – 3:24
Intro and Tweet of the Day
Speakers discuss US market potential, SpaceX IPO, rumored $3.5T Tesla merger, Anthropic's oversubscribed IPO, and Hyperbolic CEO segment.
- 3:24 – 14:18
SpaceX-Tesla Merger Talks
Bloomberg reports advancing merger talks, potential Dec/2027 announcement, $3.5T entity size, and challenges regarding control, security, and votes.
- 14:18 – 31:47
Hyperbolic CEO on Neoclouds
Jasper Zhang argues 10x annual inference demand growth lags supply, keeping GPU prices high; addresses CoreWeave debt risks and advocates for global compute grid.
- 31:47 – 37:18
Anthropic IPO Details and Market Impact
IPO is six-times oversubscribed, raising $90.4B at $1.4T valuation; highlights mixed signals amid AI stock rout shedding $6T market cap.
- 37:18 – 38:41
Hyperliquid Pricing and Investor Sentiment
DGENs price 20% pop; pre-IPO market values Anthropic at ~$1.77T; Speaker 0 views this as truer indicator than IPO target.
- 38:41 – 41:55
CapEx Cycle and Macro Risks
Anthropic's success critical for AI trade and hardware demand; strong IPO could trigger relief rally, but risks include gold selling off and geopolitical tensions.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| SpaceX | private company | neutral | Not even a year out from its record IPO, SpaceX could set another record in its reported $3.5 trillion merger with Tesla, which could hit as early |
| TeslaTSLA | equity | neutral | However, Bloomberg sources say the companies still have to figure out how to keep the China business separate, while Elon Musk has repeatedly said that Tesla has no plans to split. that part of the business off. |
| NVIDIANVDA | equity | neutral | Yeah. And at $3.5 trillion, uh, with NVIDIA drawing down, this means the combined entity could be close to NVIDIA's market cap. |
| BitcoinBTC | crypto | neutral | the firelings, which I had not seen other people talk about, which is both of Elon's companies own Bitcoin. And it's not just a little bit, it's like a lot. And these are almost like second fiddle to any other story about these companies here. |
| Anthropic | private company | neutral | We'll bring him on up here and then we'll go to the anthropic IPO. |
| AlphabetGOOGL | equity | neutral | If we take a look at that private placement, Charlie, Nvidia and Google each invested $10 billion into In the private placement, and uh AMD invested $2.5 |
| MicrosoftMSFT | equity | neutral | Broadcom, Microsoft, and Micron each invested $1 billion and TSNC getting a piece here, investing half a billion in that $26 billion private placement as well. |
| Gold | commodity | bearish | That being said, I think we should probably just do a an entire episode on is the AI trade unraveling? 'Cause there's been a lot of warning signs flashing recently, including I would add the fact that gold is selling off. That's not something that you particularly would like to see. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
SpaceX and Tesla merger talks are moving forward with an announcement possible as soon as December or early 2027.
Unidentified speaker (speaker_0) ·3:24
How to verify: Check Bloomberg News articles from November 2026 regarding SpaceX-Teska merger rumors and SEC filings for any Form S-4 or proxy statements filed by Tesla or SpaceX. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceThe combined SpaceX-Tesla entity would be worth nearly $3.5 trillion.
Unidentified speaker (speaker_0) ·3:45
How to verify: Calculate the sum of SpaceX's latest private valuation and Tesla's market capitalization as of November 2026 using financial data providers like Bloomberg or Refinitiv. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Elon Musk owns about 20% of Tesla but controls 85% of SpaceX's votes through special shares.
Unidentified speaker (speaker_0) ·6:22
How to verify: Review Tesla's latest DEF 14A proxy statement for ownership percentages and SpaceX's private shareholder agreement or recent financing disclosures for voting rights. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Tesla has a large, profitable factory business in China, while SpaceX does US defense work.
Unidentified speaker (UNKNOWN) ·6:50
How to verify: Check Tesla's 10-K filing for China revenue and operational details, and review SpaceX's contracts with the US Department of Defense via USASAC or DoD contract databases. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceSpaceX owns 18,700 BTC and Tesla owns 11,500 BTC.
Unidentified speaker (UNKNOWN) ·10:47
How to verify: Check Tesla's quarterly 10-Q or 10-K filings for Bitcoin holdings. Check SpaceX's financial disclosures or press releases for Bitcoin holdings, or verify via blockchain analysis of known corporate wallets if public. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourcePrediction market odds on Polymarket price the SpaceX-Teska merger completion before March 1, 2027, at 51.9%.
Unidentified speaker (UNKNOWN) ·12:18
How to verify: Access the Polymarket website or API to view the specific market 'SpaceX Tesla Merger Completion Date' and check the odds for the 'Before March 1, 2026' outcome as of the transcript date. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Inference demand is growing 10x every year.
Unidentified speaker (speaker_3) ·17:18
How to verify: Review reports from Epoch AI or other AI infrastructure tracking organizations regarding compute demand growth rates for inference versus training. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
CoreWeave has a more than 70% prepaid revenue mix.
Unidentified speaker (UNKNOWN) ·20:49
How to verify: Check CoreWeave's earnings reports or investor presentations for details on revenue recognition policies and the percentage of revenue recognized in advance. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Nebius and Iron are delaying some of their revenue targets.
Unidentified speaker (speaker_1) ·15:25
How to verify: Check press releases or earnings guidance updates from Nebius and Iron for any announcements regarding revised revenue targets or delays. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Nebius is moving faster on building AI solutions, including post-training solutions, compared to CoreWeave.
Unidentified speaker (speaker_3) ·29:43
How to verify: Review product launch announcements and technical blogs from Nebius and CoreWeave regarding their AI software suites and post-training capabilities. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Anthropic's IPO is six times oversubscribed and could raise as much as $90.4 billion, including $26 billion from a private placement and $64.4 billion from the IPO offering.
Unidentified speaker (speaker_1) ·31:47
How to verify: Bloomberg article reporting on the IPO details; SEC F-1 registration statement for Anthropic when filed. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceThe IPO would give Anthropic a $1.4 trillion valuation with a $70 to $78 share price.
Unidentified speaker (speaker_1) ·32:07
How to verify: Bloomberg article; SEC F-1 registration statement for Anthropic when filed. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceAI stocks have shed six trillion dollars from their collective market caps in a multi-month rout, according to MSCI research.
Unidentified speaker (speaker_1) ·32:27
How to verify: MSCI research report cited by the speaker; cross-reference with MSCI AI Index performance data. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceSpaceX holds the record for the largest IPO in history, set earlier this year.
Unidentified speaker (speaker_1) ·32:07
How to verify: SEC F-1 filing for SpaceX; historical IPO data from Bloomberg or Reuters. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceAnthropic's annualized revenue run rate rose from $65 billion at the end of July to $81 billion at the end of October.
Unidentified speaker (speaker_1) ·34:25
How to verify: Bloomberg article citing Anthropic's financials; SEC F-1 registration statement for Anthropic when filed. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceNvidia and Google each invested $10 billion into Anthropic's private placement, AMD invested $2.5 billion, and Broadcom, Microsoft, and Micron each invested $1 billion.
Unidentified speaker (speaker_1) ·35:24
How to verify: SEC Form D filing for Anthropic's private placement; press releases from respective companies. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceTSNC invested half a billion in the $26 billion private placement.
