AI trade shifts from capital war to narrative war, favoring longs via short squeeze mechanics
Summary
Russell argues the AI investment trade has evolved from a 'capital war' to a 'narrative war,' creating structural advantages for long positions. He highlights a disconnect between credit markets and operational fundamentals, noting Oracle's sales growth despite high CDS spreads and a 60% stock drop. The episode posits that high short interest can trigger squeezes that reduce dilution needs, allowing companies like SpaceX to raise massive capital. Speakers suggest that bearish narratives often backfire by creating upward liquidity pressure that invalidates short theses.
Episode sections
- 0:00 – 1:14
Introduction to AI Trade Shift
Russell introduces the shift from a capital war to a narrative war in AI markets.
- 1:14 – 3:32
SoftBank and Oracle CDS vs. Equity Disconnect
Contrasts SoftBank's rising CDS with Oracle's high CDS and growing sales despite stock drops.
- 3:32 – 5:23
Oracle Capex Reduction and OpenAI Linkage
Discusses Oracle's capex cuts and the triangular dependency between SoftBank, OpenAI, and Oracle.
- 5:23 – 6:54
OpenAI Financials and Capital Needs
Cites ChatGPT-scraped figures suggesting OpenAI needs $300 billion, highlighting market skepticism.
- 6:54 – 8:04
SpaceX Capital Raise and IPO Market Capacity
Uses SpaceX's $100 billion raise to argue equity markets can support massive AI infrastructure.
- 8:04 – 9:09
Narrative War Dynamics and Bearish Risks
Explains how dominant bearishness can create conditions for short squeezes in IPO markets.
- 9:09 – 11:48
Tesla Short Squeeze Mechanics
Details how short squeezes reduce dilution needs, invalidating short theses based on capital raising.
- 11:48 – 14:08
MicroStrategy and Bitcoin Treasury Dynamics
Illustrates how high short interest allowed MicroStrategy to fund Bitcoin purchases at a premium.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| OpenAI | private company | bearish | Um now, what links, in case you've been living under a rock, what links Softbank uh and OpenAI? So SoftBank is uh sorry, what links SoftBank and Oracle. So SoftBank is the biggest investor into OpenAI, they give OpenAI capital, and then OpenAI is Oracle's biggest customer, so they're all sort of linked together. |
| SpaceX | private company | neutral | The problem you've got is like, and this year, and we saw this, uh, seen this year a few times, it's a company like SpaceX says, Oh, we're gonna go into the AI business as well. Uh, they've got huge growth ambitions. |
| TeslaTSLA | equity | neutral | Uh and Tesla is probably the most relevant example here. So when it first listed, it was obvious that Tesla needed more capital, and it wasn't obvious that they would become the dominant EV company that they are today. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
SoftBank was offering very high yields on its bonds in Japan to raise capital for investments in OpenAI.
Unidentified speaker ·1:23
How to verify: Check SoftBank Group Corp bond prospectuses or yield data from Bloomberg/Reuters for Japanese Yen-denominated bonds issued in 2024-2025.
SoftBank's CDS has been rising while its shares have traded higher, showing a disconnect between equities and credit.
Unidentified speaker ·1:48
How to verify: Compare SoftBank Group Corp (9984.T) stock price trends with its CDS spread data from Markit or Bloomberg over the last 5-6 weeks.
Oracle's CDS has exceeded the highs seen in 2008.
Unidentified speaker ·2:18
How to verify: Check Oracle Corp (ORCL) CDS spread data from Markit or Bloomberg and compare against historical 2008 CDS levels for Oracle or comparable investment-grade issuers.
Oracle's share price is down 60% from its peak.
Unidentified speaker ·2:37
How to verify: Calculate the percentage decline from Oracle's (ORCL) all-time high to its current price using Yahoo Finance or Bloomberg.
Oracle's sales are growing at its most rapid pace since it first listed.
Unidentified speaker ·3:12
How to verify: Review Oracle's (ORCL) quarterly earnings reports (10-Q/10-K) for revenue growth rates and compare against historical data since its IPO in 1986.
