Cycle on Jul 20, 2026

I will start by listing the open threads and active watches for this topic to understand what has been investigated so far and what specific questions or catalysts we need to track.

I will read the existing notes to see what information has already been documented and understand the exact slugs and details.

Let's check the market views for some of the key tickers in our exposure map to see the latest figures: NVDA, VST (Vistra), CRWV (CoreWeave - if public, though it might be private or newly public; let's check), NBIS (Nebius), MSFT, AMZN, META, GOOGL, ORCL, VERT (Vertiv is VRT), ETN (Eaton), CEG (Constellation Energy), APLD (Applied Digital). Let's call `market_view` for some of these. First, let's look at VST to check the debt level watch. Let's do VST and VRT and CEG. Let's start with VST.

The total debt for VST is $19.91B as of 2026-03-31, which does not cross the $22 billion threshold yet. Let's see if there are other companies we should check. Let's look at CoreWeave. Is CRWV public? Let's check with `market_view` if there is a ticker. Let's search CoreWeave or look at CRWV. Wait, let's do a search first to see what's happening with CoreWeave, Nebius, Vistra, Vertiv, etc. Let's run a web search to track recent news on CoreWeave, Nebius, Lambda Labs, and the physical/power buildout companies. Let's see if there are credit spread updates, debt issuance, rating actions, or earnings releases. Let's do a search for "CoreWeave" "debt" OR "syndicated" OR "DDTL" or "rating" or "credit spread" to see if there are any updates on CoreWeave's debt.

Ah! CoreWeave is indeed public or has filed SEC reports (Nasdaq: CRWV)! Let's check `market_view` for CRWV to see its stock details, revenue, margins, and debt. This is incredibly useful! Let's pull `market_view` for CRWV.

This is outstanding! CoreWeave (CRWV) is trading on NASDAQ, has a market cap of $39.94B, but has $35.15B in total debt and an 11.6x debt-to-EBITDA ratio (as of 2026-03-31), with negative free cash flow of $4.71B (capex of $7.70B against operating cash flow of $2.98B). The stock has fallen 36.5% over the past month to $73.21, and is down 50.8% from its 52-week high, sitting at its 10.3% 52-week position. Let's look at the news mentioned here: "CoreWeave (CRWV) shares have dropped 35% following Meta's announcement of its commercial cloud service, Meta Compute, fueling fears of cannibalization in the GPU rental market. Despite this, Rosenblatt analyst John McPeake..." And we see several insider sales from CoreWeave executives (CEO, CDO, CFO) in mid-July 2026. Let's search for "Meta Compute" to see what this is all about. This seems to be a huge development. Let's run a search for "Meta Compute" and its impact on the neoclouds (CoreWeave, Nebius, Lambda Labs). This is directly related to the "nebius-meta-megadeal-and-financing" slug! Wait, let's see if we can find more about "Meta Compute" and what it does.

Let's fetch the Semianalysis newsletter and other sources to read about this Meta Compute pivot and how it affects CoreWeave, Nebius, and the broader neocloud ecosystem. Let's start with `web_fetch` for `https://newsletter.semianalysis.com/p/meta-compute-everyone-wants-to-be`. This is a prime source on the topic!

This is highly insightful. The article clarifies what "Meta Compute" represents and what's happening with Meta's datacenter and compute procurement. Let's see: 1. Meta is contracting huge amounts of capacity (over 5GW across Cloud & Colo in the first six months of 2026 alone, plus self-build activity). 2. The headlines that Meta could become a "neocloud" (Meta Compute) led to an aggressive sell-off of neoclouds like CoreWeave and Nebius. 3. However, SemiAnalysis argues that Meta's datacenter & compute procurement will accelerate, not slow down, and that the sell-off might be an overreaction. 4. Let's look at CoreWeave and Nebius's relationship with Meta: - CoreWeave and Meta expanded a $21B agreement to scale AI workloads through 2032 (building on a $14.2B contract, for a total of $35B). - Nebius signed a long-term supply agreement with Meta of up to $27 billion. - Wait, let's search for details on these contracts. Let's run a search for `CoreWeave Meta 21 billion` and `Nebius Meta 27 billion` to understand what these deals actually are. Are they CoreWeave/Nebius renting *from* Meta, or Meta renting *from* them, or is it a supplier-customer deal? Let's check!

This is extremely interesting! Let's see: - In March 2026, Meta signed a $27 billion deal with Nebius (a Dutch neocloud, NBIS) for AI cloud capacity over 5 years. - Also, Meta signed a $21 billion expanded deal with CoreWeave (CRWV) for AI cloud computing through 2032 (bringing their total commitment to $35 billion, building on a prior $14.2 billion contract). - BUT then, in early July 2026, Bloomberg and other news outlets reported that Meta is planning a "Meta Compute" commercial cloud service, where Meta would sell its own excess AI compute capacity. - This report sent shockwaves through the neocloud market, causing Nebius (NBIS) and CoreWeave (CRWV) to plummet (e.g., Nebius losing $12 billion in market cap in a day, CoreWeave falling ~35-36% over the past month). - Why? Because if Meta, which was the massive anchor tenant paying Nebius $27B and CoreWeave $21B, starts selling its own excess compute, it might: 1. Cannibalize the GPU rental market. 2. Reduce its reliance on these neoclouds (or even cancel/renegotiate contracts). 3. Signal that Meta has overprovisioned and is now trying to unload excess capacity, raising fears of "overcapacity" in the entire AI infrastructure market. Let's do a search on "Nebius Lost $12 Billion in a Day. Meta's Cloud Plan Explains Why" or "Nebius, Coreweave, and IREN Tumble on Meta's Cloud" to understand the details of Meta's "Meta Compute" announcement and how it impacts these players. Let's run a search for `Meta's Cloud Plan Explains Why` or `Nebius Lost $12 Billion in a Day`.