Unidentified speaker (speaker_1) ·35:39
How to verify: SEC Form D filing for Anthropic's private placement; press releases from TSNC. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceDGENs on Hyperliquid are pricing in a 20% pop for Anthropic, with the pre-IPO market valuing it at ~$1.77 trillion.
Unidentified speaker (speaker_1) ·37:18
How to verify: Hyperliquid platform data for pre-IPO markets; screenshot or archive of the specific market pricing at the time of the broadcast. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceAnthropic plans to issue 801.35 million new shares, while existing shareholders will sell 23.65 million shares.
Unidentified speaker (speaker_1) ·36:19
How to verify: SEC F-1 registration statement for Anthropic when filed. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Primary sourceGold is selling off, which is cited as a warning sign for the broader market.
Unidentified speaker (speaker_1) ·39:59
How to verify: Gold price data (e.g., LBMA Gold Price) for the week of November 13, 2026. For recent events, verify 2026-08-13 through 2026-11-13. Use another period only when the speaker explicitly names it.
Do your own research
- Verify Bloomberg's reporting on SpaceX-Tesla merger timelines and $3.5 trillion valuation assumptions.
- Check SEC filings for Anthropic's S-1 registration statement to confirm the $1.4 trillion valuation and $90.4 billion raise details.
- Review Hyperliquid's pre-IPO market data to validate the $1.77 trillion implied valuation and 20% upside claim.
- Analyze Nvidia, Google, AMD, and other investors' public disclosures regarding their $10 billion and $2.5 billion contributions to Anthropic's private placement.
- Assess CoreWeave's debt structure and prepaid mix to evaluate default risk claims made by Hyperbolic's CEO.
- Monitor gold price trends and geopolitical news to assess the validity of 'market stress' warnings cited by speakers.
- Compare Anthropic's projected ARR against trailing twelve-month revenue to verify the 17.3x multiple claim.
- Review recent AI stock market cap changes to contextualize the '$6 trillion shed' claim against broader market data.
Full transcriptexpand
- 0:00
- Unidentified speaker (UNKNOWN): The
- 0:00
- Unidentified speaker (speaker_0): US market is on pace for three records this year, and no, I'm not talking about a record stock market route. Not even a year out from its record IPO, SpaceX could set another record in its reported $3.5 trillion merger with Tesla, which could hit as early
- 0:20
- Unidentified speaker (speaker_0): as next month.
- 0:22
- Unidentified speaker (UNKNOWN): You're
- 0:22
- Unidentified speaker (speaker_0): listening to Blockspace Live, and we will be covering the details of SpaceX and Tesla's potential. potentially imminent combination in our first story. For our other story today, though, we will be covering potentially another record. That's Anthropic's reportedly six times oversubscribed IPO, which is on track to beat out SpaceX for the
- 0:41
- Unidentified speaker (speaker_0): largest IPO in history. Before we get to Anthropic, Topic though, we've got Jasper Zhang, the CEO of Hyperbolic on, to talk about what the hell. is happening with NeoClouds.
- 0:53
- Unidentified speaker (speaker_1): You are
- 0:54
- Unidentified speaker (speaker_0): listening to Blockspace Live Compute's daily live show, co hosted by Charlie Spears and Colin Harper of Blockspace Media. For more about Blockspace Media and more content, including written stories, you can go to Blockspace.media. Media. That's our website, Blockspace.media. We cut up we kick
- 1:14
- Unidentified speaker (speaker_0): off every show with a tweet of the day. And after this week, Colin, we have to do this one. This one comes from friend of the show, Steven Lupka, with a screenshot of the front page, the cover of The Economist,
- 1:30
- Unidentified speaker (speaker_0): which says,
- 1:32
- Unidentified speaker (UNKNOWN): for
- 1:32
- Unidentified speaker (speaker_0): those of you listening,
- 1:35
- Unidentified speaker (speaker_0): is the AI trade over?
- 1:38
- Unidentified speaker (UNKNOWN): What?
- 1:38
- Unidentified speaker (speaker_0): Whatever next, and as we know, Colin, if the economist says it's over, that is one of the best bottom signals you could possibly get
- 1:48
- Unidentified speaker (speaker_1): almost almost as good as a reverse Kramer. Uh, you know,
- 1:52
- Unidentified speaker (UNKNOWN): if
- 1:52
- Unidentified speaker (speaker_1): the economist is saying it, probably a good countercyclical. That being said, I you man, if you look at the price, I mean core weave is down another nine percent today. Believe Nebia. Nebius and Iron in the same bucket. A lot of the stocks that we cover are just like it is an unrelenting sell-off. And we're gonna be talking about it in the
- 2:12
- Unidentified speaker (speaker_1): anthropic section. Maybe, maybe that buoys everything and saves the market. It's six X oversubscribed. Not the thing you'd expect to see if investors thought the trade was truly over. I'll say
- 2:25
- Unidentified speaker (UNKNOWN): I
- 2:25
- Unidentified speaker (speaker_0): would I would ask, like, is there any money left to buy Anthropic? But apparently, there absolutely is. Maybe all the money is just fleeing, everything else going into Anthropic. We will dissect that topic here. We're gonna go talk about the titanist merger between Tesla and SpaceX
- 2:45
- Unidentified speaker (speaker_0): rumored merger after a word from our sponsor, Saluda.
- 2:54
- Unidentified speaker (speaker_2): Saluna Holdings, NASDAQ Lysoticker SLNH, develops data centers co-located with wind and solar plants, turning surplus renewable energy into computing power for AI and other intensive applications. We are proud to partner with Saluna Holdings here at Blockspace. And if you would like to learn more, please visit Saluna
- 3:13
- Unidentified speaker (speaker_2): salunacomputing.com. Computing.com.
- 3:15
- Unidentified speaker (speaker_2): That's salunacomputing.com.
- 3:24
- Unidentified speaker (speaker_0): The biggest deal in corporate history might be about a month away. Bloomberg reported yesterday that SpaceX and Tesla merger talks are moving forwards. An announcement of the merger could come as soon as December. or early 2027. It would be SpaceX buying Tesla,
- 3:44
- Unidentified speaker (UNKNOWN): and
- 3:45
- Unidentified speaker (speaker_0): the combined entity would be worth nearly $3.5 trillion, making it by far the largest merger in history.
- 3:53
- Unidentified speaker (UNKNOWN): The
- 3:54
- Unidentified speaker (speaker_0): combined entity, SpaceX and Tesla, would be a chip producing, robot manufacturing, data center building, telecommunications, and rocket. ship juggernaut.
- 4:05
- Unidentified speaker (UNKNOWN): But
- 4:05
- Unidentified speaker (speaker_0): it's not a straightforward deal. Tesla shareholders would have to give up control of their company to Elon Musk under current terms, and Elon would have to reconcile SpaceX being a US government defense contractor while also selling cars in China. We will go deeper into the complexities and considerations for the merger, but first,
- 4:25
- Unidentified speaker (speaker_0): let me bring up my on co-host Colin Harp. Harper. Colin, even though we're talking about rockets and cars, this is probably more of an AI story than it is either of those things. Do you think at a high level, zooming far out into low Earth orbit, that this deal makes sense?
- 4:43
- Unidentified speaker (speaker_1): Oh, it's probably there's no deal that you could find that makes more sense than this. This is conceivably the most obvious merger in
- 4:50
- Unidentified speaker (UNKNOWN): the world.