OpenAI needs to find $300 billion of capital to invest over the next few years.
Unidentified speaker ·5:42
How to verify: Check OpenAI's public statements, press releases, or filings regarding capital requirements and Capex plans.
Anthropic saw $17 billion in Q3 sales.
Unidentified speaker ·5:42
How to verify: Verify Anthropic's revenue figures from their official announcements, press releases, or credible financial news reports covering Q3 2024/2025.
SpaceX raised $100 billion and executed one of the largest IPOs of all time.
Unidentified speaker ·7:12
How to verify: Check SpaceX's funding rounds and IPO details from financial news sources like Bloomberg, Reuters, or SEC filings if applicable.
The global IPO market raised $600 billion in 2021.
Unidentified speaker ·7:51
How to verify: Check global IPO volume data from Refinitiv, Bloomberg, or Dealogic for the year 2021.
MicroStrategy shares traded at a premium to NAV when short interest was high, allowing the company to sell shares to buy Bitcoin.
Unidentified speaker ·12:20
How to verify: Compare MicroStrategy (MSTR) stock price to its Bitcoin holdings' net asset value (NAV) during periods of high short interest, and check SEC filings for share issuances used for Bitcoin purchases.
Do your own research
- Verify Oracle's current quarterly sales growth and CDS spread levels against Bloomberg or Refinitiv data.
- Check OpenAI's latest funding rounds and capital raise announcements to validate the $300 billion need claim.
- Review SpaceX's recent equity raise details and IPO market capacity data for 2021 versus current conditions.
- Analyze MicroStrategy's historical share issuance premiums relative to NAV during periods of high short interest.
- Assess SoftBank's current equity performance and CDS movements to confirm the disconnect with Oracle.
- Investigate ChatGPT-scraped financial figures cited for OpenAI's capital requirements for source verification.
Full transcriptexpand
- 0:00
- Unidentified speaker: Hi guys, Russell here.
- 0:02
- Unidentified speaker: Uh I have I feel compelled, or have been compelled, uh to take a closer look at the AI trade.
- 0:10
- Unidentified speaker: Um market actions in the last sort of five, six weeks have been so wildly out of sync with what I would have expected from what I've seen.
- 0:20
- Unidentified speaker: Uh that you know I've sort of been going through, you know, what could have caused it.
- 0:25
- Unidentified speaker: Is there something there?
- 0:26
- Unidentified speaker: Is there something I need to fit into sort of my model of how the world works?
- 0:31
- Unidentified speaker: Um, and so I've gone back and looked at AI.
- 0:34
- Unidentified speaker: I've looked at it from time to time.
- 0:36
- Unidentified speaker: Always felt like a sort of a bit a bit funky, if that's the right word.
- 0:41
- Unidentified speaker: And it's still a bit funky, but I took a closer look uh uh you know today and yesterday, looking over various different things.
- 0:51
- Unidentified speaker: And what I see with AI is it's interesting, it's become a very interesting trade, uh, and I think it's gonna be very problematic for a lot of people.
- 1:01
- Unidentified speaker: And what I've what I've seen is AI is traded from what I used to call a capital war almost to a narrative war.
- 1:08
- Unidentified speaker: I'm gonna explain what that means and why it's actually very bad news, particularly for the shorts, in my view.
- 1:14
- Unidentified speaker: Um, so what do I mean by that?
- 1:16
- Unidentified speaker: Well, let's sort of go through it.
- 1:18
- Unidentified speaker: Um, so earlier this year, I actually thought Softbank was probably worth a short.
- 1:23
- Unidentified speaker: And the reason for that was you know, you could see it it was offering very high uh yields on its bonds to sell in Japan to raise capital uh to invest into OpenAI and a couple of other stocks as well.
- 1:35
- Unidentified speaker: And when we looked at CDSs, it looked pretty poor.
- 1:39
- Unidentified speaker: Now, usually, and you should be aware of this, when you see a rising CDS, that's normally associated with equity weakness.
- 1:48
- Unidentified speaker: Okay.
