Cycle on Jun 8, 2026

I am starting this research cycle by checking the open threads and active watches for this topic to see what questions are currently prioritized or what catalysts we are monitoring.

I will call `market_view` for NVDA to retrieve its current financial standing, valuation multiples, recent performance, and key market data.

I will perform a web search to check the latest news on Nvidia's AI capex story, hyperscaler capex, and what the current landscape looks like in June 2026. This will help determine if there are any signs of slowdown, any updates on Blackwell, or if the capex story remains fully intact.

The search results indicate that AI spending is expected to top $1 trillion in two years (by 2027), and hyperscaler capex forecasts for 2026 are widely cited at over $600 billion or $700 billion. Let's search for the CNBC article from May 21, 2026 ("AI spending expected to top $1 trillion in 2 years. Why that estimate...") and fetch it. This is extremely recent (just a few weeks ago in our current time of June 2026) and directly relevant to Nvidia's AI capex story. Let's also look for recent articles on Nvidia's Vera Rubin Chip announcement from June 2026 (Jensen Huang made an announcement on June 7, 2026, about Vera chip using SK Hynix memory). Let's do a search on that as well. First, let's fetch the CNBC article.

This is incredibly valuable. It details Nvidia's Q1 FY2027 (ended April 2026, reported May 20, 2026) earnings call where: - Jensen Huang stated AI capex is at $1 trillion and growing toward the $3 to $4 trillion mark. - CFO Colette Kress explicitly stated: "With analysts now forecasting hyperscale capex to exceed $1 trillion in 2027 and agentic AI beginning to proliferate [across] all industries, AI infrastructure spending is on track to reach $3 to $4 trillion annually by the end of this decade." - Big Tech cloud revenues are growing robustly: Alphabet cloud up 63%, AWS up 28%, Microsoft up 40%. - Skeptics (like JPMorgan and economists) raise questions on whether the productivity gains and revenue will justify this massive spend, but the hyperscalers are forging ahead. Let's do a search on Jensen Huang's recent comments about the "Vera" or "Vera Rubin" platform and SK Hynix memory from June 2026. Let's see what the latest news is on that. This represents the next-generation architecture after Blackwell. Let's do a web search.

This is incredibly interesting! We have a major event happening right now (June 2026): 1. Nvidia GTC Taipei 2026 (or Computex 2026 / GTC 2026 Taipei) just occurred in early June 2026. 2. Jensen Huang announced the "Vera Rubin" AI platform, which is the successor to Blackwell. 3. On June 7, 2026, NVIDIA and SK Hynix announced a multiyear technology partnership. 4. "Vera Rubin systems are in full production as of June 2026 and are scheduled to begin shipping to major cloud providers — including AWS, Google..." (from TechTimes article on June 2, 2026). Wait, let's fetch these articles to get precise details. Let's first fetch the TechTimes article: `https://www.techtimes.com/articles/317539/20260602/nvidia-vera-rubin-enters-full-production-samsung-sk-hynix-micron-named-hbm4-suppliers.htm` And the Bloomberg article: `https://www.bloomberg.com/news/articles/2026-06-07/nvidia-s-ceo-says-new-vera-chip-will-use-sk-hynix-s-memory-chips` And the official NVIDIA news release: `https://nvidianews.nvidia.com/news/sk-hynix-ai-factory` Let's fetch `https://www.techtimes.com/articles/317539/20260602/nvidia-vera-rubin-enters-full-production-samsung-sk-hynix-micron-named-hbm4-suppliers.htm` first to check the production status of Vera Rubin.