- 4:51
- Unidentified speaker (speaker_1): in
- 4:52
- Unidentified speaker (UNKNOWN): I
- 4:52
- Unidentified speaker (speaker_1): won't say history, but at least in this field of AI. And what I mean by that is we've covered this on the show before. With SpaceX, you have the infrastructure layer for AI and the software layer with their own model and with services. With Tesla, you have the hardware layer, you've got the Optimus Robotics
- 5:12
- Unidentified speaker (speaker_1): line, you've got you've got uh Tesla. cyber cabs and their cars themselves, which are already using AI features for assisted for um for you know non-human driving, for self-driving, things like that. To me, this it would be one of the most incredible business combinations. And yeah, I will say that in the history of the world. You
- 5:32
- Unidentified speaker (speaker_1): there's just an immense untapped untapp potential here. And then if you add Terrafab on top of that,
- 5:38
- Unidentified speaker (UNKNOWN): for
- 5:39
- Unidentified speaker (speaker_1): what that will do for not just SpaceX, but also for Tesla with their own robotics line and with the Tesla model cars, it's a no brainer. Now, if they can iron out the national security stuff with China and also the fact that Tesla shareholders would have to give up control of the company to Elon Musk. As long as you can
- 5:59
- Unidentified speaker (speaker_1): iron that out, I don't see any reason why this wouldn't go through.
- 6:02
- Unidentified speaker (UNKNOWN): Yeah.
- 6:03
- Unidentified speaker (speaker_0): this is a very, very obvious merger. Tesla, robotics, and chips, SpaceX, everything else. Um, so let me get into the challenges here because this is there are some hoops that they're gonna have to jump through and some problems that they're gonna have to iron out. First one, um, the near-term one, kind of the obvious corporate challenge, which is the the the issue for the
- 6:22
- Unidentified speaker (speaker_0): Tesla shareholders. So Musk owns about 20% of Tesla.
- 6:27
- Unidentified speaker (speaker_0): but controls 85% of SpaceX's votes through special shares that carry voting power. So SpaceX or Elon does not control Tesla. And today Tesla shareholders can vote against Musk's Musk. If SpaceX bought Tesla under the current regime, Elon would basically get
- 6:47
- Unidentified speaker (speaker_0): control of both entities.
- 6:50
- Unidentified speaker (UNKNOWN): Then
- 6:50
- Unidentified speaker (speaker_0): we have the security issue. Tesla has a very large, profitable factory business in China, and SpaceX does US defense work. These things are kind of a no-no.
- 7:02
- Unidentified speaker (speaker_0): However, Bloomberg sources say the companies still have to figure out how to keep the China business separate, while Elon Musk has repeatedly said that Tesla has no plans to split. that part of the business off. Little question mark there.
- 7:20
- Unidentified speaker (speaker_0): Then there is a sh couple details about like the shareholder vote and transition. One being Musk called that the uh the vote that he has to convince to this Tesla and SpaceX shareholders quite a high hurdle. So that quote is uh came from
- 7:39
- Unidentified speaker (speaker_0): uh I believe a Tesla earnings call previously no the SpaceX earnings call previously. And then there's about 286 million of Musk's restricted Tesla shares from his 2018 pay package, which would vest early if Tesla changes hands. These are kind of the obvious ones, and I don't know the solution to any of them. They all seem
- 8:00
- Unidentified speaker (speaker_0): maybe like
- 8:01
- Unidentified speaker (UNKNOWN): uh
- 8:01
- Unidentified speaker (speaker_0): hurdles you can surpass. with the exception of the national security one, Colin. Um, I have a couple more things, but like of these different hurdles, which of these stands out to you as like the glaring like challenge for the companies.
- 8:17
- Unidentified speaker (speaker_1): I mean, maybe I'm over-indexing here, but look at what happened with Jensen. And I uh China versus the Pentagon 100%. The the fact that there might be some competing. national security interests and corporate interests here to me seems like one of the biggest hurdles for this merger.
- 8:35
- Unidentified speaker (UNKNOWN): Um,
- 8:35
- Unidentified speaker (speaker_1): you know, the Tesla shareholders could vote against Musk for not merging the companies. You know,
- 8:42
- Unidentified speaker (UNKNOWN): I
- 8:42
- Unidentified speaker (speaker_1): don't know if they would really ultimately want to do that if you take it, if you take a look at everything that these two companies could accomplish together, the shareholder value. from that would be immense, you know, five, ten years down the road. So I don't really see that as a huge problem. But
- 8:58
- Unidentified speaker (UNKNOWN): um
- 8:58
- Unidentified speaker (speaker_1): even I mean, Musk did call it a high hurdle himself. So maybe that I'm maybe I'm under uh appreciating the fact that some of these shareholders want to keep these businesses separate. Obviously, there are bad optics for Musk here, where you're taking two of the most valuable companies in the world, merging them together. and giving Musk more control over Tesla than he
- 9:18
- Unidentified speaker (speaker_1): previously had. So that doesn't really, you know, that runs afoul of people who think that Musk is already too powerful in the position that he's in. Right.
- 9:27
- Unidentified speaker (speaker_0): Yeah. And at $3.5 trillion, uh, with NVIDIA drawing down, this means the combined entity could be close to NVIDIA's market cap.
- 9:37
- Unidentified speaker (UNKNOWN): Um
- 9:37
- Unidentified speaker (speaker_0): and uh well. into the top 10 largest corporate entities in the world.
- 9:42
- Unidentified speaker (UNKNOWN): Um,
- 9:43
- Unidentified speaker (speaker_0): I've got a couple more details here that I think are kind of interesting, um, relevant to our show. So uh
- 9:54
- Unidentified speaker (speaker_0): the companies are already kind of like merging Emerging a little bit, especially if you look at the Terrafab Terafab deal. This is a joint project between the two companies. Obviously, space. SpaceX, or I'm sorry, a Tesla project, but the money uh for Terrafab and the like groundwork and the work on it is kind of a joint venture, although spa Tesla does
- 10:14
- Unidentified speaker (speaker_0): um call uh it is a Tesla project. And then another detail, which I think may not be as interesting to our uh our AI viewers, but to our longtime listeners.
- 10:27
- Unidentified speaker (speaker_0): I I noticed this from
- 10:30
- Unidentified speaker (UNKNOWN): uh
- 10:30
- Unidentified speaker (speaker_0): the firelings, which I had not seen other people talk about, which is both of Elon's companies own Bitcoin. And it's not just a little bit, it's like a lot. And these are almost like second fiddle to any other story about these companies here.
- 10:44
- Unidentified speaker (UNKNOWN): Um,
- 10:44
- Unidentified speaker (speaker_0): they each own a respective
- 10:47
- Unidentified speaker (UNKNOWN): uh
- 10:47
- Unidentified speaker (speaker_0): SpaceX 18.7,000 18.7 thousand and Tesla 11.5. eleven point five thousand, BTC.
- 10:53
- Unidentified speaker (UNKNOWN): The
- 10:53
- Unidentified speaker (speaker_0): combined entity would put it in the top five largest corporate holders of Bitcoin in the world, right behind Strive strategy, obviously the leading the pack. I think this is kind of an interesting uh story. So if you're like in the Bitcoin world, you can obviously uh spin this narrative however you want.
- 11:12
- Unidentified speaker (speaker_0): Um
- 11:13
- Unidentified speaker (UNKNOWN): and
- 11:14
- Unidentified speaker (speaker_0): clearly
- 11:14
- Unidentified speaker (speaker_1): they should. be valued more for the Bitcoin on the balance sheet.
- 11:17
- Unidentified speaker (UNKNOWN): Yeah, when Bitcoin
- 11:18
- Unidentified speaker (speaker_0): goes back up, we can listen to Bitcoin story.