- 1:49
- Unidentified speaker: Um, but in SoftBank's view, what we've seen is shares have tended to actually trade higher if extremely volatile.
- 1:57
- Unidentified speaker: So even though CDS has been rising, the shares of SoftBank have been giving you nothing.
- 2:03
- Unidentified speaker: There's a bit of a disconnect between the equities and the CDS there.
- 2:07
- Unidentified speaker: Um now this is in contrast to Oracle, uh, where equity valuations have gotten crushed as CDS has risen.
- 2:18
- Unidentified speaker: So if we look at Oracle CDS, they've exceeded even the 2008 high, which uh for people from 2008, you've thought I must say for a long time I never thought we'd ever see any CDSs above the 2008 highs.
- 2:31
- Unidentified speaker: And here is Oracle above that number, so it's sort of implying severe financial distress.
- 2:37
- Unidentified speaker: And with Oracle's case, you can see the share price is down 60% for our peak.
- 2:42
- Unidentified speaker: Uh, to be honest with you, you know, given the move in the CDS, it probably should be down more, should be below uh levels we saw two, three years ago.
- 2:51
- Unidentified speaker: But you know, like I said, the market's a bit funky here.
- 2:54
- Unidentified speaker: Uh speaking of Oracle's share price is down.
- 2:57
- Unidentified speaker: Now, if you were just looking at the CDS of Oracle and looked at historic of history of CDSs, you would assume that dot-com bust was on us.
- 3:07
- Unidentified speaker: Like uh it was uh you know, revenue was falling, prices falling.
- 3:12
- Unidentified speaker: Uh what's weird is when you look at Oracle's sales, uh for you know they're growing quite rapidly.
- 3:18
- Unidentified speaker: Most rapid play uh piece of sales growth since maybe it first listed, so it's actually growing, right?
- 3:26
- Unidentified speaker: But the market doesn't like this growth.
- 3:29
- Unidentified speaker: This is sort of what it's saying.
- 3:32
- Unidentified speaker: Now, what's most interesting is in this most recent quarter, the CEFO came out with a new sort of capex plan, no doubt in response to the pressure that it's uh coming under, and they've done a number of sort of wheezers to reduce the capex uh from Oracle for this investing cycle.
- 3:51
- Unidentified speaker: Now I know people are gonna come back and me say they said they were gonna do this up capex before, blah blah blah.
- 3:56
- Unidentified speaker: And I also say some of the things they're doing here, vendor financing, a lot of people hate vendor financing.
- 4:02
- Unidentified speaker: Uh, bring your own hardware so you they're gonna build the data center and then you you put the GPUs in it, and they do prepayments, a whole bunch of things basically get cash outflows down because the market is sort of saying we don't want you to spend that much.
- 4:17
- Unidentified speaker: I think that generally speaking, you see this that that's a sort of semi-bullish signal.
- 4:22
- Unidentified speaker: We see a market responding to that, you know.
- 4:25
- Unidentified speaker: Uh and so far we haven't seen anything.
- 4:28
- Unidentified speaker: Um now, what links, in case you've been living under a rock, what links Softbank uh and OpenAI?
- 4:39
- Unidentified speaker: So SoftBank is uh sorry, what links SoftBank and Oracle.
- 4:42
- Unidentified speaker: So SoftBank is the biggest investor into OpenAI, they give OpenAI capital, and then OpenAI is Oracle's biggest customer, so they're all sort of linked together.
- 4:52
- Unidentified speaker: Um so you know, so they should, in theory, move together.
- 4:58
- Unidentified speaker: So we're sort of looking at this going, okay, it sort of implies that open AI is problematic insofar as valuing it and capital.
- 5:08
- Unidentified speaker: Um, and so you know, that's really the question we're all asking ourselves here is that can open AI get the money or the capital it needs for the investing plans that's put out uh out into the market.
- 5:23
- Unidentified speaker: Um, so I asked ChatGPT to give me the financials of OpenAI.
- 5:27
- Unidentified speaker: I'm sure I just scraped it from the internet somewhere, and they look roughly in line with what I said, what I've been seeing.
