CoreWeave bonds yield 15% amid market fear of oversupply, but speakers argue fundamentals support investment through 2027
Summary
Host Josef Pschorn and analyst Jordan Chalfin discuss CoreWeave's credit profile, noting that CDS spreads widened from ~450 bps to >1200 bps despite Q3 revenue beats. Chalfin maintains a buy recommendation on CoreWeave bonds, citing 15% yields and strong near-term fundamentals driven by take-or-pay contracts. The speakers attribute the credit spread widening to technicals, sentiment, and fears of a self-fulfilling prophecy regarding CapEx funding, rather than fundamental deterioration. They highlight that while medium-term risks like capacity oversupply exist for 2028-2030, CoreWeave's unit economics and secured debt structure provide protection through 2027.
Key takeaways
- 1 98% of their business is take or pay, right? They're essentially doing nothing or very little on demand.
Unidentified speaker (speaker_0) · 12:03
- 2 Secured debt is self-amortizing at the project level, paid down by customer monthly bills, rather than sitting at the parent level.
Unidentified speaker (speaker_0) · 42:51
Episode sections
- 0:00 – 0:31
Cold open: Bond yields and self-fulfilling prophecy
Speaker 0 argues that double-digit bond yields create a self-fulfilling prophecy where investors fear CoreWeave's ability to fund CapEx, despite customers being desperate for capacity and willing to prepay for older chips. This dynamic could lead to excess capacity issues, but Speaker 0 believes this risk is more relevant to 2028-2030 than 2027.
- 0:31 – 2:04
Episode intro and market context
Host Josef Pschorn notes the recording date is November 13, 2026. He highlights that CoreWeave CDS spreads widened from ~450 bps in June to >1200 bps recently. CoreWeave beat Q3 revenue and margin guidance, causing a 20% stock rally, though the stock has since consolidated.
- 2:04 – 3:36
Buy recommendation rationale
Jordan Chalfin reiterates a buy recommendation on CoreWeave bonds driven by 15% yields and strong fundamentals expected to hold through 2027. He cites Q3 revenue and margin beats, strong new deal pricing, and the announcement of a new investment-grade facility as key factors.
- 3:36 – 5:48
Market sentiment vs. fundamentals
Chalfin attributes the market's negative sentiment to technicals, excessive tech supply, and headline noise (e.g., OpenAI IPO delay) rather than fundamentals. He describes a downward spiral where high yields trigger selling due to funding concerns, despite improving pricing and margins.
- 5:48 – 7:15
CoreWeave business model and key drivers
CoreWeave is a pure-play neocloud that leases data center power shells and buys expensive GPU servers. It signs 4-5 year take-or-pay compute rental agreements with hyperscalers, AI labs, and startups. Key credit drivers are contract pricing, GPU costs, and financing terms.
- 7:15 – 9:35
Active vs. contracted power and profitability
CoreWeave has 1.7 GW of active power (running) and 4.6 GW of contracted power (signed leases not yet online). The gap between ramp-up costs and revenue collection causes negative profits initially, but margins are rebounding toward a 17% target for Q4 after targeting mid-20% steady-state margins.
- 9:35 – 11:34
Data center delays and regulatory pushback
Neoclouds face execution and regulatory challenges, including state/local pushback and moratoriums. However, most 2027 capacity is under construction and permitted. Data centers provide tax revenue and jobs, which may mitigate political risks.
- 11:34 – 31:56
Take-or-pay contract mechanics
98% of CoreWeave's business is take-or-pay, meaning customers pay regardless of usage unless there is a material breach or delay by CoreWeave. In the current tight market, customers rarely walk away, and CoreWeave can re-contract at higher prices if a customer defaults.
- 31:56 – 33:46
Customer Prepayments and Ranking
Speakers discuss how customer prepayments function as unsecured claims and the uncertainty regarding whether they rank ahead of bondholders in a bankruptcy scenario, depending on whether payments are made to the parent or the underlying subsidiary.
- 33:46 – 36:33
Unit Economics and Contract Margins
Jordan Chalfin outlines CoreWeave's unit economics, citing a 24% operating margin and 16% pre-tax margin after interest over a five-year contract, noting that recent strong pricing offsets potentially rosy assumptions.
- 36:33 – 38:27
Lease Renewal and Capacity Risks
The speakers identify a medium-term risk where data center leases outlast compute contracts, potentially leading to overcapacity and lower renewal rates from hyperscalers who own their own infrastructure.
- 38:27 – 39:22
Financing Constraints and Cost of Capital
Chalfin explains that CoreWeave avoids unsecured debt due to high yields (15%), preferring secured debt, customer prepayments, and OEM financing, with a target cost of debt financing at or below 10%.
- 39:22 – 42:05
Profitability and Cash Flow Timing
CoreWeave continues to report net losses due to ramp-up costs and a timing mismatch between upfront capex and ratable revenue, with a break-even timeline estimated at two to four years per asset.
- 42:05 – 1:02:04
Debt Structure and Secured Lending
Chalfin notes that future funding will be primarily secured debt layered behind existing loans, which self-amortize against customer payments at the project level, rather than unsecured parent-level debt.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| AmazonAMZN | equity | neutral | They've been in this business a long time. Um, clearly, Amazon and Alphabet |
| MicrosoftMSFT | equity | neutral | Microsoft are going to be the most sophisticated in the market. |
| NVIDIANVDA | equity | neutral | billion dollars on bleeding edge Nvidia chips um for for one gigawatt of capacity. So it's it's way more expensive than the than the actual data center, despite being a you know a short lived asset. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
CoreWeave CDS spreads were ~450 bps in June and rose to >1200 bps by November 2026.
Unidentified speaker (speaker_1) ·0:31
How to verify: Check CDS pricing data for CoreWeave from Bloomberg or Refinitiv for June 2026 and November 2026. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave stock rallied ~20% after Q3 results but has since consolidated.
Unidentified speaker (speaker_1) ·0:50
How to verify: Check CoreWeave stock price history on Yahoo Finance or Bloomberg for the period following the Q3 earnings release in November 2026. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave CDS peaked at ~880 bps in December 2025 while the stock was above $100.
Unidentified speaker (speaker_1) ·4:21
How to verify: Check historical CDS data for CoreWeave and stock price history for December 2026. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave targets mid-20% operating margins on contracts, with Q4 margins expected in the 17% range.
Unidentified speaker (speaker_0) ·8:39
How to verify: Check CoreWeave's Q3 2026 earnings call transcript for margin guidance and targets. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
98% of CoreWeave's business is take-or-pay contracts.
Unidentified speaker (speaker_0) ·12:03
How to verify: Verify against CoreWeave's Q3 2026 earnings release or investor presentation detailing contract types. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
One gigawatt of data center capacity costs $15-20 billion for the power shell and $35-40 billion for Nvidia hardware (Vera Rubin).
Unidentified speaker (UNKNOWN) ·23:19
How to verify: Check CoreWeave's Q3 2026 earnings release for CapEx breakdown per gigawatt and Nvidia's Vera Rubin pricing announcements. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceCoreWeave expects $9 billion in prepayments and deposits in the next three months.
Unidentified speaker (speaker_1) ·31:22
How to verify: Verify against CoreWeave's Q3 2026 earnings release or investor presentation for forward-looking cash flow guidance. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
The Meta facility is non-recourse and valued at $8.5 billion.
Unidentified speaker (UNKNOWN) ·28:58
How to verify: Check CoreWeave's Q3 2026 earnings release for details on the Meta facility structure and size. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
US data center capacity was ~45 GW at the end of 2025 and is expected to triple over the next several years.
Unidentified speaker (speaker_0) ·22:07
How to verify: Check industry reports from Uptime Institute or similar data centers capacity trackers for 2025 and 2026 forecasts. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave's unit economics show a 24% operating margin and 16% pre-tax margin over a five-year contract.
Unidentified speaker (speaker_0) ·34:16
How to verify: CoreWeave Q3 2026 10-Q filing; check 'Risk Factors' or 'Management Discussion' for contract economics assumptions. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceNew four-to-five-year contracts are priced at $20-25 million per megawatt.
Unidentified speaker (speaker_0) ·35:15
How to verify: CoreWeave Q3 2026 Earnings Call transcript; search for 'per megawatt' or 'pricing'. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave has approximately $133 billion in Remaining Performance Obligations (RPO).
Unidentified speaker (UNKNOWN) ·49:33
How to verify: CoreWeave Q3 2026 10-Q filing; check 'Contract Obligations' or 'RPO' section. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceCoreWeave has approximately $110 billion in undiscounted future lease payments as of Q3.
Unidentified speaker (speaker_0) ·48:07
How to verify: CoreWeave Q3 2026 10-Q filing; check 'Lease Obligations' table. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceNvidia provided a backstop for CoreWeake capacity worth $6-7 billion.
Unidentified speaker (UNKNOWN) ·50:31
How to verify: CoreWeave Q3 2026 10-Q filing; check 'Related Party Transactions' or 'Commitments'. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceCoreWeave bonds are rated B-minus.
Unidentified speaker (speaker_0) ·57:25
How to verify: Credit rating agency reports (S&P, Moody's, Fitch) for CoreWeave Holdings Corp. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Oracle credit spreads are trading around 300 basis points.
Unidentified speaker (speaker_0) ·58:52
How to verify: Bloomberg or Refinitiv data for Oracle (ORCL) CDS or bond spreads. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave bonds are trading at a spread of approximately 950 basis points.
Unidentified speaker (speaker_0) ·59:15
How to verify: Bloomberg or Refinitiv data for CoreWeave bond tickers. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave expects $9 billion in prepayments in Q4, with $3 billion for recontracting existing capacity.
Unidentified speaker (speaker_0) ·47:07
How to verify: CoreWeave Q3 2026 Earnings Call transcript; search for 'prepayments' or 'Q4'. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
CoreWeave is not coming to the unsecured bond market because yields are uneconomic.