- 11:21
- Unidentified speaker (speaker_1): But I mean, I think the bigger story here, it turns out that the real Bitcoin treasury companies were the actual operating businesses along that we met along the way. Like this is kind of, I think, the
- 11:33
- Unidentified speaker (speaker_1): this is the model for a Bitcoin treasury company. Yeah. An Actual business that can execute and potentially. add Bitcoin to its treasury whenever it wants to on an ad needed basis. It's gonna be a very small slice of the pie, obviously. Right. I mean, this is nothing to SpaceX or Tesla in the grand scheme of
- 11:52
- Unidentified speaker (speaker_1): things, but
- 11:54
- Unidentified speaker (speaker_0): Yep. So uh that wraps up our
- 11:58
- Unidentified speaker (UNKNOWN): actually,
- 11:59
- Unidentified speaker (speaker_0): Charlie.
- 11:59
- Unidentified speaker (speaker_1): I have just one thing I wanted to share really quickly. Oh,
- 12:02
- Unidentified speaker (UNKNOWN): yeah, yeah,
- 12:02
- Unidentified speaker (speaker_0): okay. Show Shows us.
- 12:03
- Unidentified speaker (speaker_1): Um, just 'cause in terms of like when this will when people are saying this will likely uh hit.
- 12:10
- Unidentified speaker (UNKNOWN): If
- 12:10
- Unidentified speaker (speaker_1): we look at prediction market odds on Calchi, majority of of traders are pricing in before
- 12:18
- Unidentified speaker (UNKNOWN): uh
- 12:18
- Unidentified speaker (speaker_1): March first, twenty twenty seven. Fifty 51.9% one point nine percent are pricing in before February first, 1st, twenty 2027. twenty seven. Only thirty 33.5% are pricing it in before the end of the year. So I think kind of like with a lot of things with Elon Musk, maybe there's some signal here because he often
- 12:38
- Unidentified speaker (speaker_1): overpromises and then it takes longer. Sometimes he underdelivers. Sometimes he says that he's going to hit a target that he doesn't hit until five years later. But um all of the prediction markets for this getting a bounce on today's news that they at least think that. it will happen sometime soon.
- 12:53
- Unidentified speaker (speaker_0): Yeah, I got gut sense is that this makes too much sense for it to not happen. It's gonna take a little bit longer. I I bet you, I bet you if I were a betting man, they announce it next month and then it doesn't fully complete merger until sometime in 2027. Hard to really like know exactly when that's
- 13:12
- Unidentified speaker (speaker_0): going to be. Well, we're gonna keep on going with our show. We have Jasper, CEO of Hyperbolic, in the wings. We'll bring him on up here and then we'll go to the anthropic IPO. But before that, a word from our sponsor, Luxer.
- 13:35
- Unidentified speaker (speaker_0): This show is brought to you by Luxer's Commander, Bitcoin miner. management software built for enterprise operations. Commander gives you real-time fleet monitoring, bulk remote commands across your fleet, and intelligent miner, which is an automated profitability engine that runs every five minutes, adjusting power settings to live hash rate and energy
- 13:54
- Unidentified speaker (speaker_0): markets.
- 13:57
- Unidentified speaker (speaker_0): Show 10% more profitability versus binary mining.
- 14:00
- Unidentified speaker (UNKNOWN): And
- 14:00
- Unidentified speaker (speaker_0): Commander Pro is a hundred bucks per megawatt. or twenty-five basis point full fee added, which is roughly half the price of your competitors.
- 14:09
- Unidentified speaker (UNKNOWN): With
- 14:09
- Unidentified speaker (speaker_0): a 60-day free trial, get started at Luxer.tech forward slash.
- 14:18
- Unidentified speaker (speaker_2): GPU rental rates are through the roof, but neo clouds like Core Weave, Nebbius, and Iron are
- 14:24
- Unidentified speaker (speaker_1): getting
- 14:25
- Unidentified speaker (speaker_2): hammered
- 14:25
- Unidentified speaker (UNKNOWN): in the
- 14:26
- Unidentified speaker (speaker_2): AI
- 14:26
- Unidentified speaker (speaker_1): stock market. route
- 14:27
- Unidentified speaker (speaker_2): despite the voracious demand for
- 14:30
- Unidentified speaker (speaker_1): GPU resources. Is the neo cloud business model broken or is there a mismatch between supply and demand? Hyperbolic CEO Jasper Zing joins us today to unpack those questions and also take a look at the chaos that is fomenting in the neo-cloud neo cloud market. Let's get Jasper back on as well as It's my co-host Charlie
- 14:49
- Unidentified speaker (speaker_1): Spears. Jasper Zhang, welcome uh back to the show for the first time in virtual format. We had the pleasure of speaking with you at
- 14:57
- Unidentified speaker (speaker_2): Yada. So thanks for coming back on.
- 15:00
- Unidentified speaker (UNKNOWN): Thanks
- 15:00
- Unidentified speaker (speaker_3): for inviting me and good to see you guys again.
- 15:03
- Unidentified speaker (speaker_0): Yeah.
- 15:03
- Unidentified speaker (speaker_2): Great to
- 15:04
- Unidentified speaker (speaker_1): see you too. I wish we
- 15:05
- Unidentified speaker (speaker_2): could be
- 15:06
- Unidentified speaker (speaker_1): doing it under better auspices than what we're seeing in the market currently. And I do want to just start off with a few names, namely the ones that I mentioned in the intro. So Iron and Nebbius are delaying some of their revenue targets and their stocks are getting punished. Core weave is faring a little better on their forecasting. They've moved some revenue forward, but they're also getting
- 15:25
- Unidentified speaker (speaker_1): hammered down a lot from their highs. What's happening to the neo clouds right
- 15:30
- Unidentified speaker (UNKNOWN): now?
- 15:31
- Unidentified speaker (UNKNOWN): Yeah,
- 15:32
- Unidentified speaker (speaker_3): that'll say um I mean the the market kind of starting to worry about the delivery execution and and the debt problem of new clouds. Um I think I think there's definitely there will some be some collection correction happening in the space because like there are 300 plus new clouds, everyone have different like
- 15:52
- Unidentified speaker (speaker_3): financing history, they have different balance sheet, they have different execution like uh capabilities. Um but then I think um uh
- 16:01
- Unidentified speaker (UNKNOWN): the
- 16:01
- Unidentified speaker (speaker_3): fundamentally AI infrastructure is needed. If everyone believes that AI is gonna be the main driver for the the world's economy, then you need you need all this infrastructure to be happened uh that to kind of actually power power the AI space to move forward. And so that's that's
- 16:21
- Unidentified speaker (speaker_3): why I feel like yeah, like it's an uh like temporary correction, but that um uh
- 16:26
- Unidentified speaker (UNKNOWN): uh
- 16:26
- Unidentified speaker (speaker_3): ultimately. we're gonna need a lot more compute.
- 16:31
- Unidentified speaker (speaker_0): So uh this is against the backdrop of NeoCloud spending tens and tens of billions of dollars bringing new capacity online.
- 16:41
- Unidentified speaker (speaker_0): And I've seen some people concerned about GPU oversupply. Supply. Do you see oversupply happening or is demand still just the story just outpacing it?
