Speakers analyze AI sector bankruptcy risks using CDS data and reflexive financing models
Summary
Speaker 1 introduces the topic of identifying the first AI company to go bankrupt, citing recent market volatility and concerns over circular financing. The speaker explains how credit default swaps (CDS) measure risk, calculating a 60% implied default probability for CoreWeave over five years. Speaker 1 argues NeoClouds are more vulnerable than hyperscalers due to negative operating cash flows, while Speaker 3 discusses NVIDIA's 'reflexive' financing model tied to customer performance. Finally, the speaker cites Polymarket data showing a 27% chance of an AI bubble bursting before June 2027 based on specific adverse events.
Episode sections
- 0:15 – 3:31
Introduction and Market Context
Speaker 1 introduces the topic of identifying the first AI company to go bankrupt, citing recent market volatility and concerns over circular financing. They reference a recent 'bloodbath' in AI stocks and mention political impacts on data center construction.
- 3:31 – 6:20
Credit Default Swaps and Default Probabilities
The speaker explains how to use credit default swaps (CDS) to measure risk, citing specific costs for Microsoft and CoreWeave. They calculate an implied 60% probability of default for CoreWeave over five years and note rising risk across major AI firms.
- 6:20 – 8:44
NeoClouds vs. Hyperscalers
Speaker 1 argues NeoClouds are more vulnerable than hyperscalers due to negative operating cash flows and reliance on debt/equity financing. They identify CoreWeave as the most likely first bankruptcy candidate among NeoClouds, while Nebius is seen as more resilient.
- 8:44 – 11:16
NVIDIA's Reflexive Financing Model
Speaker 3 discusses Hunterbrook's analysis of NVIDIA's 'orbital' or 'reflexive' financing, where NVIDIA's value is tied to its investments in and relationships with customers like OpenAI and CoreWeave. They warn that a drop in compute demand could trigger a synchronized collapse.
- 11:16 – 13:37
Prediction Markets and Bubble Criteria
The speaker cites Polymarket data showing a 27% chance of an AI bubble bursting before June 2027. They detail the specific criteria required for this outcome, including significant drops in NVIDIA and SOX prices, AI bankruptcies, or GPU price collapses.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| NVIDIANVDA | equity | mixed | Nvidia Nvidia's is now being called the bank. |
| MicrosoftMSFT | equity | mixed | You have Microsoft pulling back on some of their data center investments because of delays. |
| S&P 500 | index | neutral | Simply holding the S P 500 gives you anywhere from 35 to 45% exposure to this tech sector and large tech companies, which aren't all classified under the Gix technology sector. |
| AlphabetGOOGL | equity | neutral | Hunterbrook has an even more complicated complex and comprehensive chart here that shows NVIDIA with relationships to the hyperscalers, Microsoft, Google, Amazon Web Services, to the LLM providers, Meta, |
| Meta PlatformsMETA | equity | neutral | Hunterbrook has an even more complicated complex and comprehensive chart here that shows NVIDIA with relationships to the hyperscalers, Microsoft, Google, Amazon Web Services, to the LLM providers, Meta, |
| TeslaTSLA | equity | neutral | to the likes of Tesla and SpaceX, to the likes of AI startups, NeoClouds, Nebbius, Core Weav. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
Nebius is identified as the first AI stock to go bankrupt, though the speaker immediately retracts this as 'rage bait' or a joke.
Unidentified speaker (speaker_1) ·0:15
How to verify: Check transcript for retraction; verify Nebius's actual financial status via private company filings or news reports. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
Primary sourceSidney Sweeney was filmed crying in character during the filming of 'The Housemaid's Secret' in New York last week.
Unidentified speaker (speaker_1) ·1:15
How to verify: Search news archives for reports on Sidney Sweeney and 'The Housemaid's Secret' production in the week prior to 2026-11-14. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
The four AI stocks hit hardest during the recent bloodbath were Sandisk, AMD, Micron, and Monebius.
Unidentified speaker (speaker_1) ·2:14
How to verify: Check stock price performance data for SDI (or SanDisk equivalent), AMD, MU, and MBI (or Monebius equivalent) for the week prior to 2026-11-14. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
Sandisk soared over 2,000% in less than two years.