Unidentified speaker (speaker_1) ·42:05
How to verify: CoreWeave Q3 2026 Earnings Call transcript; search for 'bond market' or 'unsecured'. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Do your own research
- Verify CoreWeave's Q3 2026 earnings report for revenue, margin beats, and guidance updates.
- Check CoreWeave's latest 10-K or 10-Q filings for details on take-or-pay contract terms, customer concentrations, and off-balance-sheet obligations.
- Review CoreWeave's capital structure filings to confirm the ratio of secured vs. unsecured debt and the specific covenants attached to secured loans.
- Analyze CoreWeave's contracted power (4.6 GW) vs. active power (1.7 GW) in recent investor presentations to assess execution risk and timeline for capacity ramp.
- Compare CoreWeave's CDS spreads and bond yields to peers like Oracle and other neoclouds to contextualize the 'distressed' pricing.
- Investigate the legal ranking of customer prepayments in bankruptcy scenarios, specifically whether they rank ahead of bondholders.
- Assess the risk of data center leases (10+ years) outlasting compute contracts (5 years) by reviewing CoreWeave's lease expiration schedules.
- Monitor hyperscaler CapEx guidance and AI demand trends to evaluate the risk of capacity oversupply in the 2028-2030 timeframe.
Full transcriptexpand
- 0:00
- Unidentified speaker (speaker_0): There's a kind of a self-fulfilling prophecy here. When the bonds reach kind of double digit yields, investors get concerned about
- 0:08
- Unidentified speaker (UNKNOWN): the
- 0:08
- Unidentified speaker (speaker_0): ability for Corleaf to fund.
- 0:10
- Unidentified speaker (UNKNOWN): It's
- 0:10
- Unidentified speaker (speaker_0): CapEx. So customers are kind of so desperate for capacity right now that they're willing to make a prepayment
- 0:17
- Unidentified speaker (UNKNOWN): on
- 0:17
- Unidentified speaker (speaker_0): an old generation of a chip.
- 0:20
- Unidentified speaker (UNKNOWN): So
- 0:21
- Unidentified speaker (speaker_0): if there's too much capacity in the industry, why would they renew with core leaves? weaves?
- 0:25
- Unidentified speaker (UNKNOWN): I
- 0:25
- Unidentified speaker (speaker_0): don't think those play out in 2027. I think that's more of like a 2028 to 2030 issue.
- 0:31
- Unidentified speaker (speaker_1): Welcome to another episode of Fix and Floating. We are recording this on Friday, the 13th of November. In June, CDS ensuring investors against the default by Core Weav cost about 450 basis points. Last week, they cost more than 1200.
- 0:49
- Unidentified speaker (UNKNOWN): On
- 0:50
- Unidentified speaker (speaker_1): Tuesday, Corviv showed a beat on revenue and margins and also raised its guidance. Next day, stock rallied by about 20%, but now the stock is barely up. Today we're gonna look at what the release means for credit investors. I'm very pleased to welcome Jordan Schalfin, head of technology
- 1:10
- Unidentified speaker (speaker_1): at Credit Sites. Jordan, welcome to the show.
- 1:13
- Unidentified speaker (UNKNOWN): Thank
- 1:14
- Unidentified speaker (speaker_0): you for having me. Excited to uh be here and talk about CoreWeath.
- 1:18
- Unidentified speaker (speaker_1): Jordan, would you mind giving our listeners a bit of background on your experience?
- 1:26
- Unidentified speaker (speaker_0): Sure. I am uh head of technology here at Credit Sites. I've been here 17 years. I cover both investment grade and high yield and a bit of distress, including Corleaf uh credits.
- 1:41
- Unidentified speaker (UNKNOWN): Uh,
- 1:41
- Unidentified speaker (speaker_0): prior to joining credit sites, I spent a few years
- 1:45
- Unidentified speaker (UNKNOWN): uh
- 1:46
- Unidentified speaker (speaker_0): as an equity research analyst at Goldman Sachs,
- 1:49
- Unidentified speaker (UNKNOWN): um,
- 1:49
- Unidentified speaker (speaker_0): as well as a couple of years on the buy side as well.
- 1:53
- Unidentified speaker (UNKNOWN): My
- 1:53
- Unidentified speaker (speaker_0): my coverage is um, I would say mostly investment grade, and that includes all of the hyperscalers as well.
- 2:03
- Unidentified speaker (UNKNOWN): So
- 2:04
- Unidentified speaker (speaker_1): Corviv is getting a lot of hate at the moment. Um, and you came out with a buy recommendation on the bonds. Um, after Tuesday, you reiterated the buy. What prompted you to do that?
- 2:19
- Unidentified speaker (speaker_0): Sure. Well, the buy recommendation is really driven by two things. It's it's one the the 15% yields on the bonds, and two, um the fact that
- 2:28
- Unidentified speaker (UNKNOWN): I
- 2:28
- Unidentified speaker (speaker_0): feel, you know, strongly that the fundamentals are going to remain very strong um through 2027.
- 2:35
- Unidentified speaker (UNKNOWN): Um there
- 2:37
- Unidentified speaker (speaker_0): are medium term risks here uh in terms of potentially having too much capacity um when the cycle turns and also a customer renewal risk as well. But I don't think those are going to play out in 2027 and our recommendations are generally um one year views. I I would say that the Q3
- 2:57
- Unidentified speaker (speaker_0): results were very supportive of my recommendation, both on the operating trends front. So uh there's a revenue beat. More importantly, there's a very strong margin beat, and the uh pricing on new deals is is really strong. Um, and that's flowing through into uh the employer. implied margin guidance for Q4
- 3:17
- Unidentified speaker (speaker_0): as well. They also address a lot of investor concerns on the call and announce that they're coming with a new uh IG rated um facility. And um I think that you know certainly addressed some fears about their ability to obtain funding.
- 3:36
- Unidentified speaker (speaker_1): What do you do you say is the model? market missing in your view at the moment
- 3:43
- Unidentified speaker (speaker_0): yeah I think and it and it kind of transcends across both investment grade and high yield but I the the the market right now is just it's very um much driven by technicals and sentiment um as opposed to fundamentals which which we talked about are are strong and actually improving. Um, There's just been so such a
- 4:03
- Unidentified speaker (speaker_0): plethora of of supply um of tech issuance in the sector and a lot of um headline noise, whether it's like OpenAI delaying an IPO. Um there's just a lot of noise out there and a lot of skittishness, I would say, uh, from investors.
- 4:20
- Unidentified speaker (UNKNOWN): In
- 4:21
- Unidentified speaker (speaker_1): December, CDS peaked at around eight 880 hundred and eighty basis points while the stock was well above a hundred bucks. By June, um CDS was down to around 450. Last week it was about 1200, the stock materially lower. What would you say changed between June and now?
- 4:41
- Unidentified speaker (UNKNOWN): Yeah,
- 4:41
- Unidentified speaker (speaker_0): so that's I think a lot of the things that we talked about, right? Um the
- 4:47
- Unidentified speaker (speaker_0): you know, all of the new supply that hit the market and in tech. Um the delayed uh IPO from from OpenAI, they had a failed acquisition of of Core Scientific.
- 4:58
- Unidentified speaker (UNKNOWN): Um
- 4:58
- Unidentified speaker (speaker_0): there's a lot of short seller criticism out there, you know, including you know Jim Chanos and uh Michael Barry. And so, you know, I would say just the technicals and the sentiment have been have been really bad. Um and it really hasn't been um it really hasn't been on the fundamental side. at all, you know, as as pricing continues to to
- 5:18
- Unidentified speaker (speaker_0): move up and be um and and be quite impressive. Um I would say the other issue too is there's a kind of like a like a downward spiral slash you know self fulfilling prophecy here. You know, in that when the when the bonds reach kind of double digit yields, yield um investors get concerned about the
- 5:37
- Unidentified speaker (speaker_0): ability for corporate core weath to fund um it's capex growth and then that contributes to to additional selling as well
- 5:48
- Unidentified speaker (speaker_1): for listeners who are less familiar with what core we've is doing could you please give us a quick wrap up of the business and especially what the key levers are for the credit.
- 6:01
- Unidentified speaker (speaker_0): Sure. Yep Absolutely. So it's actually a fairly straightforward, simple business. They're pure play neo cloud. Most of their data centers are released. They have to buy the the GPU servers, which are very expensive to put into those data centers.
- 6:19
- Unidentified speaker (UNKNOWN): And
- 6:19
- Unidentified speaker (speaker_0): then what they do is they they enter into um these off take agreements or compute rental agreements um with customers. that are either hyperscalers or AI labs or AI startups, some enterprises. And those are generally take or pay contracts that are uh four years uh on average, uh five years
- 6:39
- Unidentified speaker (speaker_0): in duration.
- 6:41
- Unidentified speaker (UNKNOWN): In
- 6:41
- Unidentified speaker (speaker_0): terms of the key drivers here, um, it's gonna be the pricing on the contracts as well as the the cost uh of those GPUs and the financing terms uh for those GPUs as well.
- 6:55
- Unidentified speaker (speaker_1): I think that uh contracted power went up by more than 900 wega megawatts in the quarter to over four point size six gigawatts now. Um about 1.7 gigawatts are already running today. What is the difference between active power and contracted power?
- 7:15
- Unidentified speaker (speaker_0): Sure. So active power is essentially
- 7:18
- Unidentified speaker (UNKNOWN): um
- 7:18
- Unidentified speaker (speaker_0): what's up and running today. Right. And that's that's as you mentioned, that's 1.7 gigawatts.
- 7:24
- Unidentified speaker (UNKNOWN): Um
- 7:25
- Unidentified speaker (speaker_0): nearly all of that is going to be chipped and and leased out to customers. You know, there's a a small lag there. You know, when when you get the active power, it takes a bit of time to put the GPU servers in and and rent them out. But nearly all of that is kind of up and running capacity. Um, the contracted power 4.6 gigawatts, this includes all of the
- 7:45
- Unidentified speaker (speaker_0): leases that they have for those power shells that have not yet commenced.