- 16:51
- Unidentified speaker (UNKNOWN): Yeah,
- 16:51
- Unidentified speaker (speaker_3): that would say like uh demand is definitely much, much higher than supply at this moment. Even for our own company, like we just like try very, very hard to get extra supply because we just keep having customers who who cannot find compute every uh anywhere. And uh I would say there are like we should separate these two problems. So uh
- 17:11
- Unidentified speaker (speaker_3): I think the the compute supply and demand. are gonna continue to be the mismatch.
- 17:18
- Unidentified speaker (speaker_3): If you look at all the like industry reports, like for example, Epoch AI, they're kind of saying uh inference demand is 10x every year.
- 17:26
- Unidentified speaker (speaker_3): On the other hand, um, I mean, supply, especially with all the news happening right now, it's continued to delay and uh it's not gonna grow as fast as 10x for sure. And so I think there were will naturally there. will be a gonna be a continual be a mismatch for a very long while. Um and and like kind of it means like by based on the supply and
- 17:46
- Unidentified speaker (speaker_3): demand relationship, the the GV price is not gonna go down. And that's what you see on on these different charts, right? Uh I think people more worry about the debt problem of a neo-cloud neo cloud company.
- 17:59
- Unidentified speaker (speaker_3): They may be uh the they delay the
- 18:02
- Unidentified speaker (UNKNOWN): uh
- 18:02
- Unidentified speaker (speaker_3): delay the delivery of the clusters, which means that they'll still need to pay the debt interest, but they they don't have any income. And so uh if they don't execute fast and they sometimes, if they break the SLA, they're gonna refund the customers.
- 18:19
- Unidentified speaker (UNKNOWN): And
- 18:19
- Unidentified speaker (speaker_3): that also means there is some risk
- 18:21
- Unidentified speaker (UNKNOWN): for
- 18:21
- Unidentified speaker (speaker_3): the neo clouds cast itself. And then let's say if a customer who defaults default in the future. there are a lot of startups who are using compute, but then they sometimes when they sign the contract, they actually don't have the balance sheet
- 18:34
- Unidentified speaker (UNKNOWN): to
- 18:34
- Unidentified speaker (speaker_3): cover the full payment. What they need is that they just want to get a compute, they want
- 18:39
- Unidentified speaker (UNKNOWN): to
- 18:39
- Unidentified speaker (speaker_3): build fast, experiment fast, and then once they hit the next milestone, they'll be able to raise more money. Or once they unlock some PMF, they will be able to generate revenue. But it's always a chance, right? A re There's a risk. here. And so if they end up not paying the contract,
- 18:55
- Unidentified speaker (UNKNOWN): then
- 18:55
- Unidentified speaker (speaker_3): neocons will kind of s like don't have the revenue stream. And that will make them like hard to pay back the the debt.
- 19:06
- Unidentified speaker (UNKNOWN): But
- 19:06
- Unidentified speaker (speaker_3): it doesn't mean that the infrastructure is gonna go away, right? Then basically the landers lenders gonna take back the cluster and they'll still need to uh put it into use and like kind of uh supporting the the global like AI AI like uh development. So um I think it's we'd like the world need to separate these
- 19:26
- Unidentified speaker (speaker_3): two problems and um and then I think that's how our view and hyperbolics just try to build solutions to make sure that
- 19:33
- Unidentified speaker (UNKNOWN): the
- 19:34
- Unidentified speaker (speaker_3): bubble is not gonna burst and everyone can monetize compute at a maximum way.
- 19:38
- Unidentified speaker (speaker_1): I I I've got uh two questions here uh for you, Jasper. One kind of building on what you just said, d w do you think that the risk for the customers for Core Weave to default on those SLAs and not give them the payments that they need to service the debt, do you think that that's a real risk currently? Uh do you think that the market sees that as a real
- 19:58
- Unidentified speaker (speaker_1): risk going into twenty 2027, twenty seven or are they just pricing and uncertainty for that happening?
- 20:04
- Unidentified speaker (speaker_3): Yeah, I mean, um, everything is probability. Everything is risk. There it it like they if if if that thing already happened, then the always price is gonna be much, much lower, right? Like and so um everyone I want just to think about there is a risk here, there's a chance that it might
- 20:24
- Unidentified speaker (speaker_3): happen.
- 20:25
- Unidentified speaker (UNKNOWN): Uh even
- 20:26
- Unidentified speaker (speaker_3): people have their different. model to estimate the risk and estimate estimate the impact to COVID. And uh
- 20:34
- Unidentified speaker (UNKNOWN): Yeah,
- 20:34
- Unidentified speaker (speaker_2): I
- 20:35
- Unidentified speaker (UNKNOWN): sorry, I
- 20:35
- Unidentified speaker (speaker_1): guess I should have asked um a little more clearly for you. Do you think it's likely over the next few years that that happens given what you've seen on the demand side?
- 20:43
- Unidentified speaker (speaker_3): No, I I don't I don't think uh that would will happen to likely that would will not gonna happen to COVID. Call with. Cove
- 20:49
- Unidentified speaker (UNKNOWN): Call
- 20:49
- Unidentified speaker (speaker_3): with like what they said on their earnings, they have a lot of uh prepayments, like 70% plus. And uh a lot of their customers uh like
- 20:59
- Unidentified speaker (UNKNOWN): even
- 20:59
- Unidentified speaker (speaker_3): they are not IG grade, they're like like much, much bigger balance sheet than kind of bigger, bigger companies. Um there might be a chance that they like some will default, but even in that case, given how strong the demand is. they'll easily find other customers. And that's that's how I think uh other
- 21:20
- Unidentified speaker (speaker_3): neocons are gonna like we're gonna
- 21:22
- Unidentified speaker (UNKNOWN): think
- 21:23
- Unidentified speaker (speaker_3): about that way as well. And hopefully the investors will also think about that
- 21:26
- Unidentified speaker (UNKNOWN): way.
- 21:27
- Unidentified speaker (speaker_0): Wanna talk about uh long term GPU contract prices and uh on demand. So for example, like Iron says that it's long long-term term GPU contract prices are rising.
- 21:39
- Unidentified speaker (speaker_0): but these long term deals can trade a discount to on demand.
- 21:44
- Unidentified speaker (speaker_3): Yeah.
- 21:45
- Unidentified speaker (speaker_0): This makes sense. But you said an interesting place in the market, hyperbolic, interesting kind of matching here. What do you think is the best indicator of real demand between on demand and long term contract pricing? Do you look at a hybrid? How do you think about measuring
- 22:05
- Unidentified speaker (speaker_0): demand?
- 22:06
- Unidentified speaker (speaker_3): Yeah, I would say uh price an indicator of the relationship between supply and demand. Uh I think different
- 22:17
- Unidentified speaker (speaker_3): contract style represent different types of demand. Uh so on demand represents represent like how immediately, immediate need how much. immediate need that people need for like for kind of yeah, like the the compute, special inference, right? And then for long-term contracts, our view it more like a
- 22:37
- Unidentified speaker (speaker_3): steady use case. Uh and then maybe represent like training and long and long-term, like a more like bigger load of inference. Um the
- 22:49
- Unidentified speaker (UNKNOWN): if
- 22:49
- Unidentified speaker (speaker_3): if you look at the the price the on-demand price and uh like long-term contract price there is a mark uh a ratio right there is a premium and it kind of adjusts along the way a change change along the way um i view that ratio to
- 23:09
- Unidentified speaker (speaker_3): represent how illiquid the market is
- 23:13
- Unidentified speaker (UNKNOWN): so
- 23:13
- Unidentified speaker (speaker_3): if let's say if um H uh BB200 is like $4.5 for long-term three years, but then um like on demand is like seven, eight dollars. The ratio is high and it kind of means that the market is illiquid.