Unidentified speaker (speaker_1) ·2:34
How to verify: Calculate percentage price change for Sandisk (or its successor entity) from 2026-11-14 to 2026-11-14. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
It costs $5.60 per year to insure Microsoft against default over the next five years via credit default swaps.
Unidentified speaker (speaker_1) ·3:51
How to verify: Check CDS market data for Microsoft (MSFT) 5-year maturity contracts around 2026-11-14. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
Primary sourceIt costs $110 per year to protect against default for $1,000 in CoreWeave debt.
Unidentified speaker (speaker_1) ·4:31
How to verify: Check CDS market data for CoreWeave debt instruments around 2026-11-14. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
The implied probability of default for CoreWeave is about 60% over the next five years.
Unidentified speaker (speaker_1) ·5:11
How to verify: Calculate implied default probability using CoreWeave CDS spreads and recovery rate assumptions. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
NeoClouds have lost half their value since an earlier piece, except for Nebius.
Unidentified speaker (speaker_1) ·7:40
How to verify: Compare NeoCloud stock indices or individual company valuations from the date of the 'earlier piece' to 2026-11-14. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
NVIDIA has an investment in OpenAI and CoreWeave, and OpenAI has an investment in CoreWeave.
Unidentified speaker (speaker_3) ·10:04
How to verify: Check SEC filings (Form D, 13F, or press releases) for NVIDIA, OpenAI, and CoreWeave investment disclosures. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
Primary sourcePolymarket says there is a 27% chance that the AI bubble bursts before June of 2027.
Unidentified speaker (speaker_1) ·11:20
How to verify: Visit Polymarket and check the current odds for the 'AI bubble burst' market resolution date. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.
Do your own research
- Verify CoreWeave's current CDS spread and compare it to the $110/year figure cited for a $1,000 notional amount using Bloomberg or Refinitiv data.
- Review CoreWeave's latest 10-K or 10-Q filings for operating cash flow statements to confirm negative cash flow trends.
- Analyze NVIDIA's investment portfolio disclosures in SEC filings to assess exposure to AI startups like OpenAI and CoreWeave.
- Check Polymarket's current odds for the 'AI bubble burst before June 2027' market to verify the 27% probability claim.
- Research recent political legislation impacting data center construction permits to assess the 'political impacts' mentioned.
- Compare Microsoft's CDS pricing against the $5.60/year figure to validate the baseline risk assessment.
- Investigate the specific criteria for the Polymarket bubble burst event, including drops in NVIDIA/SOX prices and GPU price collapses.
- Review Hunterbrook's original analysis on NVIDIA's 'orbital' financing model to verify the causal link between customer demand and NVIDIA's valuation.
Full transcriptexpand
- 0:00
- Unidentified speaker (UNKNOWN): Welcome
- 0:01
- Unidentified speaker (speaker_0): to the Nanalize Podcast. We are a boutique media and research firm specializing in disruptive innovation. Visit Nanalize.com for more details.
- 0:15
- Unidentified speaker (speaker_1): It's Nebius, okay? NeoCloud Nebbius will be the first AI stock to go bankrupt. Nah nah, it's rage bait, I'm joking. Or am I? From anthropic to core wheel. weave to Oracle to Manebius, we're gonna look at the first AI stock that's likely to go bankrupt. Why is that so important?
- 0:35
- Unidentified speaker (speaker_1): Because bankruptcies are the canary in the coal mine for any sort of financial crisis. Most of you young bucks out there have no idea what real blood on the street looks like, but don't you worry. I'll talk about last week's bloodbath in a second. but I have a very important point to make here. We believe that it's
- 0:55
- Unidentified speaker (speaker_1): highly likely in the case of the AI bubble bursting that it's accompanied by large bankruptcies or even driven by large bankruptcies. Why? Because of something called circular financing, and it's getting worse and worse as the days go on. Nvidia Nvidia's is now being called the bank.