- 7:51
- Unidentified speaker (UNKNOWN): So
- 7:52
- Unidentified speaker (speaker_0): it includes both the active power as well as those um leases not yet commenced. So there's essentially about uh the difference is is three gigawatts that that are signed. and and have not yet come online.
- 8:06
- Unidentified speaker (speaker_0): And the
- 8:08
- Unidentified speaker (speaker_0): amount that they've actually sold to customers or pre-sold to customers that shows up in RPO is going to fall somewhere between those two figures.
- 8:17
- Unidentified speaker (speaker_1): Is that also um one of the reasons why we're seeing these very negative profits?
- 8:25
- Unidentified speaker (UNKNOWN): Um
- 8:25
- Unidentified speaker (speaker_1): as there's this mismatch between um you already have to start up the um the the facilities while you're not uh yet um collecting revenue.
- 8:39
- Unidentified speaker (speaker_0): Absolutely. There's there's ramp up costs here. Um they kind of target like mid twenty percent And operating margin um on a on a contract, but it takes a couple months to kind of get to steady state uh production and and go through like normal ramp up type costs. Um so as the business is
- 8:59
- Unidentified speaker (UNKNOWN): ha
- 8:59
- Unidentified speaker (speaker_0): has done a decent job scaling, but it's still like not yet at scale. And those um that increase in active power has has has grown pretty dramatically. Um, you know, that that's gonna going to have an out outsized impact on margins. And we really saw that play out, I would say, earlier this year, where margins were
- 9:19
- Unidentified speaker (speaker_0): quite depressed. Um, but they've rebounded quite nicely sequentially throughout the year. And they're going to be, you know, decently strong, uh, probably in that like 17%
- 9:30
- Unidentified speaker (speaker_0): range for uh for Q4.
- 9:35
- Unidentified speaker (speaker_1): One key concern I would say uh one what's also getting portrayed in the media quite heavily is uh delays with data centers. Um, and I think that's also one concern uh with regards to core weave. Um what could slow down bringing that contracted power really online
- 9:55
- Unidentified speaker (speaker_1): and being able to harvest um that expected revenue.
- 10:01
- Unidentified speaker (UNKNOWN): Yeah,
- 10:01
- Unidentified speaker (speaker_0): sure. There's all types of um execution challenges that and um regulatory sort of challenges that that that neoclouts and hyperscalers have to go through um there's you know in terms of procuring the labor um and getting these things uh fur fully permitted um there's been a lot of the
- 10:21
- Unidentified speaker (speaker_0): state and local pushback i would say that the what we've seen like the blue wave outcome from the midterm elections, it's been a lot of pushback and moratoriums and things like that on on data centers.
- 10:35
- Unidentified speaker (UNKNOWN): Um,
- 10:35
- Unidentified speaker (speaker_0): and so all of these things are a challenge. I think the the good news is the 2027 capacity is um for core leave, but also for the industry, is for the most part under construction for sure, right? And and also uh for the most part ful fully permitted.
- 10:53
- Unidentified speaker (UNKNOWN): So
- 10:53
- Unidentified speaker (speaker_0): that's kind of the good news. We have to kind of wait and see what it does to future capacity beyond 2027. But I I despite all of the kind of pushback, um, data centers do bring in a lot of tax revenue. They do bring in construction jobs. And so we'll have to kind of wait and see what actually gets implemented uh in practice.
- 11:14
- Unidentified speaker (UNKNOWN): Yeah.
- 11:15
- Unidentified speaker (speaker_1): I think I also read somewhere that despite um the negativity to getting data centers into your um backyard, uh it brings also uh
- 11:26
- Unidentified speaker (speaker_1): some really high qualified jobs into more rural areas of of the US. So it's not all bad. Um
- 11:35
- Unidentified speaker (speaker_1): another key thing I want to touch about is on take or pay agreement. So my naive read of take or pay is that the customer pays for the capacity, whether they use it or not. So it doesn't really matter for core weave how much they actually use. Is that how the contract
- 11:55
- Unidentified speaker (speaker_1): works or are there any other considerations?
- 12:01
- Unidentified speaker (speaker_0): Yeah, absolutely. I mean uh
- 12:03
- Unidentified speaker (speaker_0): 98% of their business is take or pay, right? They're essentially doing nothing or very little on demand.
- 12:09
- Unidentified speaker (UNKNOWN): Um
- 12:10
- Unidentified speaker (speaker_0): and what that means is is it is contractually committed. The the customers like um can't back out. Now
- 12:17
- Unidentified speaker (speaker_0): there are certain like service level agreements and could be you know material breaches or whatnot if there's a if there's a delay,
- 12:25
- Unidentified speaker (UNKNOWN): like
- 12:25
- Unidentified speaker (speaker_0): a material delay in what in when that service comes online
- 12:29
- Unidentified speaker (speaker_0): or they're not um or the uptime isn't there. Um the customers could uh could walk. So there are, you know, exceptions um clearly written out um in those contracts. I think that in the um in the current em environment, in the current market, if a if anyone, if any of their customers walked, they could
- 12:48
- Unidentified speaker (speaker_0): recontract that out um at a higher price And so customers generally like aren't walking. And we saw that a little bit, I think, with Microsoft and some of the other neo clouds. But generally speaking, in the in this in in the current environment, um customers are not walking uh away from their contracts. Um on the on the flip side, if the
- 13:08
- Unidentified speaker (speaker_0): data center um developer has
- 13:13
- Unidentified speaker (speaker_0): you know if that if that development of data center is material delayed uh Coralief can also um potentially walk from those obligations so that that is kind of like a mitigating mitigating factor
- 13:29
- Unidentified speaker (speaker_1): can a customer um get out of contract obviously uh not super interesting for them at the moment but let's move into 2028
- 13:40
- Unidentified speaker (UNKNOWN): Yeah,
- 13:40
- Unidentified speaker (speaker_0): I mean if it if they've signed it and it's take or pay, they can't. They they would only uh be able to uh get out of the contract if there was a uh like a breach on or a failure to deliver or a failure to provide the availability um uh from from core weave.
- 13:57
- Unidentified speaker (speaker_1): What happens if core weave Weav delivers late?
- 14:02
- Unidentified speaker (speaker_0): Um if it if it is materially late, like beyond what's in the terms of their contract, um the customer can walk. But if if they did deliver late, um it's probably it's not gonna be because of availability of chips, right? Because I it's it's gonna be because of availability uh or or an
- 14:22
- Unidentified speaker (speaker_0): inability to bring up bring on that powered shell.
- 14:26
- Unidentified speaker (UNKNOWN): And
- 14:26
- Unidentified speaker (speaker_0): if that's the case, I think that core weave um
- 14:30
- Unidentified speaker (UNKNOWN): the
- 14:30
- Unidentified speaker (speaker_0): customer could walk, but core weave could also walk, essentially.
- 14:35
- Unidentified speaker (UNKNOWN): If
- 14:35
- Unidentified speaker (speaker_1): you like this episode, subscribe to Fix and Floating. You won't miss anything, and it really helps the show. And now back to the conversation. Could we classify because there's a breadth of neo clouds, you have the hyperscalers, uh who are bringing on data center capacity? Can we classify
- 14:55
- Unidentified speaker (speaker_1): those players into who is more sophisticated, also who is more experienced in developing data centers? Um, because it just seems to me that they are all everybody's just talking about this whole um
- 15:12
- Unidentified speaker (speaker_1): combined mass of of players. while it seems to me there must be at least some differentiation.
- 15:19
- Unidentified speaker (speaker_0): Yeah, no, there absolutely is. Look, the hyperscalers are going to be highly sophisticated.
- 15:26
- Unidentified speaker (UNKNOWN): Right?
- 15:26
- Unidentified speaker (speaker_0): They've been in this business a long time. Um, clearly, Amazon and Alphabet
- 15:33
- Unidentified speaker (UNKNOWN): and
- 15:33
- Unidentified speaker (speaker_0): Microsoft are going to be the most sophisticated in the market.
- 15:36
- Unidentified speaker (UNKNOWN): Um
- 15:38
- Unidentified speaker (UNKNOWN): there's
- 15:38
- Unidentified speaker (speaker_0): a mix of leasing and owning going on there. And Alphabet, I think, tends to own a bit more, and some of the other players, you know, tend to lease a bit more, but they're all fairly sophisticated in terms of bringing on capacity and managing
- 15:50
- Unidentified speaker (UNKNOWN): it.
- 15:51
- Unidentified speaker (speaker_0): The NeoClouds, a lot of them are, you know, former Bitcoin miners. They've kind of, you know, Corey've Corweav included too, right? I mean, these were natural gas traders that pivoted to Ethereum mining. that are now, you know, doing AI cloud infrastructure. And so I would say across the neo cloud space, there's there's generally um
- 16:12
- Unidentified speaker (UNKNOWN): a
- 16:12
- Unidentified speaker (speaker_0): bit less experience and and and and and less sophistication there. Um however, core we is is furthest along among the neo clouds. Like they've they've already kind of delivered, they're delivering at scale, right? They're at 1.7 gigawatts of action. active power,
- 16:28
- Unidentified speaker (UNKNOWN): and
- 16:28
- Unidentified speaker (speaker_0): that continues to scale. Whereas some of the other neo clouds have plans, you have lots of contracted capacity, but they don't necessarily have a lot of active capacity.
- 16:39
- Unidentified speaker (UNKNOWN): So
- 16:39
- Unidentified speaker (speaker_0): I think that's the differentiation there is like I I would put core weave more into that sophisticated bucket. But like for the broader neo neo-cloud cloud space, um there's a lot of uh trial and error, I would say, going on. right
- 16:51
- Unidentified speaker (UNKNOWN): now.
- 16:53
- Unidentified speaker (speaker_1): Um on top of um development uh capabilities, it also seems to me that there is also some differentiation on the software layer. Um is that something which is differentiating for Core Weave? Um, and does that help margins, or is
- 17:13
- Unidentified speaker (speaker_1): that only something which helps on the margin?