- 23:33
- Unidentified speaker (speaker_3): So because of that uh you like we our company we're trying to provide that liquidity to the world, and I think other people also, the market also gonna incentivize those other players to provide short term. And then as it kind of goes down to uh like let's say 1.3, 1.4, then at this moment
- 23:53
- Unidentified speaker (speaker_3): it's getting more balanced because there's some under utilization of on demand, then it kind of become a healthy ratio and still. ew.
- 24:02
- Unidentified speaker (UNKNOWN): I
- 24:03
- Unidentified speaker (speaker_3): I'm gonna like you try
- 24:04
- Unidentified speaker (UNKNOWN): to
- 24:05
- Unidentified speaker (speaker_3): use both.
- 24:07
- Unidentified speaker (speaker_1): So going uh looking at another part of the market where we're they're trying to find balance, you've argued that compute shortage is partly a coordination failure.
- 24:15
- Unidentified speaker (UNKNOWN): Uh
- 24:15
- Unidentified speaker (speaker_1): if Core Weav and Nevius are spending like, you know, twenty seven or like thirty billion to forty-five forty five billion dollars a year and they're still short, where is that idle capacity actually sitting?
- 24:27
- Unidentified speaker (speaker_3): You're kind of trying to it says really align with our pitch. So, like the vision, I mean I should just tell uh I given that uh we'd love to kind of share a little bit of analogy that how we think
- 24:39
- Unidentified speaker (speaker_0): yeah, go ahead.
- 24:40
- Unidentified speaker (speaker_3): All right, so compute is electricity, it's the new electricity for AI. Um GPU clusters are generators, neo clouds are powerful. power plants. So if you have a house,
- 24:54
- Unidentified speaker (UNKNOWN): you
- 24:54
- Unidentified speaker (speaker_3): don't sign a PPA contract with the power plant for electricity. You tap into the grid and then you draw whatever electricity that you need and then you pay whatever you use. Even when you're bigger, you will still have you, let's
- 25:06
- Unidentified speaker (UNKNOWN): say
- 25:06
- Unidentified speaker (speaker_3): your factory you will have a PPA to guarantee the base load, but you still need to tap into the grid to get extra electricity if you need. And then if you don't need that, you will sell back your idle electricity from your PPA. And
- 25:21
- Unidentified speaker (UNKNOWN): the
- 25:22
- Unidentified speaker (speaker_3): same here, right? Like right now, a lot of people, uh, all the AI companies, they are forced to sign PPA contract
- 25:31
- Unidentified speaker (speaker_3): with a neocast, aka power plant. plants And no one, just like electricity, no one's usage of compute is stable everyone is like this
- 25:43
- Unidentified speaker (UNKNOWN): so
- 25:44
- Unidentified speaker (speaker_3): there's always a mismatch between the contract and the real demand
- 25:48
- Unidentified speaker (UNKNOWN): and
- 25:49
- Unidentified speaker (speaker_3): so our our goal is I want to build the first global AI compute grid that allows people to able to draw compute elastically Basically, and then also sell back the idle compute
- 25:59
- Unidentified speaker (UNKNOWN): but
- 26:00
- Unidentified speaker (speaker_3): then that's kind of answer your question like because the fundamentally of the using you, the use usage of compute is never stable. So there will always idle compute happening in different layers. Let's say, okay, first, uh, on the core wave and nebula side, it's really hard to do this
- 26:20
- Unidentified speaker (speaker_3): pipping packing to make sure that they sell all the compute at all times. So there is idle compute there.
- 26:26
- Unidentified speaker (UNKNOWN): And even
- 26:26
- Unidentified speaker (speaker_3): though that they sell three year contract to a customer.
- 26:29
- Unidentified speaker (UNKNOWN): And
- 26:29
- Unidentified speaker (speaker_3): so that basically, let's say, assume like Nebula and Cole have high utilization on their compute. But then now the this like risks or like this liability now happen on the company side. The company is also not gonna use all of their three year contracts.
- 26:45
- Unidentified speaker (UNKNOWN): And
- 26:45
- Unidentified speaker (speaker_3): so they will also have idle compute up there. And I mean, we've been talking about this like for one or two. uh multiple years, that we think that the world needs to build more data centers, need
- 26:55
- Unidentified speaker (UNKNOWN): to
- 26:55
- Unidentified speaker (speaker_3): build more triple clusters, but also we need to improve the utilization of compute.
- 27:03
- Unidentified speaker (UNKNOWN): We
- 27:03
- Unidentified speaker (speaker_3): don't know why we want to waste electricity.
- 27:05
- Unidentified speaker (UNKNOWN): But
- 27:06
- Unidentified speaker (speaker_3): then we're wasting compute. Just like it's nothing. And on average, I I'll expect like it's gonna be less. than 50% utilization of compute in the world.
- 27:18
- Unidentified speaker (UNKNOWN): And
- 27:18
- Unidentified speaker (speaker_3): if we can actually reutilize those leveraging technology,
- 27:22
- Unidentified speaker (UNKNOWN): then
- 27:22
- Unidentified speaker (speaker_3): gonna solve as in some sense, like kind of like reduce the compute shortage, uh in a like in some in some way. Yeah.
- 27:32
- Unidentified speaker (UNKNOWN): So
- 27:32
- Unidentified speaker (speaker_0): you give this great analogy. First of all, one of my favorite analogies I've heard on the show so far. Um
- 27:39
- Unidentified speaker (speaker_0): that said, I wanna I wanna give you uh a Nebius and Core Weave business model here. So Nebius does more like on-demand spot rentals. Core Weave has been focusing more on like take or pay long-term contracts.
- 27:54
- Unidentified speaker (speaker_3): Yeah,
- 27:54
- Unidentified speaker (speaker_0): the market, because it does have this structural dislocation because of the analogy you just gave. That said, which of these companies do you think is best positioned from a business model standpoint, just alone, just alone on demand or on demand versus uh long-term contracts? Do you think one or the other is a
- 28:14
- Unidentified speaker (speaker_0): better structure right now, or is it just too complex to get into?
- 28:18
- Unidentified speaker (UNKNOWN): Yeah,
- 28:19
- Unidentified speaker (speaker_3): that'll say like uh it there like Like that that there are some like uh similarities between Navios Nebus and Cold Covey With and that there are also like some differences. Uh it's definitely not just as simple as like short term and long term. I think I I would say like if you look at Nebus uh compute, like their like their how how they sell the compute, I
- 28:39
- Unidentified speaker (speaker_3): I would say still majority of their compute are optic,
- 28:42
- Unidentified speaker (UNKNOWN): uh
- 28:42
- Unidentified speaker (speaker_3): long term.
- 28:43
- Unidentified speaker (UNKNOWN): Uh
- 28:44
- Unidentified speaker (speaker_3): but they're starting Start doing like short short-term term for sure. Like, but then if you look at like their meta contract, okay, Microsoft contract, this it's like long-term, it's even probably five years or even longer, right? Uh I would say uh having a mix of long-term and short-term, uh, it seems uh a better strategy,
- 29:04
- Unidentified speaker (speaker_3): so that you can still have high credit customers
- 29:09
- Unidentified speaker (UNKNOWN): to
- 29:09
- Unidentified speaker (speaker_3): underwrite, uh, to underwrite for your long-term contract, so that you can still make sure your down your downsign downside is. protected but protected but then also you'll be able to sell uh short term to kind of capture the higher higher value of the of the compute. Um I think uh Nebula's also moving faster
- 29:29
- Unidentified speaker (speaker_3): on building their AI solutions, including their like inference, right? They uh and then they also kind of just risk recently launched their like uh uh post training solutions.