- 5:33
- Unidentified speaker: So they assume uh, you know, sales of 36 billion this year, that sounds too low.
- 5:39
- Unidentified speaker: Sales of 70 billion next year, that sounds too low.
- 5:42
- Unidentified speaker: Uh given that you know Anthropics saw 17 billion in Q3 of this year.
- 5:47
- Unidentified speaker: But anyway, you know, numbers are there, and then they're sort of saying the CapEx spend, which is out there, and so you sort of end up with uh, you know, they need to find 300 billion of uh capital to invest over the next sort of uh few years, right?
- 6:04
- Unidentified speaker: So this is these are big numbers, um, you know, and the market is sort of saying actually they're gonna need 100 billion next year, and you know, you so you see the market, what the market is sort of saying uh with CDSs and some of the equities is like we don't think this money's gonna be raised, so these growth forecasts are bunk, uh, and so we need to sort of pretend that's not gonna happen.
- 6:27
- Unidentified speaker: Um, and you know, that is basically where we are, you know, and if you're bearish, it's very easy to say there's no way Anthropic and OI OpenAI can raise the cash they need.
- 6:41
- Unidentified speaker: It's very easy.
- 6:42
- Unidentified speaker: You can make all sorts of graphs.
- 6:44
- Unidentified speaker: Uh, you're basically sort of saying they're gonna have the sort of revenue profile of the Mag 7, that sort of thing.
- 6:50
- Unidentified speaker: You're very easy to make this sort of say it's impossible.
- 6:54
- Unidentified speaker: The problem you've got is like, and this year, and we saw this, uh, seen this year a few times, it's a company like SpaceX says, Oh, we're gonna go into the AI business as well.
- 7:03
- Unidentified speaker: Uh, they've got huge growth ambitions.
- 7:05
- Unidentified speaker: They're model, you know, they don't have, or didn't have anyway, uh, a frontier model like OpenAI Anthropic.
- 7:12
- Unidentified speaker: They bought in you know a bunch of companies, but they ran out of the market and they raised 100 billion.
- 7:17
- Unidentified speaker: And then there's some follow-on capital raises as well.
- 7:21
- Unidentified speaker: Uh, and it's one of the largest IPOs of all time.
- 7:24
- Unidentified speaker: So, you know, what you're left with, if you're seeing here like me, sort of thinking about markets, it's like, okay, the debt markets don't want to give uh OpenAI, Oracle, SoftBank the money they need, but the equity markets are open or could be potentially open.
- 7:41
- Unidentified speaker: Um, you know, and what we've seen in sort of you know, is and it the equity markets can be big enough to fund this growth if we want.
- 7:51
- Unidentified speaker: We can see that the globally global IPO market you know raised 600 billion in in 2021, and markets have gone up quite a lot since then.
- 8:00
- Unidentified speaker: So the capital could be there, it could be there.
- 8:04
- Unidentified speaker: The problem is the problem is that IPO markets are really narrative markets most of the time, particularly for gross stocks.
- 8:12
- Unidentified speaker: Whoever controls the narrative controls the IPO market.
- 8:16
- Unidentified speaker: Um and so the question now that is the market is trying to debate is is the AI a real story or is it a capital destruction machine?
- 8:28
- Unidentified speaker: Right now, the bears are saying it's a capital destruction machine.
- 8:32
- Unidentified speaker: Um and the thing is that uh when you have a sort of narrative war like this, is open AI or AI booming, or is a capital destruction, the bears normally lose in this story when it's a narrative story like this.
- 8:52
- Unidentified speaker: They normally lose.
- 8:53
- Unidentified speaker: Once the bear narrative becomes dominant, almost is the end of you're you're basically about to get crushed by a steamroller.
- 9:03
- Unidentified speaker: That's just my experience, and I've seen it a few times in my past.
- 9:06
- Unidentified speaker: So I'll give you an example of what I mean by that.
- 9:09
- Unidentified speaker: Uh and Tesla is probably the most relevant example here.