Let's fetch `https://247wallst.com/investing/2026/07/17/coreweave-down-35-this-analyst-reiterated-his-250-target-even-after-meta-compute-was-announced/` or `https://dslevy.substack.com/p/parsing-meta-compute` to get a structured breakdown of the "Meta Compute" announcement, the market's reaction, and the underlying numbers. Let's do a search or fetch on `https://dslevy.substack.com/p/parsing-meta-compute` first.

This is very helpful. Let's analyze the relationship between Meta, Nebius, and CoreWeave. Let's see: Under `nebius-meta-megadeal-and-financing` slug: "Nebius and CoreWeave Face Severe Backlog and Demand Risks From Meta's "Meta Compute" Pivot". Wait, let's look at the prior notes to see what we have written. Let's do a quick search on the prior notes to see what we wrote about Nebius and CoreWeave. Wait! I can fetch the prior notes themselves or I can use `read_notes` which I already did. Let's read the full note for `nebius-meta-megadeal-and-financing` to see what is already there. Wait, I can update it with the new details. Let's fetch the note for `nebius-meta-megadeal-and-financing` using `read_notes`? No, `read_notes` only gives a list of notes with titles and previews. Wait! Is there a tool to read the full body of a note? Ah, `read_notes` doesn't have a way to fetch a single note's full body unless we call it with something, but wait, `read_notes` parameters are just `topic_id` and `limit`. It returns slug, title, preview, updated_at, and source_count. It does NOT return the full body of the notes! Ah, let me check if there's any other way to read notes. No, `read_notes` returns the preview. Let's look at the preview we got from `read_notes` for `nebius-meta-megadeal-and-financing`: "The high-growth narrative of independent AI cloud providers ("neoclouds") like Nebius Group N.V. (NBIS) and CoreWeave (CRWV) has run into a major structural roadblock. In early July 2026, reports..." Wait! Let's search the web for the exact details of the Meta-Nebius $27 billion deal and the Meta-CoreWeave $21 billion ($35 billion total) deal, and how Meta's "Meta Compute" announcement in early July 2026 creates a massive risk for them. Let's do a search on `Nebius Q1 2026 results` or `Nebius revenue` and `CoreWeave Q1 2026 results` to see their financials, and see how their stock prices have behaved. Let's first pull `market_view` for Nebius (NBIS) to see its latest market data. This is a public company too! Let's do that.

This is amazing. Let's see: - Nebius Group N.V. (NBIS) has a market cap of $45.12B, with TTM revenue of $877.9M, which represents a massive 684% year-over-year growth! - It has $9.30B in cash and $9.59B in total debt (as of 2026-03-31). - But its stock has declined 36.7% over the past month, pressured by reports that Meta Platforms is establishing an internal cloud business (Meta Compute). - We see that Nebius has a $46 billion contract lineup (including the $27 billion deal from Meta). - Let's check the details of this drop: Nebius fell 36.7% in a month to $177.71 (down 38% from its 52-week high). - Let's look at the financial performance: Q1 2026 revenue was $399.0M, compared to Q4 2025 of $227.7M and Q3 2025 of $146.1M. That is hypergrowth. But its operating margin is -32.1% (operating income of $-128.0M in Q1 2026). - Let's see if we can find more details about Nebius's financing and the Meta-Nebius contract. - Let's search for "Nebius" "Meta" "contract" OR "deal" OR "revenue" OR "financing" OR "convertible" OR "debt" to see how Nebius is financing its massive capex ($2.47B in Q1 2026 alone, with free cash flow of $-214.9M). Let's do a search.

This is incredibly rich data! Let's analyze Nebius's financing structure: - In March 2026, Nebius closed a massive $4.34 billion convertible debt funding round (convertible senior notes due in 2030 and 2032, with coupon rates of 1.25% and 2.75%). - Also, Nebius raises $775 million in secured debt financing (its first secured debt financing backed by GPU infrastructure) to accelerate global buildout. - Let's look at the reddit post: "Why Comparing Nebius and CoreWeave's Debt Misses the Point" - it mentions Nebius has 1.25/2.75% interest rates, raised through convertible notes, which is a much lower cost of debt compared to CoreWeave's high-yield senior secured delayed draw term loans (like DDTL 5.0 at SOFR + 4.5%, which is around 9.8% total interest rate!). - In Q1 2026, Nebius raised its capex guidance to $20-$25B. Wait, let's fetch the article "Nebius Raises Capex to $20-$25B: A Bold Growth Move or Risky Bet?" or check the details of this capex increase. Let's do a search or fetch for `Nebius Raises Capex to $20-$25B` to get the specifics. This is exactly what the user wanted: tracking capex guidance and revisions, and the financing structures underneath. Let's search for this.

Let's fetch `https://finance.yahoo.com/markets/stocks/articles/nebius-raises-capex-20-25b-125500935.html` to get the specifics of Nebius's capex guidance. This is a crucial data point. Let's do a `web_fetch` for it.