- 29:42
- Unidentified speaker (UNKNOWN): And
- 29:43
- Unidentified speaker (speaker_3): and so I I think gradually all these new clouds wanted to kind
- 29:46
- Unidentified speaker (UNKNOWN): of
- 29:47
- Unidentified speaker (speaker_3): build higher higher value added services uh to capture higher upside of uh GPU hour. In the end, uh it's all about how how much you can make per GPU hour. So they uh uh I think I need to move faster and I know I know Call of the is also making making uh progress on
- 30:06
- Unidentified speaker (speaker_3): that end. Uh but right now I think Nabulas kind of have a better t software suite.
- 30:14
- Unidentified speaker (UNKNOWN): Well
- 30:14
- Unidentified speaker (speaker_1): let's hope that uh those stock prices can catch up to those GPU hours because I know we got a lot of listeners who are wondering
- 30:20
- Unidentified speaker (UNKNOWN): what the
- 30:20
- Unidentified speaker (speaker_1): hell is happening to the market. But seems like everything's kind of taking a beating today. We're gonna be covering anthropic on our new next section. Maybe that'll save it, but gonna let you go for now, Jasper. Thank you so much for joining. Hope you have a good weekend, man, and uh best of luck with everything y'all are building over there at Hyperbolic.
- 30:38
- Unidentified speaker (speaker_3): Thanks a lot.
- 30:39
- Unidentified speaker (UNKNOWN): Thank you,
- 30:39
- Unidentified speaker (speaker_0): Jasper.
- 30:42
- Unidentified speaker (UNKNOWN): We're
- 30:43
- Unidentified speaker (speaker_0): gonna keep on rolling. Let's cover the anthropic IPO. We'll get to that after a word from our sponsor, LIGO. Ligos.
- 30:57
- Unidentified speaker (speaker_0): This episode is brought to you by Ligos Finance. Hedge funds are getting liquidated. Is your Bitcoin safe? Everybody's getting hacked. Binance. But think back to FTX and Celsius. Don't be the next victim of a counterparty risk.
- 31:14
- Unidentified speaker (UNKNOWN): If you
- 31:15
- Unidentified speaker (speaker_0): are working with another loan provider, do yourself a favor before it's too late. Check out Ligos Finance, which is Blockspace's preferred non-custodial Bitcoin lender using Bitcoin native smart contracts to protect your stack. With Ligos, you always know where your Bitcoin is. Hold your keys, no wrapping, no bridging, no
- 31:34
- Unidentified speaker (speaker_0): rehypothecation. Get competitive rates as low as 10% APR. For more information, go to Legos.finance.
- 31:47
- Unidentified speaker (speaker_1): Anthropic's IPO is six times oversubscribed as Wall Street races for a piece of the first publicly traded AI lab in history. Bloomberg first reported that Anthropic has seen overwhelming demand on the offering that could raise as much as $90.4 billion, including $26 billion from a
- 32:07
- Unidentified speaker (speaker_1): private placement and $64.4 billion from the IPO offering itself, which would make it the largest IPO in history, a record currently held by SpaceX, which was set earlier this year. The raise would give Anthropic a $1.4 trillion valuation with a $70 to $78 share price.
- 32:27
- Unidentified speaker (speaker_1): and the immense 6X demand is sending a mixed message about investor appetite for AI. Indeed, it comes as a at a time when AI stocks are bleeding from a multi-month route that has shed six trillion dollars from their collective market caps, according to MSCI research. Charlie, here's
- 32:47
- Unidentified speaker (speaker_1): how I see it, man. Anthropic's IPO is either going to save or break the market, or it's going to suck up all the liquidity from everything else as investors and traders look to something else to juice their bags with everything else down.
- 33:03
- Unidentified speaker (UNKNOWN): So,
- 33:03
- Unidentified speaker (speaker_1): what's your take? What do you think is gonna happen when it IPOs? Do you think we're so back, or do you think people are just gonna be rotating into an AI lab?
- 33:10
- Unidentified speaker (speaker_0): Why not both? I mean, it's a whole week away from today, so which is two or three. narrative cycles worth of narratives in AI. So we can be so back and so over a couple more times before that IPO goes live. It certainly feels like a lot of the money is exiting, or D you might even say just
- 33:30
- Unidentified speaker (speaker_0): rotating from
- 33:31
- Unidentified speaker (UNKNOWN): the
- 33:32
- Unidentified speaker (speaker_0): AI trade into anthropic. It's been the one everybody's looking forward to. It's the first frontier lab. uh IPO. So uh I'm not surprised.
- 33:44
- Unidentified speaker (UNKNOWN): I
- 33:44
- Unidentified speaker (speaker_0): you love to see it. If there's any hope in this dark route across all AI stocks and Jensen getting ousted and NVIDIA down, it's the fact that our North Star anthropic is six times oversubscribed.
- 33:59
- Unidentified speaker (UNKNOWN): I
- 33:59
- Unidentified speaker (speaker_0): don't know what happens after it IPOs, but Space. SpaceX has done pretty well despite huge share lockups. In fact, it's roughly parody with its IPO price, which actually is pretty good considering only six percent shares were circulating at the outset. I think Anthropic will be a great bellwether. I have a hard time seeing how it
- 34:19
- Unidentified speaker (speaker_0): doesn't turn out great in the first few days, given the fact that it's massivers. So we'll have to
- 34:24
- Unidentified speaker (UNKNOWN): see.
- 34:25
- Unidentified speaker (speaker_1): Yeah, I would imagine. that we'd probably see a pop, and we have more on that in a second with what the DGENS on Hyperliquid are pricing in. But first, I want to get a few more numbers here from the Bloomberg article. So, first off, one of the ones that stood out to me, Anthropics annualized revenue run rate has risen 25%
- 34:44
- Unidentified speaker (speaker_1): from 65 sixty five billion at the end of July to 81 eighty one billion at the end of October. That puts its valuation of one point four trillion at roughly seventeen point three times its ARR. And I think it's important to note here that you could argue that the market is assigning a multiple to the best number it can find. Typically
- 35:04
- Unidentified speaker (speaker_1): you'd want to use something like a conventional trailing twelve 12 month revenue multiple because that eighty 81 one billion has not been realized, right? That's projecting out. revenue based on what they are currently earning for the next 12 months. So that being said, you got to take what you can get. And a lot of investors are
- 35:24
- Unidentified speaker (speaker_1): taking anything that they can get on this one for sure. If we take a look at that private placement, Charlie, Nvidia and Google each invested $10 billion into In the private placement, and uh AMD invested $2.5
- 35:38
- Unidentified speaker (UNKNOWN): million.