This is a massive development. Let's analyze what we have here. 1. The Vera Rubin platform is in *full production* as of June 2026. This is the successor to Blackwell. Wait, let's look at the prior findings. One of the watches was: "Track when Nvidia starts its first official production shipments of the Vera Rubin GPU platform, which is currently sampling and expected to ship in Q3/Q4 2026." According to the TechTimes article: - "Jensen Huang confirmed at his GTC Taipei 2026 keynote on June 1 that Nvidia's Vera Rubin AI platform is now in full production... First customer shipments are scheduled to begin this summer, with broad availability targeting the second half of 2026. Among the first cloud providers confirmed for early deployment are AWS, Google Cloud, Microsoft Azure, and Oracle." Wait, does this resolve the watch `019e9c33-1526-73d2-9154-7e661457a981` ("Nvidia Vera Rubin First Production Shipments")? The watch description says "Track when Nvidia starts its first official production shipments of the Vera Rubin GPU platform... Fires when Nvidia officially announces first shipments of the Vera Rubin platform, confirming the transition timeline from Blackwell." The article says: "First customer shipments are scheduled to begin this summer, with broad availability targeting the second half of 2026." So they haven't officially shipped yet (they start "this summer"). Wait, the CPU side (Vera CPU) is already shipping! "Nvidia hand-delivered the first systems to Anthropic, OpenAI, SpaceX, and Oracle in mid-May." But the GPU platform (Vera Rubin) is "scheduled to begin shipping to major cloud providers... starting this summer". So the watch hasn't fired yet, but it is extremely close. We should keep the watch open as it's scheduled for Q3/Q4 2026. Or is it? Let's check if there are other details. Let's fetch the Bloomberg article from June 7, 2026 to see if there are any additional details about SK Hynix and Vera Rubin. Let's also look for the official news from Nvidia: `https://nvidianews.nvidia.com/news/sk-hynix-ai-factory` or search for it. Let's search for the Bloomberg article.

Let's fetch the MSN article (which mirrors the Bloomberg article): `https://www.msn.com/en-us/money/companies/nvidia-s-ceo-says-new-vera-chip-will-use-sk-hynix-s-memory-chips/ar-AA2523zW?ocid=finance-verthp-feeds` to get the core details of the SK Hynix partnership and the Vera chip.

Ah, the MSN page is mostly empty or sponsored content. Let's try to search for the text of the Bloomberg article or the official Nvidia announcement since we have the URL `https://nvidianews.nvidia.com/news/sk-hynix-ai-factory` and `https://www.crnasia.com/news/2026/components-and-peripherals/nvidia-and-sk-hynix-announce-multiyear-technology-partnershi`. Let's fetch `https://www.crnasia.com/news/2026/components-and-peripherals/nvidia-and-sk-hynix-announce-multiyear-technology-partnershi`.

This is excellent. Let's analyze what we have here: - NVIDIA and SK Hynix announced a multiyear technology partnership on June 7-8, 2026. - The partnership focuses on co-developing next-generation memory for AI factories, including memory for Nvidia Vera Rubin AI supercomputers, Vera CPUs, RTX Spark-powered PCs, and Jetson Thor robotic computing platforms. - SK Hynix has peaked at a record $1.12 trillion valuation (in South Korean Won, likely, but the article says "third company in Asia to top the US$1 trillion market cap" - wait, in Korean Won it would be 1,000 trillion Won, which is 1 quadrillion KRW, or perhaps the article means KRW 1,000 trillion, which is about $720-$800 billion USD, or they mean KRW. Let's verify SK Hynix's market cap or just quote the article directly). - SK Hynix will use NVIDIA CUDA-X libraries, AI, and NVIDIA PhysicsNeMo framework to speed up semiconductor simulation, design, and manufacturing. Let's do a search on the $36 billion Anthropic-Google TPU Debt Deal. The watch thread `019ea15d-1286-7d6e-9dcc-200f0309cec7` says: "Track the official closing and final terms of the historic $36 billion private-credit debt package arranged by Apollo and Blackstone to buy Google TPUs for Anthropic, as well as Broadcom's $35 billion custom chip financing platform. ... Fires when the $36 billion Anthropic TPU debt deal officially closes..." Let's search for "Anthropic Google TPU debt deal Apollo Blackstone" to see if there are any updates in June 2026. This is highly relevant to the overall AI capex story.

Oh, wow! Look at that last result: "Apollo Global Management and Blackstone finalized a $35 billion debt financing package for Anthropic on June 5, 2026. The deal is designed to..." Wait, let's fetch that CryptoBriefing article: `https://cryptobriefing.com/apollo-blackstone-anthropic-35b-debt-deal/` And let's do a search on "Apollo Blackstone Anthropic finalized June 2026" or similar to see if there is a finalization. Let's fetch that CryptoBriefing article first.