This article is a goldmine of specific data points! Let's list the incredible facts we just learned: 1. **Nebius (NBIS) Capex and Backlog**: - Nebius raised its 2026 CapEx guidance to $20-$25 billion (up from $16-$20 billion). - Its capacity expansion is backed by contracts with Meta (the $27 billion deal over 5 years) and Microsoft (MSFT). - It closed Q1 2026 with $9.3 billion in cash and cash equivalents, supported by operating cash flow of $2.3 billion, a $4.3 billion convertible notes offering, and a $2 billion NVIDIA investment. 2. **CoreWeave (CRWV) Capex and Backlog**: - CoreWeave's capex for 2026 is estimated to be $31 billion to $35 billion, with $6.8 billion spent in Q1 2026 (though the market view we pulled for CRWV showed capex of $7.70B for the quarter ended 2026-03-31, which is of similar magnitude). - CoreWeave has a contracted revenue backlog of roughly $100 billion. - It has been financing this via delayed draw debt facilities (DDTL) aligned with data center operationalization. 3. **Microsoft (MSFT) Capex**: - Microsoft spent $31.9 billion in capex in Q3 of fiscal 2026 (two-thirds on GPUs/CPUs). - Microsoft's fiscal fourth-quarter capex is expected to exceed $40 billion. - For calendar year 2026, Microsoft plans to invest a staggering ~$190 billion in capex (including $25 billion due to component pricing pressures). Wait! Let's verify Microsoft's current market view and capex figures using `market_view` for MSFT. Let's do that!

Let's look at Microsoft's numbers: For the quarter ended 2026-03-31, Microsoft's Capex was $30.88B (almost matching the $31.9B reported in the article). Its cash flow from operations was $46.68B, and free cash flow was $15.80B. Let's pull the market views for some of the other hyperscale buyers: Meta (META), Google (GOOGL), Amazon (AMZN), and Oracle (ORCL). Let's do that to get a comprehensive view of the hyperscalers' capital expenditures and financial positions. Let's start with META.

Let's check Meta's numbers: For the quarter ended 2026-03-31, Meta's Capex was $19.00B (reflecting massive investment, up significantly year-on-year). Operating cash flow was $32.23B, and free cash flow was $13.23B. And Meta's stock rose 13.8% over the past month to $646.01, driven in part by the bullish outlook on its "Meta Compute" commercial initiative and its ability to monetize its massive compute capacity, while the neoclouds (CoreWeave and Nebius) collapsed. This is a classic divergence between the buyer/enabler (Meta) and the leveraged middlemen (neoclouds)! Let's pull the market views for GOOGL, AMZN, and ORCL as well. Let's start with GOOGL.

Let's check Google's numbers: For the quarter ended 2026-03-31, Google's Capex was $35.67B. Operating cash flow was $45.79B, and free cash flow was $10.12B. This is massive! Google's capex is even larger than Microsoft's for that quarter ($30.88B). Let's pull the market view for AMZN.

Let's check Amazon's numbers: For the quarter ended 2026-03-31, Amazon's Capex was $44.20B! Operating cash flow was $26.03B, and free cash flow was $-18.17B. This is a massive negative free cash flow due to extreme capex! Let's look at ORCL. Let's pull the market view for ORCL.

Wait, look at Oracle! Oracle's stock has declined 30.9% over the past month and is trading 61.1% below its 52-week high, sitting at its 1.1% 52-week position! And look at its debt: $156.19B in total debt against $31.29B in cash, reflecting a Debt/EBITDA of 5.1x. And its capex for the latest quarter (ended 2026-05-31) was $16.49B, resulting in a negative free cash flow of $-1.87B. Wait, why did Oracle's stock crash? SemiAnalysis mentioned: "SpaceX’s pricing on the Google deal is even higher than what we observe on the on-demand or short-term rental market... This effectively leaves only two companies that can truly capitalize on this new market: Oracle and Meta. For the former, we see this as a massive blow. Enter Meta. Given their laser-focus on building Superintelligence, they probably didn’t put enough thought into the Cloud option... As such, we expect Meta to announce such a deal soon." Wait, let's search for "Oracle" "Meta" "Meta Compute" OR "competition" to understand why Oracle's stock is crashing. Is it due to Meta Compute or is there another reason? Let's search for `Oracle stock crash July 2026` or `Oracle "Meta Compute"`.

Wow, this is a spectacular find! Let's look at the Oracle situation: - Oracle's stock has plunged over 60% in 2026 (currently trading at $126.41, down 61.1% from its 52-week high, and down 30.9% in the last month). - This is driven by massive, debt-funded AI datacenter investments. Oracle's total debt has ballooned to $156.19 billion (as of May 31, 2026), representing a Debt/EBITDA of 5.1x! - In its recent earnings report, Oracle guided for fiscal year 2026 (FY26) CapEx of approximately $50 billion, which will be funded by significant new debt issuance. - Investors have dumped Oracle bonds and stock over worries that this massive, debt-funded AI infrastructure bet might fail, especially as "Meta Compute" and other developments threaten the pricing and demand for cloud compute. - This is a massive, second-order financial exposure! If the AI buildout slows, Oracle is sitting on $156B in debt and $50B in annual capex that it may not be able to monetize profitably. Let's look at the power/physical buildout players. The named entities are: Vertiv (VRT), Eaton (ETN), Constellation (CEG), Vistra (VST). We already pulled VST (Vistra). Let's pull VRT, ETN, and CEG. Let's start with VRT (Vertiv). Let's call `market_view` for VRT.

Vertiv (VRT) has a market cap of $111.22B, TTM revenue of $10.84B (up 30.1% YoY), and has $3.26B in debt against $2.15B in cash. This is a very clean balance sheet compared to Oracle or the neoclouds (Debt/EBITDA of 1.4x). Free cash flow in Q1 2026 was $654.2M (capex was $112.6M, which is quite modest). Let's check CEG (Constellation Energy). Let's call `market_view` for CEG.

Let's look at Constellation Energy: - CEG is trading at $252.39, down 37.3% from its 52-week high, sitting at its 9.6% 52-week position. - Total debt is $22.47 billion, with Debt/EBITDA of 2.8x. - Quarterly capex was $1.27 billion, leading to negative free cash flow of $850.0 million for the quarter ended 2026-03-31. - Why did CEG decline so much from its high? Let's check: - "Constellation Energy secured approximately $2.24 billion in annual gross capacity payments from the PJM auction for June 2028, but nuclear facilities accounting for 83% of this capacity may reduce the net earnings benefit..." - And: "Constellation Energy (NASDAQ: CEG) Stock Price Trades At Deep Discount As New York Moratorium Tightens Power Outlook" - Wait, let's look at Eaton (ETN) as well. Let's pull the market view for ETN.