- 35:39
- Unidentified speaker (speaker_1): billion. Broadcom, Microsoft, and Micron each invested $1 billion and TSNC getting a piece here, investing half a billion in that $26 billion private placement as well. Bloomberg also reported that investors are being prioritized for share allocation in the oversubscribed raise. They warned that prospective
- 35:59
- Unidentified speaker (speaker_1): buyers may not receive or may or will probably receive smaller allocations. than they requested. Makes sense. It's six times oversubscribed, but existing investors specifically are being prioritized over new ones for the raise. Anthropics prospectus also revealed that the company plans to issue for this 801.35 million new shares in the offering,
- 36:19
- Unidentified speaker (speaker_1): while existing shareholders, which includes emplo employ employees, will sell 23.65 twenty-three point six five million shares during the offering existing. shares, not new ones. Ultimately, I think that if they hit the upper bound of ninety-five million or excuse me, ninety billion, ninety point four billion in this, or even if they
- 36:39
- Unidentified speaker (speaker_1): get eighty-five billion from this raise, that's going to have a huge impact on obviously their massive capex spend that they're planning over the next few years. Roughly eighty-five $85 billion in that price. net proceeds at the IPO midpoint would give Anthropic substantial liquidity for compute commitments and reduce its near-term uh
- 36:58
- Unidentified speaker (speaker_1): reliance on borrowing, which is very important. They're still not investment grade, not a publicly traded company. They might not be for some time after they go public either. We'll take a minute to get there, probably. Charlie, they'll definitely get every penny of what they're asking for here, given that it's over. oversubscribed. But they are launching into
- 37:18
- Unidentified speaker (speaker_1): a tough market. And I would remind you, as I said earlier, the D Gens think that they're gonna get like a twenty percent pop in the first few days of trading. Djens pricing in anthropics IPO valuation at just under 18 uh just under 1.8 trillion. It's at about 1.77 trillion currently. Do you
- 37:38
- Unidentified speaker (speaker_1): think it runs like SpaceX after debut?
- 37:41
- Unidentified speaker (speaker_0): Do you even call hyperliquid traders Dejens anymore? Because I would call them sophisticated market participants. So I would take hyper hyperliquid's pre-IPO market here that we sh that we're showing, which is what 1.7 trillion and the and the IPOs at at what uh targeted to be 1.4. I would take hyperliquid as
- 38:01
- Unidentified speaker (speaker_0): a much truer indicator. indication of what the price is. So I at right now I think we're gonna gonna get a bump on the first day. I think I right now, um, based upon all signals that we can have, anthropic is a buy. Obviously, as I said, anything can happen in six days. Thank God two of those
- 38:21
- Unidentified speaker (speaker_0): days are on the weekend. Although I don't think weekends really matter anymore for the markets. Um Uh so uh yeah, I'm optimistic. And it anthropic does kind of have to succeed. Like if it's a complete flop and failure, then I don't know what we do because they they are spending a lot of the a lot of
- 38:41
- Unidentified speaker (speaker_0): the market, a lot of the AI trade depends upon them honoring their compute commitments, which again, the IPO is not the only source of funds for that, but like it's a great bellwether for how much real demand there is in the market, both to own Frontier Labs and to spend and like to predict what the CapEx cycle looks like on the short term.
- 39:01
- Unidentified speaker (speaker_0): Because if the CapEx cycle dies, then
- 39:05
- Unidentified speaker (speaker_0): we're I don't know, we're gonna have to start different
- 39:07
- Unidentified speaker (speaker_1): We're gonna have to pivot to a Zcash Cash podcast. So please, Dario, do not mess this up. Yeah,
- 39:13
- Unidentified speaker (speaker_0): we're gonna have after to we're gonna have after to just be a politics podcast, which you don't want.
- 39:18
- Unidentified speaker (UNKNOWN): I might want it. Yeah. But I don't think
- 39:20
- Unidentified speaker (speaker_1): anyone else does.
- 39:21
- Unidentified speaker (speaker_0): No, yeah.
- 39:22
- Unidentified speaker (speaker_1): I I do think that this it maybe it's overstating it, but if they have a strong IPO, things go well, and it doesn't just immediately tank, then I think that will be a good sign for at least a relief rally from what we're seeing right now. Because I I do have to reiterate.
- 39:39
- Unidentified speaker (speaker_1): multiple names are down double digit days this week. Like it's just double digit day after double digit day. It has been brutal. There was a brief rally earlier in the week, but it's kind of back to business, the bloody business that we saw last week. So if I if anthropics IPO does go over well, we could be looking at
- 39:59
- Unidentified speaker (speaker_1): sunnier skies ahead. That being said, I think we should probably just do a an entire episode on is the AI trade unraveling? 'Cause there's been a lot of warning signs flashing recently, including I would add the fact that gold is selling off. That's not something that you particularly would like to see. Um, if you're thinking about the
- 40:19
- Unidentified speaker (speaker_1): fact that global liquidity is in a good place. So there are a lot of things I think that are lingering right now that are spooking investors, but who knows? Maybe the first I AI lab going public in history can save us all. So we'll have to see.
- 40:34
- Unidentified speaker (speaker_0): Look, just think about the last four Iran wars we've had in the past year or two, and think about how absolutely bleak it felt at those times. And then realize we've got four more Iran wars ahead of us. So uh that's uh, you know, that's bullish you if you ask
- 40:49
- Unidentified speaker (UNKNOWN): me.
- 40:51
- Unidentified speaker (UNKNOWN): I
- 40:51
- Unidentified speaker (speaker_0): kid, I kid, I kid. We're gonna wrap it up before we get too far down a rabbit hole and we get demonetized for saying the four-letter word Iran. Thank you for listening to Block Space Live Compute's daily live show, hosted by Charlie Spears and Colin Harper. We are with Blockspace Media. If you want more Blockspace content, go to our website,
- 41:11
- Unidentified speaker (speaker_0): blockspace.media. And if you made it to the end of the show, I have a gift for you. Our group chat on X is still open. You can find the link to that in our Twitter bio or X bio as they call it these days. Go join that. Would love to chat and argue with you
- 41:31
- Unidentified speaker (speaker_0): in our group chat. Thank you for listening. This wraps up this week. Hoping for a better week in the markets. We'll see you. Monday. I'm Charlie. I'm Colin. Adios.
More from Blockspace: AI News
JPMorgan Slashes IREN Target and Banks Revalue AI Stocks. Plus, Will Jensen Return to NVIDIA?
Episode discussion: Wall Street cuts AI stock targets amid execution delays; Luxor CEO discusses AI cyberattacks
Speakers’ claims: Hosts analyze JPMorgan's reduction of IREN's price target to $55, attributing the move to delays at IREN's Childress data center and rising financing costs. Luxor CEO Nick Hansen joins to detail an AI-enabled cyberattack on his fleet management software, noting a shift in internal security resource allocation from 10-20% to 70-80%. The episode covers broader sector repricing by Goldman Sachs, Morgan Stanley, and Bank of America for stocks like IREN, Terawulf, and Nebius, citing execution risks and interest rates rather than fundamental AI demand collapse. Hansen criticizes Ocean mining pool for refusing to assist hack victims, arguing their 'permissionless' stance is a choice rather than a technical limitation. Finally, speakers argue that 'PowerShells' hold more residual value than 'Neoclouds' because physical power infrastructure is harder to replace than compute capacity.
Jensen Huang Resigns as NVIDIA CEO. Plus Riot's $28B Corsicana Lease and Nebius' Q3 Earnings
Episode discussion: Jensen Huang's resignation, Riot's $28B Corsicana lease, and Nebius Q3 earnings discussed
Speakers’ claims: Hosts and guests discuss the resignation of NVIDIA CEO Jensen Huang, with Jay Patel arguing Wall Street underestimates his role in securing neo-cloud financing. The episode analyzes Nebius's Q3 results, noting a massive revenue jump offset by stock declines due to non-recurring revenue and delayed targets. Riot Platforms CEO Jason Les details a $28 billion Corsicana lease utilizing a modular infrastructure model to reduce costs. The discussion covers financing structures, yield on cost, and Riot's vertical integration in data center engineering.