This is absolutely incredible! Let's analyze what we have here: - On June 5, 2026, Apollo Global Management and Blackstone finalized a $35 billion debt financing package for Anthropic (evolved from the earlier $36 billion figure in May). - The transaction is a private credit deal structured to fund Google TPU acquisitions. - The deal is split into three tranches, with Broadcom providing residual-value guarantees on the senior tranches (backstopping the deal by guaranteeing the chips retain a certain value). - This is one of the largest private credit transactions in history. - In May 2026, Anthropic completed a massive $65 billion Series H funding round, valuing it at $965 billion post-money, and is advancing plans for an IPO. - This directly resolves the watch thread `019ea15d-1286-7d6e-9dcc-200f0309cec7` ("Closing of the $36 Billion Anthropic-Google TPU Debt Deal"). We can call `record_event` and `resolve_thread` on this! Let's do a search to double-check if there are any other details or articles about this finalized deal, just to verify and gather more quotes/details. Let's do a web search.

Let's fetch the Bloomberg article via ET CIO or other mirrors, or let's search for "Apollo wraps up $35 billion debt to buy AI chips for Anthropic - ET CIO" and fetch that URL: `https://cio.economictimes.indiatimes.com/news/investments/apollo-wraps-up-35-billion-debt-to-buy-ai-chips-for-anthropic/131577366`. This will give us the authoritative mainstream financial reporting of the deal.

This is an absolutely phenomenal source! It contains exact terms, pricing, tranches, coupons, equity, residual value support details, and Hock Tan's comments on Broadcom's "AI XPV platform" with Apollo and Blackstone. Let's analyze the details of the $35 billion deal: - Total size: $35 billion (finalized June 5, 2026). - Structure: Special-purpose vehicle (SPV) owned by Apollo's Atlas SP Partners (which provided $800 million in equity). The SPV raises capital to purchase Google custom TPUs and lease them to Anthropic. - Tranches: 1. **Tranche A1**: $6 billion, backed by Broadcom, sold to banks at a coupon of 1 percentage point over Treasuries. 2. **Tranche A2**: $24 billion, backed by Broadcom, priced at par with a 5.75% coupon, sold to institutional investors (like Apollo's Athene insurance arm). 3. **Tranche B**: $4.5 billion, no Broadcom backing, sold at par with an 8.5% coupon. - Backstop: Broadcom provides a "residual value support" agreement. If Anthropic defaults and the sold chips don't make debt investors whole, Broadcom covers 100% of the shortfall for A1 and A2. - Broadcom's Custom Chip Platform: Hock Tan (Broadcom CEO) announced the "AI XPV platform" with Apollo, Blackstone, and others to deploy over 20 gigawatts of compute capacity through 2028. - Anthropic status: Confidential US IPO filing, raised $65 billion in a Series H round late last month (May 2026) valuing it at $965 billion post-money. Wait, this is an incredible piece of evidence that the AI capex story is not only intact but is scaling to unprecedented heights, using highly sophisticated private credit and structured finance mechanisms. Let's do a quick search on other hyperscaler capex developments or any indications of a "Blackwell cliff" or slowdown. Wait, let's look at the prior findings: - `hyperscaler-capex-surge-2026`: "The core narrative supporting the artificial intelligence boom—aggressive capital expenditure by the world's largest technology companies—remains fully intact and has accelerated." We should update or write a new finding about this. - `nvidia-q1-2027-record-financials-agentic-ai`: "Nvidia's financial results for its fiscal first quarter of 2027 (ended April 26, 2026, reported on May 20, 2026) prove that the massive AI capital expenditure surge is translating directly into..." - `broadcom-q2-2026-earnings-ai-demand`: "On June 3, 2026, Broadcom (AVGO) reported its fiscal Q2 2026 earnings, delivering blowout results..." Wait, let's pull Broadcom's current public-market view to see if there is any additional context on AVGO. Let's call `market_view` for AVGO.