- 1:15
- Unidentified speaker (speaker_1): of AI for good reason. Wyatt's gonna discuss that a little bit later, but let's get on to talking about the elephant in the room. Something terrible happened last week, and I can't stop thinking about it. Sidney Sweeney was filmed crying in character during the filming of The Housemaid's Secret last week in New York. I
- 1:35
- Unidentified speaker (speaker_1): hope that poor woman's okay. Another equally traumatic event. that happened last week was the AI stock bloodbath. Bloodbath. As always, we kept our Nanalize premium subscribers up to date on the situation in our Nanalize market open. But for the rest of you, take a look at this
- 1:54
- Unidentified speaker (speaker_1): chart here. This is the NASDAQ 100 over the past five years, and it's not really that bad when you look at it in the context of a longer time frame, is it? Now, yes, I know the crisis mainly affected high beta AI stocks. So take a look at this even better slide. These are the four
- 2:14
- Unidentified speaker (speaker_1): AI stocks that were hit the hardest during last week's bloodbath Sandisk, AMD, Micron, and Monebius. You know, this always gets me nobody bats an eyelid when Sandisk soars over 2,000% in less than two years. But heaven forbid it should drop
- 2:34
- Unidentified speaker (speaker_1): 20%, giving it a year-to-date return of only 200%. We need to stop slobbering over short-term price changes like a bunch of monkeys and focus on what systemic changes may have happened here. Most recently, there are concerns.
- 2:51
- Unidentified speaker (speaker_1): that the democratic midterm wins in the United States are going to slow down data center construction because of things like regulation, moratoriums, and the like. And probably most tangible would be the signs that are showing access to easy capital for AI expansion is starting to tighten. Evidence, well,
- 3:11
- Unidentified speaker (speaker_1): some AI IPOs are being pulled back due to lack of interest from investors like Firmus in Australia. You have Microsoft pulling back on some of their data center investments because of delays. And most importantly, I think is that new debt is becoming more costly because it's more costly to insure it. This is
- 3:31
- Unidentified speaker (speaker_1): something we measure with tools called credit default swaps. This is a really cool metric that any investor out there can. use to monitor the situation. So here we've listed out some of the hyperscalers, and I'll just show you really quickly how to read this. So let's take Microsoft as an example here. It costs
- 3:51
- Unidentified speaker (speaker_1): $5.60 per year to insure Microsoft against default over the next five years. And for most large AI companies out there, you can find these instruments. What you can then do is see what the cost of these credit default swaps has been doing over time.
- 4:11
- Unidentified speaker (speaker_1): For example, how far are they off their all-time high? Right? If they're close to the all-time high, shows you that risk is still pretty high or perceived as being high. And we can expand this out to other names. And of course, the name that's been in the news. lately, Core Weave. So it costs $110 a year
- 4:31
- Unidentified speaker (speaker_1): to protect against default for $1,000 in core weave debt. That's very high. And when we look at the change of these credit default swaps over the past month, you can see they've all been rapidly appreciating. In other words, this debt is getting riskier as time goes on. Nvidia, look at that appreciation in the risk for NVIDIA. NVIDIA's
- 4:51
- Unidentified speaker (speaker_1): debt. What's going on there? Well, Wyatt's going to talk a bit more about that in a second. But the other thing that we can do with the credit default swap that's quite interesting is that you can calculate an implied probability of default. In the case of Corweave, that's about a 60% likelihood they default over the next five years. That's crazy,
- 5:11
- Unidentified speaker (speaker_1): right? And when you consider the likelihood of default. there can be problems that are not so obvious to investors. It's not just a matter of looking at their debt and then their cash flows to see if they can make their interest payments. You also need to consider things like contractual commitments, covenants for their debt that may be quite
- 5:30
- Unidentified speaker (speaker_1): obscure, off-balance sheet debt that you can't even see. And it's the job of credit ratings agencies to look through. all this stuff. For a good number of AI stocks out there, we can look to credit ratings agencies for signs of trouble. We can see how they change their outlook over time. Now I know what
- 5:50
- Unidentified speaker (speaker_1): a lot of you are thinking right now. Too many Sydney Sweeney jokes lately. How about not enough Sydney Sweeney jokes?