Eaton is trading at $399.99, down 8.2% from its 52-week high, which is quite resilient compared to Vertiv (-23% from high), Constellation (-37.3%), and Vistra (-28.4%). Wait! Let's check Applied Digital (APLD), which is mentioned in the prior findings as having CoreWeave exposure and high-yield leverage. Let's pull the market view for APLD.

This is incredible! Let's see: - Applied Digital (APLD) has a market cap of $7.37B and TTM revenue of $319.3M, but it has $2.83B in total debt (as of February 28, 2026). Its Debt/EBITDA is a staggering 203.3x! - It had a quarterly net loss of $99.3 million on $126.6 million in revenue, and negative free cash flow of $720.2 million (with capex of $775.2 million). - The stock has plunged 43.4% over the past month to $25.79, and is down 48.1% from its 52-week high, with RSI(14) at 14 (extremely oversold!). - The news mentions: "Applied Digital Plunges As Nvidia Walks Away, Dilution Fears" - "Applied Digital (APLD) shares are falling due to a broad sector selloff in semiconductor and AI stocks, intensified by fears of share dilution from recent financing moves. Adding to the negative sentiment, Nvidia has reportedly walked away..." - Wait, let's search for "Applied Digital" "Nvidia" "walks away" OR "withdrew" OR "cancelled" to see what happened. This is a massive piece of negative news and directly relates to the circular financing/vendor-financing thesis! Let's do a search.

Let's fetch `https://finance.yahoo.com/news/investors-reacting-applied-digital-apld-221109704.html` to find out what is in that article "How Investors Are Reacting To Applied Digital (APLD) Massive Debt Deal, Nvidia Exit and CoreWeave Risks". This is exactly the synthesis we need! Let's do a `web_fetch` for it.

This is extremely interesting! Let's see: - Applied Digital (APLD) announced a massive **$2.15 billion private offering of 6.750% senior secured notes due 2031** to fund its Polaris Forge 2 AI data center campus. - Nvidia exited its equity stake (which it acquired in September 2024 as part of a $160M round; by early 2026, Nvidia had sold its entire stake). - Applied Digital has high customer concentration in **CoreWeave**, which is a key customer but is itself facing legal and financial pressures. - Wait, let's search for "Applied Digital Polaris Forge 2" or "Polaris Forge 2" to see who the "investment grade hyperscaler" is. Is it Microsoft or Meta, or is it someone else? Let's run a search.

Ah! Look at this: - Applied Digital signed a $5 billion AI Factory lease with a U.S.-based investment-grade hyperscaler at its Polaris Forge 2 North Dakota campus (200 MW). - Also, it signed a massive 430 MW data center lease with an unnamed hyperscaler at its Polaris Forge 3 campus (which is a 15-year lease valued at $7.5 billion). - But at its Polaris Forge 1 campus, the primary tenant is CoreWeave. - Let's look at the financing: To fund the Polaris Forge 2 buildout, Applied Digital issued $2.15 billion in 6.750% senior secured notes due 2031. - And Nvidia, which had backed Applied Digital with equity, sold its entire stake. - Let's check the credit risk and leverage: Applied Digital has $2.83 billion in total debt against $1.73 billion in cash, and is running massive quarterly net losses (e.g. $99.3 million loss in the latest quarter). Its Debt/EBITDA is 203.3x! - This massive leverage means that Applied Digital is extremely sensitive to project execution, construction delays, or any slowdown/renegotiation by its hyperscaler or neocloud tenants (CoreWeave at Polaris Forge 1). Wait, let's look at the "neocloud" and "vendor-financing loop" aspects. We have existing notes: 1. `nvidia-circular-vendor-financing-loop` / `nvidia-circular-financing-loop`: - "The explosive growth of the artificial intelligence infrastructure buildout relies on a highly fragile, circular capital-recycling loop between NVIDIA (NVDA), independent "neocloud" providers, and..." 2. `coreweave-liquidity-crisis-and-syndicated-debt`: - "CoreWeave's Extreme Leverage, Widening Losses, and the Bifurcated Cost of GPU Debt..." 3. `nebius-meta-megadeal-and-financing`: - "Nebius and CoreWeave Face Severe Backlog and Demand Risks From Meta's "Meta Compute" Pivot..." 4. `applied-digital-coreweave-exposure-and-debt`: - "Applied Digital Exposes Physical Buildout to CoreWeave Credit Risk and High-Yield Leverage..." 5. `utility-leverage-and-power-buildout-exposure`: - "Power Buildout and Utility Exposure to AI Capital Spending Slowdown..." Let's check if there are any other specific updates on the CoreWeave DDTL 5.0 facility secondary market trading prices or credit spreads. The watch is "CoreWeave DDTL 5.0 Secondary Market Pricing Track" with trigger "CRWV". Let's see if there is any news on CoreWeave DDTL 5.0 facility trading prices or credit spreads. Let's do a search for `CoreWeave DDTL 5.0` or `CoreWeave debt secondary market` or `CoreWeave credit spread`.

Let's fetch `https://davefriedman.substack.com/p/coreweave-issued-yet-more-debt` or `https://davefriedman.substack.com/p/coreweave-piles-on-more-debt` to get the breakdown of CoreWeave's debt. Let's do a search or fetch for `https://davefriedman.substack.com/p/coreweave-issued-yet-more-debt`. Wait, let's search for the URL. Let's fetch it.