- 17:18
- Unidentified speaker (speaker_0): Yeah, I look a little bit. I I'll I'll give you kind of both perspectives here. And this is like so core we will say that um it's MFU, which is like model flops utilization, um, is is higher than the industry. Um and and the reasons for that is they're their pure play in AI infrastructure. They've kind of designed the software stack specifically
- 17:38
- Unidentified speaker (speaker_0): for GPU as a service. And so they don't have as much, you know, technical debt per se, you know, relative relative to the hyper hyperscaler. So they've done things like remove the hypervisor layer and other things on the software side to make it specifically tuned to this AI cloud business. They they tend to
- 17:58
- Unidentified speaker (speaker_0): get good pricing and margins. And they also have gotten like pretty good industry rankings as well. So like for example, the semi analyst um cluster max rankings, they they got like the platinum status tier. So there are some proof points. Um however, kind of like broadly speaking, they're they're they are
- 18:18
- Unidentified speaker (speaker_0): competing against very sophisticated players, like we talked about, all the hyperscalers. Um, I don't think there's a sustainable advantage here. Um, I would argue that this is more of a commodity. And at the end of the day,
- 18:32
- Unidentified speaker (UNKNOWN): um
- 18:33
- Unidentified speaker (UNKNOWN): this
- 18:34
- Unidentified speaker (speaker_0): the winners are going to be the ones that have large economies of scale and like an ability to execute. Um it they're doing a decent job on both fronts, but you know, I don't I don't necessarily think there's like an inherent or dirt. durable uh advantage on the on the software side
- 18:54
- Unidentified speaker (speaker_0): for them.
- 18:57
- Unidentified speaker (speaker_1): Microsoft I think is around or was two-thirds of revenue last year and is now um a bit more over one-third. OpenAI and Anthropic are also large parts of the backlog. I think backlog rose to over one hundred and thirty billion. Um
- 19:15
- Unidentified speaker (speaker_1): most of that come quite early in the quarter, so we haven't seen that many new deals. Um how much of the backlog depends on customers that are not investment grade?
- 19:29
- Unidentified speaker (UNKNOWN): Yeah,
- 19:30
- Unidentified speaker (speaker_0): I would say kind of
- 19:32
- Unidentified speaker (speaker_0): most of it is is not investment grade, right? And so it's really kind of their investment grade customers or meta is is is
- 19:40
- Unidentified speaker (UNKNOWN): the
- 19:40
- Unidentified speaker (speaker_0): biggest customer um that they have. And Microsoft has has went from being the biggest customer as now kind of like relatively you know smaller relative to Meta um and and and and OpenAI. And so I would say you know the majority of the the customers are are not investment grade.
- 19:59
- Unidentified speaker (UNKNOWN): Um
- 19:59
- Unidentified speaker (speaker_0): They're starting to get more enterprise business. uh enterprise customers, but that is still like fairly small. So I'd say it's it's AI labs and AI startups um and hyperscalers are kind of the big buckets there.
- 20:14
- Unidentified speaker (speaker_1): Understood. And the enterprise customer also more spot buyers, I would say, or short term buyers.
- 20:21
- Unidentified speaker (speaker_0): Yeah, they they tend to be doing the um more like three year contracts. Um so that's what the recent like the
- 20:30
- Unidentified speaker (UNKNOWN): the
- 20:30
- Unidentified speaker (speaker_0): recent facility they did, the five point five facility, was specifically targeted towards enterprise customers and those and they tend to buy in like two, three year contracts. Like they don't think five years out.
- 20:43
- Unidentified speaker (speaker_1): If you think about pricing, I think management um said that four to five year pri uh contracts now sign at twenty to twenty-five million per megawatt a year, and a short contract is up to forty million. Um the existing book was closer to ten to twelve million. Why is the
- 21:04
- Unidentified speaker (speaker_1): pricing curve so inverted?
- 21:08
- Unidentified speaker (speaker_0): Yeah, sure. So um it's just we're in such a tight market environment right now.
- 21:15
- Unidentified speaker (UNKNOWN): And
- 21:15
- Unidentified speaker (speaker_0): so I think I think inverted pricing is actually quite normal for tight markets. And so
- 21:23
- Unidentified speaker (speaker_0): there's a shortage of AI compute, and that's really driven by the shortage of powered shells.
- 21:29
- Unidentified speaker (UNKNOWN): And
- 21:30
- Unidentified speaker (speaker_0): so customers are willing to pay up for that capacity Um to meet cab customer demands, right? And to um to to grow their revenue and satisfy. and customers, it's really about customer attention.
- 21:42
- Unidentified speaker (UNKNOWN): Um
- 21:43
- Unidentified speaker (speaker_0): so
- 21:44
- Unidentified speaker (speaker_0): even if it's hurtful to margins,
- 21:48
- Unidentified speaker (UNKNOWN): right,
- 21:48
- Unidentified speaker (speaker_0): even if it's uneconomical, um, there's reasons to to that customers want
- 21:54
- Unidentified speaker (UNKNOWN): it. Um
- 21:55
- Unidentified speaker (speaker_0): there's also kind of a view that like, hey, just because we're supply constrained today doesn't mean we're gonna be supply constrained three, four years from now.
- 22:03
- Unidentified speaker (speaker_0): Right. And
- 22:05
- Unidentified speaker (UNKNOWN): um there
- 22:07
- Unidentified speaker (speaker_0): is a lot of supply coming on, right? We had about 45 gigawatts uh of data center capacity in the US at the end of 2025. That's probably like tripling over the next several years.
- 22:17
- Unidentified speaker (UNKNOWN): And
- 22:18
- Unidentified speaker (speaker_0): so I don't think the the current um we can't extrapolate the current um market dynamics that far in the future. And then just kind of generally speaking, the way that the cloud business works is the longer term commitment you and you make and the larger commitment you make, you you get kind of more favorable and uh better pricing
- 22:38
- Unidentified speaker (speaker_0): for that.
- 22:41
- Unidentified speaker (speaker_1): If you think about um the operating business model and think about um one specific project project, so core. CoreViv mostly doesn't own its own data center. At least it's the land and the power shell comes then from developers like Core Scientific. And
- 23:00
- Unidentified speaker (speaker_1): then it puts its own chips inside. Could you explain that relationship with the landlords?
- 23:10
- Unidentified speaker (speaker_0): Yeah, sure. So um
- 23:12
- Unidentified speaker (speaker_0): the developer, you know. finances and funds and pays for the pays for the data center. And that, you
- 23:19
- Unidentified speaker (UNKNOWN): know,
- 23:19
- Unidentified speaker (speaker_0): for like a one gigawatt data center, that could be somewhere on the order of 15 to 20 billion dollars.
- 23:26
- Unidentified speaker (UNKNOWN): Um
- 23:26
- Unidentified speaker (speaker_0): they they they lease that out to to to Corviv for a long-term lease. That could be like 15 or 16 years. It's generally about
- 23:36
- Unidentified speaker (UNKNOWN): two
- 23:36
- Unidentified speaker (speaker_0): $2 billion a year roughly in rent. plus energy. You're at
- 23:41
- Unidentified speaker (UNKNOWN): at
- 23:41
- Unidentified speaker (speaker_0): least $500 million a year. You know, um, so from a cash cost perspective,
- 23:47
- Unidentified speaker (UNKNOWN): um
- 23:48
- Unidentified speaker (speaker_0): Correav
- 23:49
- Unidentified speaker (UNKNOWN): can
- 23:49
- Unidentified speaker (speaker_0): be out, you know, call it two and a half, three billion dollars a year, you know, between data center rent and and and energy and you know, maybe some other maintenance costs as well. Um, but that's just the power show. chill. Like core Corey leave is responsible. for buying all the hardware and installing that um into the data center.
- 24:08
- Unidentified speaker (speaker_0): And that and that hardware and that hardware it has gone up over time, right? So for um like hopper chips it was, you know, um it was over twenty five billion and black ball was over thirty. And now Vararubin it's over thirty 35. five. So it's really like the the the the tech hardware CapEx costs is like 35 to 40
- 24:28
- Unidentified speaker (speaker_0): billion dollars on bleeding edge Nvidia chips um for for one gigawatt of capacity. So it's it's way more expensive than the than the actual data center, despite being a you know a short lived asset.
- 24:44
- Unidentified speaker (speaker_1): So we are talking around 60 sixty billion for the whole package. Um what I find personally appealing. about the credit is that it's essentially project finance of a holding company on top. So each box holds a customer contract, chips, and a loan secured on both. Um
- 25:05
- Unidentified speaker (speaker_1): so could you take one data center and walk us through how it is financed from the customer contract down to the loan.
- 25:17
- Unidentified speaker (speaker_0): Yeah, sure. So most of CoreWe's funding is done through these delayed draw term loans.
- 25:25
- Unidentified speaker (UNKNOWN): Um
- 25:25
- Unidentified speaker (speaker_0): it's the asset level
- 25:27
- Unidentified speaker (UNKNOWN): uh
- 25:28
- Unidentified speaker (speaker_0): funding that you talked about.
- 25:30
- Unidentified speaker (UNKNOWN): Um
- 25:30
- Unidentified speaker (speaker_0): those are collateralized by the GPUs.
- 25:35
- Unidentified speaker (UNKNOWN): They're
- 25:35
- Unidentified speaker (speaker_0): collateralized by importantly by the customer contract. contracts as well as the leases.
- 25:42
- Unidentified speaker (UNKNOWN): And
- 25:42
- Unidentified speaker (speaker_0): in most cases, they also have a core weave parent guarantee.
- 25:57
- Unidentified speaker (speaker_0): the cost of that hardware
- 25:59
- Unidentified speaker (UNKNOWN): um
- 25:59
- Unidentified speaker (speaker_0): using those facilities.
- 26:01
- Unidentified speaker (UNKNOWN): Um
- 26:01
- Unidentified speaker (speaker_0): Those are delayed. draw term loans, which means the money doesn't get borrowed until CoreWeave needs it to buy the CapEx.