- 5:58
- Unidentified speaker (speaker_2): How about that? How about that? How about that? Now
- 6:00
- Unidentified speaker (speaker_1): I know what a lot of you are thinking. Credit ratings agencies weren't of any use during the 2008 financial crisis. And you're exactly right. A sufficiently sized cataclysmic crisis doesn't give investors time to respond. That's the entire point of it. Because of the circular nature of the AI
- 6:20
- Unidentified speaker (speaker_1): economy, a crash is likely to be sudden and powerful. Neo clouds are going to be the first. to fail because they rely on debt for their lofty valuations. You see that happening right now. This simply captures current sentiment because if it becomes more difficult for them to raise money, those growth stories start to dry
- 6:40
- Unidentified speaker (speaker_1): up. It comes down to this: the core business for these neo clouds is not generating positive operating cash flows or earnings sufficient to service their debt. The debt servicing is simply coming from customer advances or new financing or even selling equity. Whenever we bring this up, that neo
- 7:00
- Unidentified speaker (speaker_1): clouds are likely to be the first to fail, nobody disputes that. They simply focus on reasons why the AI trade is going to keep going. This is the compute demand story, fair enough. But for neo clouds in particular, when they can can't access capital, that growth story dries up. Hyperscalers, on the other hand,
- 7:20
- Unidentified speaker (speaker_1): they have very strong cash flows that they can fund build-outs with. They don't need to take on debt, even though some are. And hyperscalers are actually major customers for some neo clouds, and that's a real problem, right? You never want to have your competitor as your customer. We talked about this in an earlier. piece this year,
- 7:40
- Unidentified speaker (speaker_1): which looked at problems in the neo cloud sector. Now, since that piece, neo clouds have lost half their value, except for Nebius, which has a stronger balance sheet and is more resilient to turmoil. The first AI company to go bankrupt is likely to be a neo cloud, and it's likely to be Core Weave because
- 8:00
- Unidentified speaker (speaker_1): they're the most likely to struggle when it comes. to refinancing their debt or paying their interest obligations. Of the NeoClouds, Nebius is probably the best positioned at the moment because of their strong balance sheet. So ask yourself this: why did this recent bloodbath have you so worried? Is it because you're overexposed to
- 8:20
- Unidentified speaker (speaker_1): the AI trade? Well, yes, because everyone is. Simply holding the S P 500 gives you anywhere from 35 to 45% exposure to this tech sector and large tech companies, which aren't all classified under the Gix technology sector. So let's talk a little bit more about NVIDIA,
- 8:40
- Unidentified speaker (speaker_1): which is presently our largest position that we've been holding for over a decade now.
- 8:44
- Unidentified speaker (speaker_3): Putting aside Jensen Wong's recent scandal, today I want to look at some concerns raised by Hunterbrook, which is an investigative journalism company. They came up with this really cool piece called The Comprehensive Guide to the Bank of NVIDIA, which made some really cool points. Essentially, they're saying that NVIDIA has evolved from just a chipmaking company to a company that supports the entire AI industry. There have been
- 9:04
- Unidentified speaker (speaker_3): these concerns going around about circular financing, right? You've seen the giant bubble chart where NVIDIA supplies a customer, they have an investment in that customer. Hunterbrook has an even more complicated complex and comprehensive chart here that shows NVIDIA with relationships to the hyperscalers, Microsoft, Google, Amazon Web Services, to the LLM providers, Meta,
- 9:24
- Unidentified speaker (speaker_3): Anthropic, OpenAI, to the likes of Tesla and SpaceX, to the likes of AI startups, NeoClouds, Nebbius, Core Weav. All of these are interconnected. And they're saying that it's not circular financing, it's reflexive. So the reflexive property in math, going back to our high school math days, means that one thing equals itself, meaning
- 9:44
- Unidentified speaker (speaker_3): that the value of NVIDIA is dependent on the value of its customers, of its investments, of its suppliers. So lately, all of the value of these investments have been growing together. However, when the opposite happens, when demand for AI compute begins to underwhelm supply of AI compute, right? Because we're in a massive. build-out right now, the
- 10:04
- Unidentified speaker (speaker_3): opposite is going to happen. And you're going to see the reflexive financing in action. Let me give you an example. Nvidia has an investment in OpenAI. OpenAI is a customer of NVIDIA. NVIDIA has an investment in Core Weave. Core Weave is a customer of NVIDIA. OpenAI has an investment in Core Weave. There's this loop. And Hunter Brook Brooke calls it orbital financing
- 10:24
- Unidentified speaker (speaker_3): as opposed to circular financing, which implies some motion. throughout this orbit. As soon as one component of this orbit, of this reflexive financing loop, starts to struggle, it impacts the rest of the group.