This is a spectacular set of findings! The Dave Friedman article is a masterclass in structured credit analysis for the AI infrastructure space. Let's outline the core insights: 1. **CoreWeave's Debt Stratification & Bifurcation (DDTL 4.0 vs DDTL 5.0)**: - In March 2026, CoreWeave closed an $8.5B delayed draw term loan (DDTL 4.0) through *CoreWeave Compute Acquisition Co. VIII LLC*. It was rated A3 by Moody's / A(low) by DBRS, and priced at SOFR + 2.25% (floating) / ~5.9% (fixed). It is backed by a single anchor customer of investment-grade (IG) credit (Aa2), likely Microsoft. It is non-recourse to the parent. - In May 2026, CoreWeave closed a $3.1B delayed draw term loan (DDTL 5.0) through *CoreWeave Financing DDTL V LLC*. It was rated Ba2 by Moody's / BB+ by Fitch, and priced at SOFR + 4.50% (floating). It is backed by two large, non-investment grade customers (undisclosed, likely unrated frontier AI labs). It is unconditionally guaranteed by CoreWeave (parent) and key subsidiaries, making it recourse. - Despite having *more* credit protection (parent guarantee), DDTL 5.0 prices 225 basis points higher in spread and is rated 4 notches lower than DDTL 4.0. - This proves the "Inversion": GPU debt is *not* asset-backed in practice. It is contract-backed, with GPUs acting merely as a recovery floor. The credit is determined by *who* is renting the GPUs, not the GPUs themselves. - This creates a massive stratification: Neoclouds with IG customers (like Microsoft) get cheap capital (SOFR + 2.25%); those with unrated AI labs are structurally pinned to high-yield pricing (SOFR + 4.50% or worse), crushing their unit economics and margins. - DDTL 5.0 is the first publicly syndicated HPC-backed debt facility, designed to enable secondary market trading. Wait, let's look at the active watch: `CoreWeave DDTL 5.0 Secondary Market Pricing Track` (ID: `019f1273-ce52-7485-a89d-8b5383b6d175`) The trigger says: `Monitor secondary market trading prices and credit spreads of CoreWeave's $3.1 billion syndicated DDTL 5.0 facility. A drop in secondary trading prices below par or a widening of the spread over SOFR beyond 4.50% would signal a demand softening or credit deterioration in the neocloud sector.` Wait! In July 2026, we have the Meta Compute announcement which crashed neocloud stocks (CoreWeave down 36.5%, Nebius down 36.7%). This is a massive market shock! Let's see if we have evidence that the secondary market pricing of DDTL 5.0 has dropped or if spreads have widened. Wait, did we find any specific secondary market trading price or spread widening? The Dave Friedman article was written on May 28, 2026. In July 2026, we have the "Meta Compute" announcement. The Meta Compute announcement occurred around July 1, 2026, which sent CoreWeave and Nebius down ~35-36%. Let's see: we don't have direct secondary market price quotes (like 95 cents on the dollar), but we know that neocloud equity prices have crashed, and the threat of Meta Compute (Meta selling its own excess compute) directly threatens the pricing and demand of the GPU rental market, which would widen spreads and push secondary market prices of DDTL 5.0 below par due to credit risk concerns for the underlying non-IG customers. Wait, let's record an event or update the watch if we have concrete data. If we don't have the exact bond price, we can keep the watch open as `resolves: false` and note the Meta Compute shock as a major ongoing credit catalyst. Wait, let's also check the Vistra Corp Debt Watch: `Vistra Corp Debt and Leverage Track` (ID: `019eca5a-01d4-7c6d-ab25-99092c634700`) The total debt of VST is $19.91B (as of 2026-03-31), which is below the $22B threshold. We will keep this watch open and record that VST's debt remains at $19.91B as of the latest filing, but it has expanded its accounts receivable facility to $1.25B in July 2026 to enhance liquidity. Let's organize our findings into separate high-quality notes! We have several distinct themes that we should write or update: 1. **CoreWeave's Debt Stratification and the "Inversion" of GPU Debt (DDTL 4.0 vs DDTL 5.0)**: - This directly updates/extends: `coreweave-liquidity-crisis-and-syndicated-debt` (CoreWeave's Extreme Leverage, Widening Losses, and the Bifurcated Cost of GPU Debt). - Let's reuse the slug `coreweave-liquidity-crisis-and-syndicated-debt` to update this note. It is a perfect fit! - We will detail the structure of DDTL 4.0 ($8.5B, SOFR + 2.25%, A3/A-low, non-recourse, single IG customer Aa2) and DDTL 5.0 ($3.1B, SOFR + 4.50%, Ba2/BB+, recourse parent-guaranteed, two non-IG customers). - Explain the "Inversion": GPU-backed debt is actually contract-backed debt, where the off-taker's credit determines the rating and spread, not the physical GPU collateral. This creates a severe stratification problem for neoclouds. - Mention CoreWeave's current market view (as of July 20, 2026): Market Cap $39.94B, Total Debt $35.15B, Debt/EBITDA 11.6x, TTM revenue of $6.23B, net margin of -25.6%, and negative free cash flow of $4.71B (Q1 2026 capex of $7.70B vs operating cash flow of $2.98B). - Also mention that CoreWeave's stock (CRWV) has crashed 36.5% over the past month to $73.21, down 50.8% from its 52-week high, following Meta's "Meta Compute" announcement. 