- 26:09
- Unidentified speaker (UNKNOWN): And
- 26:09
- Unidentified speaker (speaker_0): they're also amortizing. So as CoreWeave
- 26:14
- Unidentified speaker (UNKNOWN): um
- 26:15
- Unidentified speaker (speaker_0): collects the
- 26:17
- Unidentified speaker (UNKNOWN): uh
- 26:17
- Unidentified speaker (speaker_0): payments from its customers, it it first gets used to pay down um the loan and then excess amount gets divided and back up to the parent.
- 26:32
- Unidentified speaker (speaker_1): Um for listeners um who aren't familiar with the term, what is a delayed draw term loan?
- 26:41
- Unidentified speaker (UNKNOWN): Yeah,
- 26:41
- Unidentified speaker (speaker_0): so basically it's it's just um the loan
- 26:45
- Unidentified speaker (speaker_0): Corey Corweave Venters enters into an agreement with the lender,
- 26:48
- Unidentified speaker (UNKNOWN): um
- 26:49
- Unidentified speaker (speaker_0): and but the the borrowing doesn't happen immediately when when the loan is signed. Um it's delayed draw, which means that that core weave um borrows the money when it needs it. It's basically when it when it has to outlay the cash to buy the the GPU servers. Um they'll draw down the loan.
- 27:08
- Unidentified speaker (UNKNOWN): Now
- 27:08
- Unidentified speaker (speaker_0): these these loans are they're also amortizing. Um so you know, Corweve Corweave is is paying down, you know um essentially principal and interest. They're using the the customer um payments to uh to to pay down the loan over time. And so um
- 27:27
- Unidentified speaker (UNKNOWN): if
- 27:27
- Unidentified speaker (speaker_0): you think about it, most of their contracts are five years.
- 27:32
- Unidentified speaker (UNKNOWN): Um
- 27:33
- Unidentified speaker (speaker_0): and the the loans are designed to kind of match that maturity profile.
- 27:39
- Unidentified speaker (speaker_0): of the contracts. And so um when the cost term of a contract ends, those loans will be fully paid off.
- 27:50
- Unidentified speaker (speaker_1): How much of the delayed draw term loans have recourse to core weave itself?
- 27:59
- Unidentified speaker (speaker_0): Um almost all of them. So the exception
- 28:02
- Unidentified speaker (UNKNOWN): is a very good thing.
- 28:03
- Unidentified speaker (speaker_0): is the Meta facility. This is their the 4.0 facility. This is an investment grade contract uh customer.
- 28:11
- Unidentified speaker (UNKNOWN): And
- 28:11
- Unidentified speaker (speaker_0): they basically didn't need the the parent guarantee from Corweev. Um because you have you're contractually committed by by by Meta uh essentially.
- 28:22
- Unidentified speaker (UNKNOWN): Um
- 28:23
- Unidentified speaker (speaker_0): so and I think generally speaking, the the most valuable part.
- 28:27
- Unidentified speaker (speaker_0): um here is is the collateral, right? It's the customer contracts and it's the the value of the GPUs, whereas the the parent guarantee um is sort of like the less valuable aspect. And so if you have a customer like Meta, um you don't you don't need that parent guarantee. Um and so that facility is non Non-recourse. recourse. Core Weave does have another
- 28:47
- Unidentified speaker (speaker_0): IG facility coming so it's very possible that one looks like uh the meta one as well. Um and the meta one I think is is eight and a half uh billion dollars. So it's a it's a it's a rather large
- 28:58
- Unidentified speaker (UNKNOWN): um
- 28:59
- Unidentified speaker (speaker_0): you know loan size once that's fully drawn upon.
- 29:05
- Unidentified speaker (speaker_1): What happens if um let's say a non-investment grade Great um
- 29:11
- Unidentified speaker (speaker_1): client is unable to uh fulfill its obligations? and essentially defaults um on the deal it has um with Corby. So let's say Anthropic goes bust in in two years um and they can't pay um for the compute. Um what happens um
- 29:31
- Unidentified speaker (speaker_1): as an afterthought?
- 29:34
- Unidentified speaker (UNKNOWN): Yeah,
- 29:34
- Unidentified speaker (speaker_0): so there's a there's a few different types of collateral here, right? There's there's the collateral from the it's backed by the backed by the customer contracts.
- 29:43
- Unidentified speaker (UNKNOWN): Right.
- 29:43
- Unidentified speaker (speaker_0): It's it's all but it's also backed by the GPUs, um, and it's backed by the at by the leases too. So there's like the the right to use the lease as an asset.
- 29:51
- Unidentified speaker (UNKNOWN): Um
- 29:52
- Unidentified speaker (speaker_0): so there's lots of different ways for for creditors to get their money back. Um they could they could recontract it out to another customer,
- 30:00
- Unidentified speaker (UNKNOWN): um,
- 30:01
- Unidentified speaker (speaker_0): they could sell the GPUs.
- 30:03
- Unidentified speaker (speaker_0): They could
- 30:05
- Unidentified speaker (speaker_0): monetize the leases. And they also have a potential claim on anthropic as well, you know, through uh through a hypothetical um default type type type type type scenario.
- 30:19
- Unidentified speaker (speaker_1): But if that doesn't work, what happens? What is their claim, um, their recourse towards core weed? weave?
- 30:28
- Unidentified speaker (speaker_0): Correct. So for the for the facilities which are mostly recourse, they have a they have a parent guarantee. And then uh core weave would have to
- 30:36
- Unidentified speaker (UNKNOWN): um
- 30:37
- Unidentified speaker (speaker_0): would have to cover the short shortfall. So core weave is on the hook, you know, outside of all if they if the creditors can't get made whole, like from all that collateral they have and those like those different avenues that we talked about, um
- 30:50
- Unidentified speaker (UNKNOWN): that they
- 30:51
- Unidentified speaker (speaker_0): would look to the to the parent guarantee um to to obtain, you know, to get like additional funds um to to help cover the the creditors in that in that box.
- 31:02
- Unidentified speaker (UNKNOWN): This
- 31:02
- Unidentified speaker (speaker_1): year, core we've collected prepayments on seventy percent of its deals, some of them up to hundred percent of the contract value. So they got essentially all the contract value in advance. which is obviously a big cash inflow um now. Um and it expects about
- 31:22
- Unidentified speaker (speaker_1): nine billion in prepayments and deposits in the next three months. So another big cash inflow. Um what do the prepayments do to the economics of a data center?
- 31:36
- Unidentified speaker (speaker_0): Yeah so um I think the the market condition conditions are extremely tight right now. And so I think that a lot of Neo Clouds and hyperscalers included, like Oracle, are seeking out prepayments and demanding prepayments. And I think a lot of customers are happy to do it. And so I
- 31:56
- Unidentified speaker (speaker_0): think that um it could change the economics somewhat, right? A customer that's making a prepayment uh obviously wants something in return. for that. But I think first and foremost, in this extremely tight market environment, um customers are just are desperate to get the capacity. Um and you know, making a
- 32:15
- Unidentified speaker (speaker_0): prepayment um is is essentially the cost of doing business right
- 32:20
- Unidentified speaker (UNKNOWN): now.
- 32:21
- Unidentified speaker (speaker_1): So most of their clients, even the the hyperscalers, are just struggling for computing. compute and they are very m very happy to secure it. Corbeev potentially doesn't need to give up that much in pricing and margin power at the moment.
- 32:38
- Unidentified speaker (speaker_0): Correct. Correct. In the current environment. That's right.
- 32:43
- Unidentified speaker (speaker_1): If something goes wrong, does the customer who prepaid rank ahead of the bondholders?
- 32:51
- Unidentified speaker (UNKNOWN): This
- 32:51
- Unidentified speaker (speaker_0): is kind of a tricky question. Um
- 32:55
- Unidentified speaker (speaker_0): Yeah, a customer prepayment is is like uh an unsecure claim, but I I think it really depends on where that prepayment was made and and like is it is it a prepayment to to customer parent? Is core leaf parent, or is it a is it a prepayment to um
- 33:12
- Unidentified speaker (UNKNOWN): to
- 33:13
- Unidentified speaker (speaker_0): the to the underlying box?
- 33:15
- Unidentified speaker (UNKNOWN): um
- 33:16
- Unidentified speaker (speaker_0): that you know the the wholly owned subsidiary, right? Um that that the that the customer has the the actual contract with. Um so I think you know I'm not entirely sure
- 33:28
- Unidentified speaker (UNKNOWN): um
- 33:30
- Unidentified speaker (speaker_0): where that would would rank.
- 33:32
- Unidentified speaker (UNKNOWN): If
- 33:32
- Unidentified speaker (speaker_1): you're looking for more credit credit-related related content, check out fixedandfloating.com. And now back to the show.
- 33:46
- Unidentified speaker (speaker_1): There was one in August at SOFA plus 550, so materially higher. The next one should be investment grade, as you mentioned, yielding about seven to seven and a half percent, um, financing more than 90% of the capex. So at the top of the rate. range you gave us earlier. How much does Corviv
- 34:06
- Unidentified speaker (speaker_1): actually make on a contract once it is financed like that? So what's really the equity return on that investment?
- 34:16
- Unidentified speaker (speaker_0): Yeah, so the kind of the unit economics that they've laid out is um over the lifetime of a contract, it's kind of like 24% operating market. margin and 16%
- 34:32
- Unidentified speaker (speaker_0): kind of a pre-tax margin
- 34:35
- Unidentified speaker (speaker_0): after after all those interest uh payments uh for the financing. So that's that's kind of the unit economics. Um I think those have maybe like a bit of rosy assumptions in them, uh including the cost of financing. And so maybe it's a bit
- 34:55
- Unidentified speaker (speaker_0): lower than that. But the the offset here is that the pricing, recent pricing has been like really strong. They've gone through a number of price increases. On the recent call, they talked about, I think you mentioned it earlier, like the twenty to twenty-five million dollars per megawatt on a new.
- 35:15
- Unidentified speaker (speaker_0): four, five year contract. And so, um, you know, I would say those two things kind of offset and that that's, you know, in the current environment, maybe maybe the Indian economics are are holding.