- 9:12
- Unidentified speaker: So when it first listed, it was obvious that Tesla needed more capital, and it wasn't obvious that they would become the dominant EV company that they are today.
- 9:20
- Unidentified speaker: And so everyone was like, okay, they're gonna need more capital.
- 9:23
- Unidentified speaker: I don't want to lend someone and I'm gonna short their shares.
- 9:26
- Unidentified speaker: It was very common across sort of hedge fund industry.
- 9:29
- Unidentified speaker: Um, the problem was, and I've put the graph here of short interest in Tesla, is that when you have a large short interest, you're almost always gonna get a short squeeze.
- 9:40
- Unidentified speaker: All right, particularly if the company is growing, right?
- 9:42
- Unidentified speaker: It's non-financial, you know, it's growing, your risk is you're gonna get a short squeeze, some positive news flow, and the shares go up.
- 9:50
- Unidentified speaker: Um, and the thing is that the weird thing, if you look at the financial models that these companies base their shorts on, so Edgeon's base their shorts on, is basically sort of saying, okay, share price at this level of share price, this is the amount of dilution they would need to raise capital.
- 10:06
- Unidentified speaker: So shares should fall this much.
- 10:08
- Unidentified speaker: But if the shares start spiking, the amount of shares I need to issue to raise the capital I need actually start shrinking, which actually creates more upward uh upward pressure on the share price because that dilution risk is going down.
- 10:20
- Unidentified speaker: The higher the shares go up, the less dilution there's going to be.
- 10:23
- Unidentified speaker: I've seen this many times, even all the way back in 2002.
- 10:26
- Unidentified speaker: This is not, I'm using Tesla, but this is not a new, not a new theme in markets.
- 10:32
- Unidentified speaker: Heavily shorter stocks on a basis of capital dilution tend to fail as shorts.
- 10:38
- Unidentified speaker: It's why I don't why I never do them and never suggest them because they don't really work.
- 10:43
- Unidentified speaker: My experience, they don't really work.
- 10:45
- Unidentified speaker: Um, and so if you look at like this graph I put up here, it's a little bit hard to say, see, but I've circled it.
- 10:52
- Unidentified speaker: The bottom chart showing the volume trade in Tesla.
- 10:55
- Unidentified speaker: So when that's gone up, basically means they've issued more shares.
- 10:58
- Unidentified speaker: You can see where the short interest was very high, you know, and where it's sort of come down is because the shorts have come into this equity raise and they've reduced their shorts.
- 11:08
- Unidentified speaker: But whenever it goes up, when the short interest ratio is high, you know, that's when stock spikes, they sell some shares, and the capital problems go away.
- 11:16
- Unidentified speaker: Um now that's something we've seen a few times.
- 11:19
- Unidentified speaker: Now look at Oracle.
- 11:21
- Unidentified speaker: Uh now, Oracle, the short interest is elevated.
- 11:25
- Unidentified speaker: Okay, it's elevated.
- 11:26
- Unidentified speaker: So the market is obviously saying they're gonna have to issue more shares.
- 11:29
- Unidentified speaker: To be fair, though, they have issued more shares, uh, but you know, they're obviously thinking they're gonna be even more insurance going forward.
- 11:35
- Unidentified speaker: What they said above is that may not be the case.
- 11:38
- Unidentified speaker: So, you know, a tricky, very tricky short.
- 11:41
- Unidentified speaker: I would think showing Oracle here is a is a disaster waiting to happen.
- 11:46
- Unidentified speaker: I could be wrong.
- 11:48
- Unidentified speaker: Um now, Tesla, you may think, is a wonderful currency.
- 11:51
- Unidentified speaker: You see all the time in markets, bearishness creates upward pressure on prices.
- 11:56
- Unidentified speaker: You see it all the time.
- 11:58
- Unidentified speaker: Uh, another good one, a more relevant one maybe for for for people, is strategy, also used to be known as micro strategy.
- 12:05
- Unidentified speaker: So the market used to be super bearish on this because of its Bitcoin strategy type thing.
- 12:10
- Unidentified speaker: You spy a lot of Bitcoin, we've seen it as a way of shooting Bitcoin if you're bearish on that.