2. **Nebius and CoreWeave Face Severe Backlog and Demand Risks From Meta's "Meta Compute" Pivot**: - This directly updates/extends: `nebius-meta-megadeal-and-financing` (Nebius and CoreWeave Face Severe Backlog and Demand Risks From Meta's "Meta Compute" Pivot). - Let's reuse the slug `nebius-meta-megadeal-and-financing` to update this note. - Detail the "Meta Compute" announcement in early July 2026, where Meta plans to rent out its excess compute capacity (monetizing its massive AI infrastructure as productive capital). - Explain why this sent shockwaves through the neocloud market: Meta is the massive anchor tenant that recently signed a $27 billion deal with Nebius (NBIS) over 5 years, and a $21 billion expanded deal with CoreWeave (CRWV) through 2032 (totaling $35 billion). - If Meta begins selling its own excess compute, it directly threatens to cannibalize the GPU rental market, reduce its reliance on neoclouds, and signal overprovisioning. - Consequently, neocloud stocks tumbled ~35-37% in July 2026 (CoreWeave down 36.5% to $73.21, Nebius down 36.7% to $177.71). - Contrast this with Meta's own stock (META), which rose 13.8% over the past month to $646.01, showing a stark divergence between the buyer/enabler (Meta) and the leveraged middlemen (neoclouds). - Detail Nebius's financials: Market Cap $45.12B, TTM revenue $877.9M (+684% YoY), $9.30B in cash against $9.59B in total debt. Operating margin of -32.1% and Q1 2026 capex of $2.47B. - Detail Nebius's financing structure: closed a massive $4.34B convertible senior notes offering in March 2026 (due 2030/2032 with low coupon rates of 1.25% and 2.75%), plus $775 million in its first secured debt financing. This lower cost of debt contrasts sharply with CoreWeave's high-yield DDTL 5.0 (SOFR + 4.50%). 3. **Applied Digital Exposes Physical Buildout to CoreWeave Credit Risk and High-Yield Leverage**: - This directly updates/extends: `applied-digital-coreweave-exposure-and-debt` (Applied Digital Exposes Physical Buildout to CoreWeave Credit Risk and High-Yield Leverage). - Let's reuse the slug `applied-digital-coreweave-exposure-and-debt` to update this note. - Detail Applied Digital's (APLD) recent developments: - Plunged 43.4% over the past month to $25.79, down 48.1% from its 52-week high, with RSI(14) at 14. - Pivot from crypto mining to 1 GW of AI infrastructure. - Announced a massive **$2.15 billion private offering of 6.750% senior secured notes due 2031** to fund its Polaris Forge 2 AI data center campus. - It has signed a $5 billion lease with a U.S. investment-grade hyperscaler at Polaris Forge 2, and a $7.5 billion lease at Polaris Forge 3. - However, its Polaris Forge 1 campus is leased to CoreWeave, creating massive customer concentration risk. - Nvidia, which had backed Applied Digital with a $160 million equity round in September 2024, has completely exited its equity stake as of early 2026, sparking dilution and funding fears. - Balance sheet is highly fragile: $2.83 billion in total debt against $1.73 billion in cash, with a Debt/EBITDA of 203.3x! It ran a quarterly net loss of $99.3 million on $126.6 million in revenue, and negative free cash flow of $720.2 million. 4. **Oracle's High-Yield Leverage and Soaring Capex Fail to Convince Bond and Stock Markets**: - Let's mint a new slug: `oracle-debt-and-capex-debacle`. - Detail the massive crash of Oracle (ORCL): - Oracle's stock has plunged over 60% in 2026 (currently trading at $126.41, down 61.1% from its 52-week high, and down 30.9% in the last month). - Driven by a massive, debt-funded AI datacenter buildout. Total debt has ballooned to $156.19 billion (as of May 31, 2026), representing a Debt/EBITDA of 5.1x! - Oracle guided for fiscal year 2026 (FY26) CapEx of approximately $50 billion, funded by significant new debt issuance. - This massive leverage has caused investors to dump Oracle's bonds and stock, as concerns mount over whether this debt-fueled bet can be monetized, especially with Meta Compute entering the market and threatening GPU rental pricing. - In the latest quarter (ended May 31, 2026), Oracle spent $16.49 billion in capex against operating cash flow of $14.62 billion, resulting in a negative free cash flow of $-1.87 billion. 5. **Power Buildout and Utility Exposure to AI Capital Spending Slowdown**: - This directly updates/extends: `utility-leverage-and-power-buildout-exposure` (Power Buildout and Utility Exposure to AI Capital Spending Slowdown). - Let's reuse the slug `utility-leverage-and-power-buildout-exposure` to update this note. - Detail the power players' performance: - **Vistra Corp (VST)**: Market cap $52.41B, total debt $19.91B, Debt/EBITDA 2.9x. Gross annual run-rate of $1.296B from the PJM capacity auction. Expands accounts receivable securitization facility to $1.25B to enhance liquidity. Stock is down 28.4% from its 52-week high, showing some softness. - **Constellation Energy (CEG)**: Market cap $90.13B, total debt $22.47B, Debt/EBITDA 2.8x. Stock has declined 37.3% from its 52-week high to $252.39, despite TTM revenue of $29.87B (+63.8% YoY). Spent $1.27B in Q1 2026 capex, leading to negative free cash flow of $850.0 million. PJM capacity auction secured $2.24B but nuclear facilities may reduce net benefit, and New York moratorium tightens power outlook. - **Eaton (ETN)**: Market cap $155.32B, total debt $21.83B, Debt/EBITDA 3.4x. Resilient stock, down only 8.2% from 52-week high. - **Vertiv (VRT)**: Market cap $111.22B, total debt $3.26B, Debt/EBITDA 1.4x. Down 23.0% from its 52-week high to $289.56. This is an incredibly comprehensive, detailed, and highly analytical mapping of the second-order exposures of the AI buildout! Let's write these notes one by one. Let's start by updating `coreweave-liquidity-crisis-and-syndicated-debt`. Let's construct the markdown and call `write_note`. Wait! Let's check the rules for `write_note`: - Carry 1-2 representative verbatim quotes, copied exactly, each kept with the link it came from. - Name and link the actual thing. - No YAML frontmatter. First heading is `# <Title>`. - Pass the slug in the parameter. Let's write the first note: `coreweave-liquidity-crisis-and-syndicated-debt`. Let's draft it. Quote from Dave Friedman Substack: > "This is the inversion: GPU-backed debt is not actually an asset-backed product in the way market participants describe it. It’s a contract-backed product with GPU collateral providing a recovery floor. The rating agencies are pricing the off-taker’s ability to pay through the contract term. The asset is the backstop, not the engine." And: > "The compounding effects are punitive. Higher financing costs mean thinner spread to GPU lease revenue, which means less margin to absorb obsolescence risk, which means less room to underprice incumbents to win the very investment-grade anchor tenants that would let them break onto the cheaper track." Let's call `write_note` for `coreweave-liquidity-crisis-and-syndicated-debt`.