- 35:29
- Unidentified speaker (UNKNOWN): Um,
- 35:30
- Unidentified speaker (speaker_1): you mentioned the sixteen percent. Is that an a return on asset or s or is that certain a return on equity? So only under ten percent they are chipping in.
- 35:39
- Unidentified speaker (speaker_0): Um that's kind of the like sort of like a pre-tax margin is is is the way that I think about that. Um
- 35:46
- Unidentified speaker (speaker_1): on the whole asset.
- 35:48
- Unidentified speaker (speaker_0): So
- 35:50
- Unidentified speaker (speaker_0): yeah over like the duration of like the five year
- 35:53
- Unidentified speaker (UNKNOWN): uh
- 35:54
- Unidentified speaker (speaker_0): the five years that a five year contract essentially.
- 35:58
- Unidentified speaker (UNKNOWN): Um
- 35:59
- Unidentified speaker (speaker_0): now
- 36:01
- Unidentified speaker (speaker_0): there's potential upsider upside or downside to that, right? So depending on because they own the the chips, you know, once um you know once those are fully paid off. And so if they could recontract out that capacity uh at reasonable rates, they could see upside to that figure. But if they can't find a customer
- 36:21
- Unidentified speaker (UNKNOWN): um,
- 36:22
- Unidentified speaker (UNKNOWN): you know,
- 36:23
- Unidentified speaker (speaker_0): to renew that capacity, then then there's obviously a lot of downside.
- 36:27
- Unidentified speaker (speaker_0): But for the sort of the initial five years, that's kind of where it lands.
- 36:33
- Unidentified speaker (speaker_1): That makes totally sense. I think you just mentioned um being able to put a new contract um on the um existing chips. Um one thing we haven't talked about is that they're still on the hook for another probably 10 ten years on on the powered shell.
- 36:53
- Unidentified speaker (speaker_1): Um is that an issue that they need to pay another ten years um two billion a year, which um kind of is very dilutive for um the equity.
- 37:08
- Unidentified speaker (UNKNOWN): Yeah,
- 37:08
- Unidentified speaker (speaker_0): no, it it's uh it's kind of a major issue, right? That this is a a pretty material medium term risk for core Corleave. leave
- 37:14
- Unidentified speaker (UNKNOWN): um
- 37:15
- Unidentified speaker (speaker_0): that they have these
- 37:17
- Unidentified speaker (UNKNOWN): um you
- 37:18
- Unidentified speaker (speaker_0): know, w we might we could potentially overbuild data center capacity, particularly in the United States, just given how many of these like gigawatt plus projects that are coming online um over the next several years.
- 37:31
- Unidentified speaker (UNKNOWN): I
- 37:31
- Unidentified speaker (speaker_0): think one of the issues is that um some of their customers
- 37:36
- Unidentified speaker (UNKNOWN): are,
- 37:36
- Unidentified speaker (speaker_0): you know, the hyperscalers
- 37:39
- Unidentified speaker (speaker_0): and they like meta and and microsoft they own their own data centers and they have long-term leases themselves on data centers. So if there's too much capacity in the industry, why would they renew with CoreWeath? And so I think I think there's um a material like medium term risk
- 37:57
- Unidentified speaker (speaker_0): um that
- 37:58
- Unidentified speaker (speaker_0): that there's just you know on renewals that there's kind of too much capacity out there.
- 38:05
- Unidentified speaker (speaker_1): If you talk about uh cost of capital, um and especially the cost of capital driven um by uh delayed term loan um rates and in the past unsecured financing at the parent level, is there a point where a new contract doesn't make sense?
- 38:27
- Unidentified speaker (UNKNOWN): Yeah,
- 38:27
- Unidentified speaker (speaker_0): I know it's a good question. I I would say there's kind of a few factors besides the cost of financing. It's also like the cost of the chips and the and the um and the and the pricing on new contracts. So depending on kind of where all those things are um at any given time, I don't think they've kind of reached reach that point where
- 38:47
- Unidentified speaker (speaker_0): it doesn't make sense. Um I think we've we've seen them pull back from the unsecured market, right, in double digit yields. And so I think that kind of gives you an indication of they want to be at like 10% or lower in terms of uh cost of debt financing.
- 39:01
- Unidentified speaker (UNKNOWN): Um
- 39:02
- Unidentified speaker (speaker_0): in the car and market, I you know there's maybe a bit more wiggle room. Um, but I think that um you know as long as they kind of stay under that, you know, 10% ten percent level um it you know it it it makes sense to to continue to to move forward in you know, given the demand that we're currently seeing.
- 39:22
- Unidentified speaker (UNKNOWN): We
- 39:22
- Unidentified speaker (speaker_1): touched upon operating margins before. Um, they were about ten percent this quarter, net loss was still about seven hundred million
- 39:32
- Unidentified speaker (UNKNOWN): uh
- 39:32
- Unidentified speaker (speaker_1): USD. Um, could you explain for listeners why the company still loses money?
- 39:39
- Unidentified speaker (speaker_0): Yeah, so there's a few issues here. Um one is the ramp up costs, right? Because it's, you know, um it takes a few months to kind of get profitable for each new equipment that you install. Um it just there's takes takes time to kind of when you from when you start paying the lease to get it installed and start, you know, uh collecting from the customer and then
- 39:59
- Unidentified speaker (speaker_0): ramping up in an efficient manner. Um and since they're not like a huge business yet, right? They just basically have pure play and they're they're kind of gaining scale as we speak, um, there is there's sufficient ramp up costs. I think the other issue, like when you think about a free cash flow perspective, there's a huge
- 40:19
- Unidentified speaker (speaker_0): timing related impact here where Corleave Corleaven, and the whole business,
- 40:23
- Unidentified speaker (UNKNOWN): you
- 40:24
- Unidentified speaker (speaker_0): have upfront capex and and rattable revenue And so I think historically it's probably taken close to like four years, I would estimate, to get to that, you know, free cash flow cumulative break-even.
- 40:39
- Unidentified speaker (speaker_0): With the recent pricing where it is, it that's obviously compressed maybe it's like two to three years. My numbers are more conservative, right? Than the numbers you're probably hearing elsewhere. I'm including overhead and things and like that. and not talking about, you know, extrapolating short term contracts at at 40 million a uh a megawatt. But that's kind of how
- 40:59
- Unidentified speaker (speaker_0): I think about it. So there's uh there's the the ramp up impact on margins, but there's also this timing, really painful timing related free cash flow burn. That doesn't mean it's not profitable or cash generating for the life of the business, um, but it's a it's a pretty major headwind, you know, as you're um as you're as you're buying the capex.
- 41:21
- Unidentified speaker (speaker_1): Despite that time related uh impact, they now guide uh to a positive net adjusted net income next year. What has to go right?
- 41:33
- Unidentified speaker (speaker_0): I I think they they need to execute, they need to continue to to to to scale. Um uh in terms of their active power. Um so if they see material delays to that, that's gonna be an issue. They need to kind of continue to uh sign new customer uh contracts at uh at profitable uh pricing and margins.
- 41:55
- Unidentified speaker (UNKNOWN): Um
- 41:55
- Unidentified speaker (speaker_1): on Tuesday management said that they won't come to the bond market because it's at current yields, it's uneconomic for them. Um
- 42:05
- Unidentified speaker (speaker_1): what does that mean for bondholders? Um, especially as more and more funding is coming now ahead of them.
- 42:14
- Unidentified speaker (speaker_0): Yeah, so um
- 42:17
- Unidentified speaker (speaker_0): most of it they're they're out they're primed, right? They're they're there's a lot of they're layered essentially behind a lot of um secured debt, and and that secured debt is going to continue to increase. Um and the the unsecured debt is gonna be probably be a bit more flattish because they've said they they they're not coming to unsecured. They're basically shut out of that market with fifteen percent yields
- 42:37
- Unidentified speaker (speaker_0): anyway. They've already done so much in terms of convert, so I'm not sure how much more they could do there. Um so I think most of most of the debt going forward is gonna going to be um at that at that secured um level.
- 42:50
- Unidentified speaker (UNKNOWN): So it's
- 42:51
- Unidentified speaker (speaker_0): it's not great you know to be to be layered behind all this debt. But keep in mind all of that secured debt basically sits at those at those boxes. They are self-amortizing, you know, as the as the customers pay um their their their monthly bills, those are used to kind of pay down the debt uh at those at those boxes. So
- 43:11
- Unidentified speaker (speaker_0): it's not like the secured secure debt is is sitting at the um at the parent level.
- 43:16
- Unidentified speaker (speaker_1): If you think about debt in general, I think it increased by about twelve billion in the quarter to around 47 billion and gross leverage is now around eight times
- 43:27
- Unidentified speaker (UNKNOWN): uh
- 43:27
- Unidentified speaker (speaker_1): EBDA. On consensus numbers, the company still burns about sixty 60 billion of free cash flow from twenty 27 seven to twenty 29. nine. Um could he give you us a split of all the financing forms where that money should come from?
- 43:45
- Unidentified speaker (speaker_0): Yeah, so uh we talked about like that 78 to 90 percent loan to cost for the delayed draw term loans. So that that's gonna have to be the bulk of
- 43:53
- Unidentified speaker (UNKNOWN): it.
- 43:54
- Unidentified speaker (speaker_0): Um instead of doing unsecured debt, they're gonna going lean to more heavily into customer prepayments.
- 44:00
- Unidentified speaker (UNKNOWN): If
- 44:00
- Unidentified speaker (speaker_0): you think the the way about prepayment tip typically works, is let's say you enter into a five-year contract, maybe the customer pays you up front for your five.
- 44:10
- Unidentified speaker (UNKNOWN): Right.