- 10:36
- Unidentified speaker (speaker_1): So far, we've only talked about public companies going bankrupt, but honestly, a overly hyped private company going bankrupt can be a catalyst that pops the AI bubble as well. So I think right behind neo NeoClouds clouds being the most likely to go bankrupt first, you probably have startups that have overextended
- 10:56
- Unidentified speaker (speaker_1): themselves. And a good reason for that is simply just look at how the venture capital community is just throwing money at AI hand over fist. No business model, no customers, no website, no problem. Here's a billion dollars. Now, one other thing I recently saw that's quite interesting is. polymarket,
- 11:16
- Unidentified speaker (speaker_1): so that's a prediction market, says there's a 27%
- 11:20
- Unidentified speaker (speaker_1): chance that the AI bubble bursts before June of 2027. Well, immediately you ask, what does that mean? Well, it's quite interesting. So they say that three at least of these events need to happen in a 90-day window for that bet to resolve to AI bubbles. bubble has
- 11:39
- Unidentified speaker (speaker_1): burst. They are NVIDIA down 50% from an all-time high, SOX, the semiconductor ETF, down 40% from an all-time high, OpenAI or Anthropic declaring bankruptcy, OpenAI being acquired, H100 GPU rental prices falling
- 11:59
- Unidentified speaker (speaker_1): to a dollar or lower for five consecutive days. or any major AI hardware supplier seeing their price, their stock price collapse 50% or more. And you can see the list of names here. Again, three or more of these events need to happen during a 90-day window. So, by all measures, the
- 12:19
- Unidentified speaker (speaker_1): AI bubble is not bursting at all. This is just normal price volatility. One or more bankruptcies is major bankruptcies is likely to accompany any real AI crisis. And the first company to go bankrupt, it's likely to be core weave based on what everything points to today. Just remember, no
- 12:39
- Unidentified speaker (speaker_1): matter how strong demand is for AI compute, there are always unknown unknowns that can happen, right? Think like a COVID COVID-style style moratorium on AI because that whole cybersecurity situation gets out of hand. That's why you should always be aware of how much tech risk you have. And for those of you thinking about doubling down
- 12:59
- Unidentified speaker (speaker_1): on neo clouds, you ought to give this next piece which talks about some of the risks in that space. Thanks so much for taking the time.
- 13:07
- Unidentified speaker (UNKNOWN): Thank
- 13:07
- Unidentified speaker (speaker_0): you for listening to the Nanalize Analyze Podcast. If you found this information useful, please share this episode with a friend. This helps us to continue to provide thorough research for you. Want more research like this? Want to know what we're invested in and what stocks we're avoiding? Head to Nanalize.com and consider becoming a premium annual subscriber to get access to
- 13:27
- Unidentified speaker (speaker_0): premium articles, webinars, and our extensive tech stock catalog. Thank you for your time.
More from Nanalyze
How To Invest In The Future Of Nuclear: Four Stocks
Episode discussion: Nanalyze identifies four nuclear stocks via ETF overlap analysis, excluding pre-revenue SMRs.
Speakers’ claims: Speaker_1 outlines a methodology of analyzing seven nuclear ETFs to identify common top holdings, finding BWXT as the only stock in the top 15 of all seven. The speaker critiques nuclear SPACs and pre-revenue Small Modular Reactor (SMR) companies like ACLA and TerraNova, citing past predictions of their decline and a policy against investing in companies with no revenues or market caps below $1 billion. Utility companies are evaluated for nuclear exposure, with Constellation Energy retained for its ~90% nuclear operations while others are dropped. The speaker proposes a 'mini nuclear ETF' consisting of BWXT, Cameco, Constellation Energy, and Centrus Energy to provide diversified exposure while avoiding ETF fees.
It's Boring, But It Pays 15% Guaranteed for Life
Episode discussion: Speakers argue 15% guaranteed annuities are inferior to 60-40 portfolios due to inflation, opportunity cost, and lack of liquidity.
Speakers’ claims: Speaker 1 introduces a hypothetical New York Life annuity offering a 15% nominal return but argues it is misleading due to inflation and opportunity cost. Speaker 2 compares this to a 60-40 stock-bond portfolio, demonstrating that the portfolio preserves principal and offers higher cumulative value even in poor market scenarios. Speakers warn that annuity issuers profit from mortality risk and high commissions, advising against allocating more than 10% of assets to a single issuer. The episode concludes that asset allocation is a superior solution to sequence of returns risk compared to annuities.