I will now write the updated note for CoreWeave's leverage and debt structure, incorporating the groundbreaking insights from its recent DDTL 4.0 and DDTL 5.0 syndicated debt facilities and its latest public market performance.

Now I will update the second note, `nebius-meta-megadeal-and-financing`, to document the "Meta Compute" announcement, its impact on the neoclouds (Nebius and CoreWeave), and the financial profile of Nebius (NBIS), including its low-cost convertible debt financing versus CoreWeave's high-yield DDTLs. Let's draft and write this note.

Now I will write the updated note for Applied Digital, `applied-digital-coreweave-exposure-and-debt`, incorporating its latest Q1 2026 financials, its massive $2.15 billion senior secured notes due 2031, its customer concentration in CoreWeave, and the complete exit of Nvidia from its equity stake. Let's draft and write this note.

Now I will write a brand new note on Oracle's debt and capex debacle, `oracle-debt-and-capex-debacle`, detailing its massive stock drop, its ballooning debt of $156.19 billion, and its aggressive $50 billion FY26 capex guidance. Let's draft and write this note.

Now I will update the fifth note on the power buildout, `utility-leverage-and-power-buildout-exposure`, detailing the latest performance of Vistra (VST), Constellation Energy (CEG), Vertiv (VRT), and Eaton (ETN). Let's draft and write this note.

Perfect! Now, let's review our active watches and threads to see if we should record any events or resolve any threads. The threads/watches we retrieved are: 1. `CoreWeave DDTL 5.0 Secondary Market Pricing Track` (ID: `019f1273-ce52-7485-a89d-8b5383b6d175`): - This is an ongoing watch tracking secondary market trading prices and credit spreads of CoreWeave's $3.1 billion syndicated DDTL 5.0 facility. - We don't have a specific secondary price quote (e.g. 98 cents) or a precise spread print from this cycle, so we shouldn't resolve it or record a one-shot event. It remains an ongoing, highly relevant watch. 2. `Vistra Corp Debt and Leverage Track` (ID: `019eca5a-01d4-7c6d-ab25-99092c634700`): - This ongoing watch tracks Vistra's total debt levels, looking for it to cross $22 billion. - VST's total debt is currently $19.91 billion as of the latest Q1 2026 balance sheet, so it has not crossed $22 billion yet. This watch remains open. Let's check if there are any other threads that we can resolve. Let's list the threads again to check if there are any other open threads. No, we only had those two watches in the list. Wait! Let's register a new watch for Oracle's debt and capex or for the Applied Digital fiscal Q4 earnings on July 27, 2026! Yes! Applied Digital is hosting its conference call on Monday, July 27, 2026, to discuss its fiscal Q4 and full year 2026 results. This is a perfect dated catalyst! Let's register a watch for it. Let's call `record_watch`. - `title`: Applied Digital Fiscal Q4 and Full Year 2026 Earnings Release - `cardinality`: `one_shot` - `watch_type`: `dated_catalyst` - `fire_date`: `2026-07-27T21:00:00Z` - `entity`: `APLD` - `description`: Applied Digital will release its fiscal Q4 and full year 2026 results after the market close on July 27, 2026. This will provide updated numbers on its cash burn, total debt, and the status of its Polaris Forge campuses. Let's also register a watch for Oracle's total debt levels crossing $170 billion or its capex trajectory. Or let's watch for any credit rating actions on CoreWeave or Nebius. Let's register the Applied Digital watch first.

Excellent! Are there any other watches we should register? Let's think. We have documented a massive structural shift where Meta is launching "Meta Compute" to rent out excess compute. We can register a watch for the official launch of Meta Compute or any announcements about Meta's first commercial cloud customers. Let's do that! - `title`: Meta Platforms Official Meta Compute Commercial Cloud Launch - `cardinality`: `one_shot` - `watch_type`: `tripwire` - `entity`: `META` - `description`: Monitor the official launch, pricing, and first commercial cloud customers of Meta's "Meta Compute" initiative. Any contract wins or pricing undercuts will directly impact the neoclouds (CoreWeave, Nebius). Let's register this watch.

Now, let's call `finish_cycle` to complete this research cycle. I have updated four notes with highly detailed, primary-source-backed evidence and registered two high-value watches. This is a highly successful cycle!