- 44:10
- Unidentified speaker (speaker_0): So you could kind of get what you need, right? Between a combination, if you're if your loan to cost is 80%, and then you're getting a uh call it like a 20% prepayment. Um, that kind of gets you most of the way there. Um now they do have overhead and other costs and things like that, and like not everything. every contract has prepayments. And so they do need additional sources
- 44:30
- Unidentified speaker (speaker_0): of funding. Um, I think the the other area that they they're they're going to to lean heavily into, um,
- 44:37
- Unidentified speaker (UNKNOWN): I
- 44:37
- Unidentified speaker (speaker_0): would say is is the OEM uh financing, which has started to grow, but I think they could do more there. Um, specifically, Dell is a a major supplier of of AI servers.
- 44:50
- Unidentified speaker (UNKNOWN): Um
- 44:51
- Unidentified speaker (speaker_0): and and to to Core Leave and and the whole industry, and they've got a really strong balance sheet. Um Dell's leverage is well below its own target. Um so there's a lot of room there to uh to extend more financing to customers and help like continue to fuel their own growth. And so I think I think you know, Corleave is is going to uh take
- 45:11
- Unidentified speaker (speaker_0): advantage of that in my opinion.
- 45:14
- Unidentified speaker (UNKNOWN): I think
- 45:14
- Unidentified speaker (speaker_1): that's a really strong point. Um, I think that the general issue as well is that the two biggest um uses of compute, anthropic and OpenAI, they don't have a strong balance sheet, right? So to facilitate uh um AI build out, you need these other very strong balance sheets sheet to
- 45:34
- Unidentified speaker (speaker_1): facilitate that. 130,000 GPUs came of off contract in 2027 and another 70k in 2028. Most of the data center leases run much longer. What happens once such a contract runs out and Corviv is still on the hook of the lease?
- 45:54
- Unidentified speaker (speaker_0): Yeah, so um
- 45:57
- Unidentified speaker (UNKNOWN): they
- 45:57
- Unidentified speaker (speaker_0): need to there's a few different options, right? They they could recontract it out to the existing customer, they could recontract
- 46:02
- Unidentified speaker (UNKNOWN): it.
- 46:03
- Unidentified speaker (speaker_0): out to a to a new customer. Um, they could they could kind of serve it up as demands, you know, on demands. Um there's a there's a variety of options that they could do um once those once those contracts expire. Um the
- 46:18
- Unidentified speaker (UNKNOWN): good
- 46:19
- Unidentified speaker (speaker_0): news for them in in the current environment, the stuff that is expiring is actually being recontracted out at pricing. that's higher than their original rates, which is which is kind of remarkable. That's like not a normal situation, right? Like these things depreciate at you know over time and um
- 46:37
- Unidentified speaker (UNKNOWN): you
- 46:37
- Unidentified speaker (speaker_0): know as as you know the technology advances.
- 46:42
- Unidentified speaker (speaker_1): So even for older chips.
- 46:47
- Unidentified speaker (speaker_0): Yeah, for older chips, they're they're recontracting out at higher rates. than they had originally. Um, and and I think what's what's really was a uh I think a powerful kind of comment that that management made um on the on the Q3 call was they talked about the nine billion of prepayments that they're that they're getting in Q4.
- 47:07
- Unidentified speaker (speaker_0): Three billion of those prepayments
- 47:11
- Unidentified speaker (speaker_0): are for recontracting out existing capacity.
- 47:15
- Unidentified speaker (speaker_0): So customers are kind of so desperate for capacity right now that they're willing to make a prepayment
- 47:22
- Unidentified speaker (UNKNOWN): on
- 47:22
- Unidentified speaker (speaker_0): an old generation of a chip
- 47:26
- Unidentified speaker (speaker_0): that that there's no CoreWeaf doesn't even have any capex associated with it. There's like no that they already own the the that capex. The debt is fully paid off on that. Um They own they own the GPUs and um the custom customers are willing to to prepay uh even for the older generation chips.
- 47:47
- Unidentified speaker (speaker_1): Um cob has about ninety billion in of future lease payments for listeners who are not familiar with um leases. What exactly are future lease payments and why use? an undiscounted number?
- 48:07
- Unidentified speaker (speaker_0): Okay. Yep. And the numbers actually increased a bit. Uh it's about 110 billion as of Q three from the most recent um 10 Q filing. And so the way leases work
- 48:22
- Unidentified speaker (speaker_0): is
- 48:25
- Unidentified speaker (speaker_0): when the leases actually commence
- 48:29
- Unidentified speaker (UNKNOWN): and
- 48:29
- Unidentified speaker (speaker_0): the power shell is made available to Coreweave,
- 48:32
- Unidentified speaker (UNKNOWN): it
- 48:33
- Unidentified speaker (speaker_0): comes on the balance sheet.
- 48:35
- Unidentified speaker (UNKNOWN): And
- 48:35
- Unidentified speaker (speaker_0): it comes on the balance sheet
- 48:38
- Unidentified speaker (speaker_0): at present value of those future uh lease payments.
- 48:42
- Unidentified speaker (UNKNOWN): And
- 48:42
- Unidentified speaker (speaker_0): that and that's for both operating leases and financing leases.
- 48:46
- Unidentified speaker (UNKNOWN): Now,
- 48:46
- Unidentified speaker (speaker_0): in addition to what's on the balance sheet, there's also leases that have not yet commenced.
- 48:53
- Unidentified speaker (speaker_0): Those are off balance sheet and they're typically disclosed in the K's and the Q's.
- 49:00
- Unidentified speaker (speaker_0): Those are not yet discounted to present value. Those are undiscounted or future value of the payments.
- 49:08
- Unidentified speaker (UNKNOWN): What
- 49:08
- Unidentified speaker (speaker_0): I like to do is get kind of um
- 49:13
- Unidentified speaker (UNKNOWN): a
- 49:14
- Unidentified speaker (speaker_0): total undiscounted.
- 49:15
- Unidentified speaker (speaker_0): number to keep things simple. What's the total amount of future rent that they have to pay?
- 49:22
- Unidentified speaker (UNKNOWN): And
- 49:22
- Unidentified speaker (speaker_0): then we could use that to estimate future rent payments. We could compare that with the RPO, which is $133 billion.
- 49:33
- Unidentified speaker (UNKNOWN): Um
- 49:34
- Unidentified speaker (speaker_0): so I just I think it's a lot more simple to kind of think about the undiscounted total amount
- 49:41
- Unidentified speaker (UNKNOWN): of
- 49:41
- Unidentified speaker (speaker_0): future lease payments.
- 49:45
- Unidentified speaker (speaker_1): Does the circular financing uh worry you?
- 49:51
- Unidentified speaker (speaker_0): Yeah, uh for sure. I mean the the circular financing, this whole industry, this whole AI build-out out is has been kind of perpetuated by circular financing. You know, think back to the original. deal between Microsoft and and OpenAI, where they invest in OpenAI and OpenAI
- 50:08
- Unidentified speaker (UNKNOWN): um,
- 50:08
- Unidentified speaker (speaker_0): you know, essentially agrees to use them exclusively. It's not exclusive anymore, but um, you know, it's that was kind of the first circular deal and and almost every deal since then has it has had some circular component. Um Clearly you have Nvidia driving a lot of these circular type investments. Um They've they've provided they're they're an equity owner.
- 50:28
- Unidentified speaker (speaker_0): They've provided equity uh for Core Weave.
- 50:31
- Unidentified speaker (UNKNOWN): And
- 50:31
- Unidentified speaker (speaker_0): they've also provided a backstop on um on JPUs. Like there's a certain amount of capacity that if that if Core Weave doesn't sell to other customers, they'll step in and buy it. But it's only like six or seven billion dollars. It's not um
- 50:47
- Unidentified speaker (speaker_0): that meaningful relative to like the $133 billion of RPO that they have from customers and the the substantial amount of more RPO that they're going to need um to to to to kind of fill up all of their um the contracted capacity from all their data centers. The other kind of thing I think about
- 51:07
- Unidentified speaker (speaker_0): as it relates to this is Corweev Corleave knows it has this backstop, which means um it it just kind of emboldens them to be a bit more aggressive. So I'm not actually sure that it provides much uh safety uh in the way of uh creditors.
- 51:28
- Unidentified speaker (speaker_1): You just mentioned that um Corweav actually needs to um sign a lot more um deals to um essentially fill up all the uh contracted compute could you shed a bit more light on how we can think about um the sales activity of
- 51:48
- Unidentified speaker (speaker_1): of core weave in the future
- 51:51
- Unidentified speaker (UNKNOWN): and
- 51:51
- Unidentified speaker (speaker_1): how that makes it really a sustainable um structure I would say
- 51:56
- Unidentified speaker (speaker_0): yeah so um
- 51:58
- Unidentified speaker (UNKNOWN): you
- 51:59
- Unidentified speaker (speaker_0): know we talked about contracted capacity at like 4.6 four point gigawatts six
- 52:04
- Unidentified speaker (speaker_0): Um
- 52:05
- Unidentified speaker (UNKNOWN): they
- 52:05
- Unidentified speaker (speaker_0): don't disclose how much of that is sold to customers, but we could kind of like do some financial gymnastics to try and estimate that figure. Um we do have the backlog of of 133 billion, right? Um so we could estimate, you know, what's the average duration on that and try and
- 52:25
- Unidentified speaker (speaker_0): determine um you know how much is actually contracted. So I've estimated probably there's like two gigawatts or so um that's not yet sold to customers. And I I think the the comments on the on the Q three call kind of support that because the company said that they've um they haven't released start to sell 2028
- 52:45
- Unidentified speaker (speaker_0): capacity yet and they're waiting for that to to season further, which kind of makes sense because like pricing is still going up.
- 52:51
- Unidentified speaker (speaker_1): They are also trying to kind of time when they sell the capacity. So it doesn't make sense for them to sell everything now because you still have this um inverted curve
- 53:04
- Unidentified speaker (UNKNOWN): that
- 53:04
- Unidentified speaker (speaker_1): a but a pricing is still pretty pretty high.