- 12:14
- Unidentified speaker: So short interest as a percentage of outstanding used to balloon at various times uh in this company's history.
- 12:20
- Unidentified speaker: And what that meant was when you had a very large short interest, was the shares actually traded a premium to NAV.
- 12:26
- Unidentified speaker: And we saw that across all the Bitcoin treasury companies, and so they were incentivized to sell shares at a premium to then buy more Bitcoin.
- 12:34
- Unidentified speaker: So what would happen was yes, maybe the premium of the shares would reduce, but the underlying asset was going up because the bearishness was feeding the liquidity of the crypto space.
- 12:45
- Unidentified speaker: I think it's no uh I think it's no surprise that crypto has struggled as the bearishness of uh the bearishness in strategy and other sort of crypto treasury companies has disappeared.
- 12:57
- Unidentified speaker: You know, as people have not gotten bearish on them, then they've been struggled.
- 13:01
- Unidentified speaker: So I'm showing bearishness is a real problem for the bears here.
- 13:04
- Unidentified speaker: They've won the narrative war, but they could well lose the capital war.
- 13:07
- Unidentified speaker: And that is the problem.
- 13:09
- Unidentified speaker: And I think the way I've way I understood it, which made it really much easier to understand, is that bearish bearish bits make the conditions that make it easier to raise capital.
- 13:21
- Unidentified speaker: Why?
- 13:22
- Unidentified speaker: Because every short seller is a guaranteed buyer at some point in the future, and every long owner is actually constrained because they can't sell because they've lent the share out to a short seller.
- 13:33
- Unidentified speaker: So basically there's no the there's no downward liquidity, there's no selling liquidity in the market.
- 13:38
- Unidentified speaker: It can only be upward liquidity.
- 13:39
- Unidentified speaker: And then your problem is if the spike then creates sort of capital raising conditions that this company is looking for, your whole short thesis falls apart.
- 13:49
- Unidentified speaker: Uh and this is what I this is what I see.
- 13:51
- Unidentified speaker: I see the bears are winning the narrative story.
- 13:54
- Unidentified speaker: They're very excited about markets, but I think the problem is they're creating conditions to lose the capital story, which is what we really care about.
- 14:02
- Unidentified speaker: Anyway, I hope that makes sense.
- 14:04
- Unidentified speaker: Uh, it makes sense to me.
- 14:05
- Unidentified speaker: Um, take care, we'll talk again soon.
- 14:07
- Unidentified speaker: Ciao.
More from Capital Flows and Asset Markets
MORE THOUGHTS ON GLD/TLT AND GLD/SPX
Episode discussion: Russell analyzes why traditional gold-based recession signals failed in 2026, citing US fiscal deficits and share buybacks.
Speakers’ claims: Russell argues that the traditional model of gold outperforming the S&P 500 as a recession warning failed in 2026, despite gold's surge in 2025. He attributes the S&P 500's resilience to a strong correlation between US fiscal deficits and corporate share buybacks since the Global Financial Crisis. The speaker notes that defensive stocks and high-yield bonds hit highs despite rising yields, contradicting standard valuation models. Additionally, Russell highlights how Chinese fiscal impulses drive their equity markets and gold prices, while US energy exports provide a fiscal buffer.
IS S&P 500 THE ONLY INFLATION HEDGE?
Episode discussion: S&P 500 defies inflation logic via buybacks and energy independence
Speakers’ claims: Russell argues that the S&P 500 has become a 'default risk-free asset' that defies traditional valuation metrics, similar to how Japanese Government Bonds (JGBs) acted as a 'widowmaker' short in the 1990s and 2000s. He attributes this resilience to massive share buybacks, favorable tax policies, and US energy independence, which keep capital domestic. Despite rising yields and inflation, the index has maintained high valuations, contradicting historical parallels to the 1970s. Russell advises against consensus non-S&P 500 trades, specifically labeling gold miners as a 'loser trade,' and recommends maintaining long positions in high-quality buyback companies while shorting cyclical assets.