I have successfully completed the research cycle. Here is a summary of the findings and updates made to the exposure map: ### 1. **CoreWeave's Leverage and the "Inversion" of GPU Debt** * **Slug**: `coreweave-liquidity-crisis-and-syndicated-debt` (Updated) * **Key Findings**: * CoreWeave's capital structure is pushed to its limits, carrying **$35.15 billion in total debt** against $3.02 billion in cash (as of March 31, 2026), representing a **Debt-to-EBITDA of 11.6x**. * Free cash flow was **-$4.71 billion** in Q1 2026, driven by **$7.70 billion in capex** against $2.98 billion in operating cash flow. * Analysis of its **DDTL 4.0 ($8.5B)** and **DDTL 5.0 ($3.1B)** facilities reveals a profound structural "inversion": GPU-backed debt is actually contract-backed debt. DDTL 4.0 (backed by single investment-grade tenant, likely Microsoft) priced at **SOFR + 2.25%**, whereas DDTL 5.0 (backed by two unrated frontier AI labs) priced at **SOFR + 4.50%** despite carrying parent guarantees. The off-taker's credit profile, not the physical GPU, drives the rating and spread. * CoreWeave shares (CRWV) crashed **36.5% over the past month to $73.21** (down 50.8% from its 52-week high) following Meta's commercial cloud announcement, prompting pre-arranged stock sales by CoreWeave's CEO, CFO, and CDO. ### 2. **Meta's "Meta Compute" Pivot and Neocloud Backlog Risks** * **Slug**: `nebius-meta-megadeal-and-financing` (Updated) * **Key Findings**: * In early July 2026, reports emerged of Meta Platforms' (META) commercial cloud initiative, **"Meta Compute"**, to rent out excess AI compute capacity. * This is a severe threat to neocloud backlogs. Meta is the massive anchor tenant that recently signed a **$27 billion deal with Nebius (NBIS)** and a **$21 billion expanded deal with CoreWeave (CRWV)**. * If Meta commercializes its compute, it directly threatens to cannibalize the GPU rental market and reduce its reliance on third-party cloud providers. * Consequently, neocloud shares tumbled over the past month (Nebius down 36.7% to $177.71; CoreWeave down 36.5% to $73.21), while Meta's stock rose **13.8% to $646.01**, showing a stark divergence. * Nebius TTM revenue reached **$877.9 million (+684% YoY)**, with cash of $9.30B against $9.59B in total debt. Its 2026 capex guidance was raised to **$20-$25 billion**. However, Nebius maintains a lower cost of debt than CoreWeave, having closed a **$4.34 billion convertible senior notes offering** in March 2026 at low coupon rates of 1.25% and 2.75%. ### 3. **Applied Digital’s Physical Buildout and CoreWeave Exposure** * **Slug**: `applied-digital-coreweave-exposure-and-debt` (Updated) * **Key Findings**: * Applied Digital (APLD) has pivoted hard to AI data centers, reaching a 1 GW milestone. However, its balance sheet is exceptionally fragile, carrying **$2.83 billion in total debt against $1.73 billion in cash** (Debt-to-EBITDA of **203.3x**) and running a quarterly net loss of **$99.3 million** on free cash flow of **-$720.2 million**. * To fund its 200 MW Polaris Forge 2 campus, it issued **$2.15 billion in 6.750% senior secured notes due 2031**. * While Polaris Forge 2 and 3 are leased to investment-grade hyperscalers, its operational Polaris Forge 1 campus is anchored by CoreWeave, creating massive customer concentration risk. * Nvidia (NVDA) has **completely exited its equity position** in APLD as of early 2026, sparking dilution and funding fears that sent APLD stock down **43.4% over the past month to $25.79** (RSI at 14). ### 4. **Oracle's High-Yield Leverage and Capex Debacle** * **Slug**: `oracle-debt-and-capex-debacle` (New Finding) * **Key Findings**: * Oracle Corporation (ORCL) has emerged as one of the most heavily levered physical buildout players. Its total debt has ballooned to **$156.19 billion** (as of May 31, 2026), representing a Debt-to-EBITDA of **5.1x**. * Oracle guided for an aggressive **$50 billion in FY26 capex**, funded by significant new debt issuance. * In the latest quarter, Oracle spent **$16.49 billion in capex** against operating cash flows of $14.62 billion, leading to a negative free cash flow of **-$1.87 billion**. * The stock has plunged **30.9% over the past month** and is trading **61.1% below its 52-week high** as investors dump Oracle bonds and shares, fearing that its debt-fueled OCI buildout cannot be monetized profitably in a market threatened by Meta Compute. ### 5. **Power and Physical Buildout Exposure** * **Slug**: `utility-leverage-and-power-buildout-exposure` (Updated) * **Key Findings**: * **Constellation Energy (CEG)** shares have declined **37.3% from their 52-week high** to trade at $252.39, burdened by **$22.47 billion in total debt** (Debt-to-EBITDA of 2.8x) and negative free cash flow of **-$850.0 million** (capex of $1.27B). Regulatory challenges in New York and nuclear-capacity capacity payment limitations from the PJM auction have weighed on its valuation. * **Vistra Corp (VST)** shares declined **28.4% from their 52-week high** to $155.44. VST carries **$19.91 billion in total debt** (Debt-to-EBITDA of 2.9x) but secured a gross run-rate of **$1.296 billion** from the PJM capacity auction. VST expanded its accounts receivable facility to **$1.25 billion** to enhance liquidity. * Equipment providers remain fundamentally stronger: **Vertiv (VRT)** has a clean balance sheet (Debt-to-EBITDA 1.4x, $654.2M positive quarterly free cash flow) but fell **23.0% from its high** to $289.56; **Eaton (ETN)** remains highly resilient, down only **8.2% from its high** to $399.99. ### New Watches Registered 1. **Applied Digital (APLD) Fiscal Q4 Earnings Release (Dated Catalyst - July 27, 2026)**: Track cash burn, total debt levels, and Polaris Forge campus buildout updates. 2. **Meta Platforms (META) Meta Compute Launch (Tripwire)**: Monitor official commercial cloud launch, pricing, and first customer wins to assess direct cannibalization of the neocloud market.