- 53:08
- Unidentified speaker (speaker_0): Yeah they the pricing is still kind of moving higher on new contracts. So that's one reason
- 53:14
- Unidentified speaker (UNKNOWN): And
- 53:15
- Unidentified speaker (speaker_0): and that's certainly the reason that they gave. Um, the other kind of thing to think about is
- 53:22
- Unidentified speaker (speaker_0): each new year there's like a new generation of chips and we don't necessarily know the pricing of those chips yet
- 53:30
- Unidentified speaker (UNKNOWN): and
- 53:30
- Unidentified speaker (speaker_0): we don't know the efficiency yet of those chips. And so it's easier to kind of sell to customers the existing architectures. Um because that's how these things are priced, right? They're priced on like what type of GPU, right? And then, you know, the the the duration essentially and the
- 53:50
- Unidentified speaker (speaker_0): amount. So it it's just, I think, a bit more challenging to sell future generations
- 53:59
- Unidentified speaker (speaker_0): um when we don't
- 54:02
- Unidentified speaker (UNKNOWN): know
- 54:02
- Unidentified speaker (speaker_0): the price or the um efficiency or the or the energy efficient efficiency as well.
- 54:13
- Unidentified speaker (speaker_1): Is um bring your own chips becoming a thing there? So you have for example um anthroping coming um with its own chips um only using the the shell Corviv has leased and there is they're an essentially um only providing the um the software layer and um
- 54:32
- Unidentified speaker (speaker_1): essentially managing the facility
- 54:37
- Unidentified speaker (UNKNOWN): yeah
- 54:37
- Unidentified speaker (speaker_0): so I don't think Corviv has done any of these like sort of bring your own hardware type arrangements. Um
- 54:44
- Unidentified speaker (speaker_0): you know Oracle is certainly doing some of them and I think it really depends on the
- 54:51
- Unidentified speaker (UNKNOWN): um
- 54:52
- Unidentified speaker (speaker_0): the chip architecture as well. And so, you know, if you have like um
- 55:00
- Unidentified speaker (speaker_0): you know, if you have a um like OpenAI has you know its own custom ASICs like jalapeno,
- 55:09
- Unidentified speaker (UNKNOWN): um
- 55:10
- Unidentified speaker (speaker_0): and maybe they're getting funding there um from uh a you know f
- 55:14
- Unidentified speaker (UNKNOWN): for
- 55:14
- Unidentified speaker (speaker_0): from an industry partner.
- 55:16
- Unidentified speaker (UNKNOWN): Um you
- 55:17
- Unidentified speaker (speaker_0): know similarly with with TPUs. Um I just I'm not sure how much
- 55:23
- Unidentified speaker (UNKNOWN): of
- 55:23
- Unidentified speaker (speaker_0): this kind of bring your own chips we're seeing on the on the Nvidia side of things.
- 55:30
- Unidentified speaker (speaker_1): In some ways, Core Weav looks pretty close to a read. Um you have long leases on one side and and contracted rent. um on on the other. Um how do you value a company like that?
- 55:45
- Unidentified speaker (speaker_0): Yeah so this is a um
- 55:48
- Unidentified speaker (UNKNOWN): I
- 55:48
- Unidentified speaker (speaker_0): don't know I I guess I I don't necessarily I don't know if I look at it a read because they're they're kind of like on the other end of the lease
- 55:55
- Unidentified speaker (UNKNOWN): um
- 55:56
- Unidentified speaker (speaker_0): where they are the they're essentially the tenant um and they're they're basically look like a like a cloud provider,
- 56:04
- Unidentified speaker (UNKNOWN): um,
- 56:04
- Unidentified speaker (speaker_0): just a very specific type of of of cloud provider. Um you know, it's hard to value a company like this because it's it's so extremely capital intensive. Um and I I think that, you know,
- 56:17
- Unidentified speaker (speaker_0): certainly could could create some some boom bust, you know, particularly on the um on the on the on the data center leasing side of things.
- 56:26
- Unidentified speaker (UNKNOWN): Um
- 56:28
- Unidentified speaker (speaker_0): and so you know depreciation is just such a high part of their costs.
- 56:33
- Unidentified speaker (speaker_1): Is EBTA the right measure when uh chips need to be replaced pretty much every few years?
- 56:43
- Unidentified speaker (UNKNOWN): Yeah,
- 56:43
- Unidentified speaker (speaker_0): I look at the end of the day, I don't think it matters so much what metric you're using as long as you're handicapping it appropriately. So I if you do use IPITA, I would kind of make sure it's uh somewhat modest multiple, given how capital intensive it is and you know how much depreciation there is in the business. Um, you know, you could look at
- 57:03
- Unidentified speaker (speaker_0): a um
- 57:05
- Unidentified speaker (UNKNOWN): you know
- 57:05
- Unidentified speaker (speaker_0): an EVD oper operating profit, or you could kind of look at like free cash flow on a normalized basis, like over the duration of a customer contract and um and and value it that way. as
- 57:15
- Unidentified speaker (UNKNOWN): well
- 57:17
- Unidentified speaker (speaker_1): what are the agencies looking at I mean uh bonds are still rated double B minus
- 57:25
- Unidentified speaker (speaker_0): bonds are clearly showing a lot more stress than than than where the ratings are.
- 57:31
- Unidentified speaker (UNKNOWN): Um
- 57:32
- Unidentified speaker (speaker_0): I think the the market is looking at you know some of these risks
- 57:36
- Unidentified speaker (UNKNOWN): um
- 57:38
- Unidentified speaker (speaker_0): the agencies are
- 57:42
- Unidentified speaker (UNKNOWN): um I
- 57:43
- Unidentified speaker (speaker_0): don't I think I think are taking kind of a wait and see approach. Um so they haven't they're not going to necessarily downgrade just because the market is is um
- 57:54
- Unidentified speaker (speaker_0): panicking.
- 57:56
- Unidentified speaker (UNKNOWN): Um
- 57:56
- Unidentified speaker (speaker_0): I think that they're waiting to see, you know, if
- 57:59
- Unidentified speaker (UNKNOWN): um
- 58:00
- Unidentified speaker (speaker_0): if fundamentals change, if there are any kind of real signs of stress here for for Corweave? Um if they can't get the funding they need or you know fundamentals turn or things like that. But um you know clearly there's a there's a pretty large disconnect here between the ratings and where the um the bonds are trading.
- 58:20
- Unidentified speaker (speaker_1): How do um the bonds compare with other AI credits, like for example, Oracle or the other new? neo clouds.
- 58:29
- Unidentified speaker (UNKNOWN): Yeah.
- 58:30
- Unidentified speaker (speaker_0): So Corey's the only one that has
- 58:33
- Unidentified speaker (speaker_0): non-convertible debts, uh bonds uh among the um among the neo clouds.
- 58:41
- Unidentified speaker (UNKNOWN): Um
- 58:42
- Unidentified speaker (UNKNOWN): I
- 58:42
- Unidentified speaker (speaker_0): I would say there's like signs of stress throughout the industry. Oracle um you know trading around 300 basis points per point spread.
- 58:51
- Unidentified speaker (UNKNOWN): Um
- 58:52
- Unidentified speaker (speaker_0): that's a triple B name that's trading like a double B, right? So not not that kind of different from from from Corviv, which is like you know, a B rated credit that's trading kind of like a like a triple C, um, or or somewhere in between. So um,
- 59:11
- Unidentified speaker (UNKNOWN): but
- 59:11
- Unidentified speaker (speaker_0): like, you know, Oracle is 300, right? And and Corweave Core Weav is like
- 59:15
- Unidentified speaker (speaker_0): 950 or something like that. So um they're just, you know, a very both showing s signs of stress, both for very similar reasons. But Oracle has outside of this like essentially neo cloud business that Oracle has, they they're broadly diversified business. They already have like there's
- 59:35
- Unidentified speaker (speaker_0): no really incremental overhead, like the corporate overhead to support it. And they've got database, they've got applications, you know, they have traditional cloud infrastructure. Um, so they just, you know, it's and um and substantially lower leverage. So just um both showing signs of stress, but like different, you know, one is IG and the other one is is
- 59:55
- Unidentified speaker (speaker_0): high yield/slash distress.
- 59:59
- Unidentified speaker (speaker_1): Before we wrap up, um, what does the market get wrong about? CorViv right now?
- 1:00:05
- Unidentified speaker (speaker_0): Yeah, I I think the market is just very swept up in the um the technical pressure um and the and the negative sentiment around AI. Um I think that the
- 1:00:16
- Unidentified speaker (UNKNOWN): um
- 1:00:17
- Unidentified speaker (speaker_0): the fundamentals are strong and they're actually improving. And I think core Coralief we've has has done a nice job executing well uh among all the various of news. neo clouds. Um there are medium-term risks, right? You know, one, I think we talked about before, like they might have too much capacity when the
- 1:00:37
- Unidentified speaker (speaker_0): cycle turns. And then two, there's customer renewal risk as well. Um, but I I don't think those play out in 2027. I think that's more of like a 2028 to 2030 issue, right? So it's gonna take a few years. for that to happen. Um
- 1:00:53
- Unidentified speaker (UNKNOWN): and
- 1:00:54
- Unidentified speaker (speaker_0): so I I think that you know fundamentals remain uh quite strong and are actually actually improve um in the near future. And then also I would say that um I've felt and continue to feel that they'll probably get the funding that they need in the near term, just because it's mostly based on
- 1:01:14
- Unidentified speaker (speaker_0): the collateral. and the customer contracts, it's less reliant on that parent guarantee.
- 1:01:29
- Unidentified speaker (speaker_0): Thank you. Appreciate that, Joseph. Um, you know, all of my research reports are published on Creditsites.com.
- 1:01:38
- Unidentified speaker (UNKNOWN): It's a
- 1:01:38
- Unidentified speaker (speaker_0): platform.
- 1:01:39
- Unidentified speaker (speaker_0): uh for uh credit investors. I cover, you know, all of the hyperscalers as well as Core Weave, hardware OEM, software companies. Um, listeners can also find me on LinkedIn.
- 1:01:51
- Unidentified speaker (speaker_1): Jordan, thank you very much for coming on the show. If you enjoyed uh this one on Core Weave, uh please follow and subscribe wherever you get your podcasts. And see you next time.