This is incredibly rich context. Let's look at the Broadcom (AVGO) and Nvidia (NVDA) stock performance: - NVDA: $205.10, down 6.2% in the last session, down 2.9% in the last month, down 12.9% vs its 52-week high. - AVGO: $385.73, down 7.9% in the last session, down 13.7% in the last 5 days, down 19.9% vs its 52-week high. This indicates a sharp market correction or "AI jitters" as noted in the news: "AI Jitters Drive 15x Volume Surge in TSEL ETF... investor reassessment of AI spending." Yet, the actual fundamentals and the finalized deals (like the $35 billion Anthropic TPU deal, and Nvidia's Vera Rubin platform entering full production) show that the capex story is completely intact and scaling rapidly. Let's do a quick search on "AI jitters" or "AI correction" or "market correction June 2026" to see what is driving this divergence between the stock price action and the actual capex/deals. This will help us write a highly nuanced, expert-level pair of findings: 1. One focused on Nvidia's AI capex story remaining robust, the launch of Vera Rubin, and the massive $3-4 trillion capex forecast by Jensen Huang/Colette Kress. 2. One focused on the $35 billion Anthropic TPU debt deal, which has just finalized, illustrating how private credit and structured finance are funding this massive AI capex. Let's search the web for "AI correction" or "AI stocks pullback June 2026" to understand the market sentiment.

Let's fetch the CNN or Fortune article to understand the stock sell-off on June 5, 2026. Let's fetch the Fortune article: `https://fortune.com/2026/06/05/tech-stocks-market-bloodbath-fed-rate-hikes-jobs-report-ai-fueled-chip-boom/` or `https://www.cnn.com/2026/06/05/markets/stock-market-sell-off-fed`. Let's try Fortune first. If that fails or is paywalled, we'll try CNN. Let's start with Fortune.

This is super helpful. So here is the macro backdrop and the market narrative: - Market sell-off on Friday, June 5, 2026. - The Nasdaq sank 4% (worst since April 2025). The S&P 500 lost 2.6%. - Why? A combination of: 1. Fears about the AI boom's longevity. 2. Broadcom's "disappointing guidance" late Wednesday, June 3 (when it reported earnings). Wait! Broadcom's guidance is described as "disappointing" in this article, or there was some concern. Wait, let's look at the prior findings: "On June 3, 2026, Broadcom (AVGO) reported its fiscal Q2 2026 earnings, delivering blowout results... However, in a stark reminder of..." Let's check how Broadcom's guidance was received. 3. Worries about rate hikes from the Federal Reserve after a strong jobs report (172,000 jobs added, nearly double Wall Street forecasts). 4. Geopolitical context: higher oil prices caused by the "Iran war". 5. 10-year Treasury yield rose to 4.532%. This explains the "AI jitters" and the recent correction. It's a classic macro-driven and sentiment-driven pullback, while the underlying fundamentals (finalization of the $35B Anthropic TPU deal on June 5, 2026, and Nvidia's Vera Rubin platform entering full production in June 2026) remain exceptionally strong. Let's do a search on Broadcom's Q2 FY2026 earnings to see what exactly was "disappointing" or what the guidance was, to ensure we represent Broadcom accurately. Let's do a web search.

Let's fetch the Yahoo Finance article: `https://finance.yahoo.com/markets/stocks/articles/broadcom-q2-2026-earnings-ai-111613207.html` to see why the stock sank on Q2 earnings. This is highly relevant because Broadcom is a major AI chip player alongside Nvidia, and its performance directly feeds into the market's assessment of "Is Nvidia's AI capex story still intact?" Let's fetch it.

This is incredibly helpful! Let's analyze what happened with Broadcom's earnings to see how it ties to the "Nvidia AI capex story": 1. Broadcom's Q2 FY2026 revenue was $22.19 billion (up 48% YoY), beating Wall Street's $22.13 billion estimate. Adjusted EPS was $2.44, beating the $2.40 estimate. 2. AI semiconductor revenue was $10.8 billion (up 143% YoY), beating the company's own forecast. 3. However, the stock slid (down 12% after-hours on Wednesday, June 3, and closed down 7.9% on Friday, June 5) because: - Broadcom's Q3 AI semiconductor revenue target was set at $16 billion, which came in *below* the whisper/analyst expectations of $17.2 billion (even though it represents >200% YoY growth!). - Broadcom left its full-year AI chip revenue target unchanged (instead of raising it), reiterating that it expects AI semiconductor revenue to be "in excess of $100 billion" for the fiscal year. - Hock Tan acknowledged on the call that Google would likely draw on multiple chip suppliers (adding to competitive fears). - He warned that rapidly growing AI semiconductor sales (which have lower margins than software) were diluting the company's overall gross margins. 4. Yet, Broadcom's strategic outlook is incredibly strong: - Broadcom and Meta extended their custom AI chip partnership through 2029. - Anthropic signed a compute agreement with Google and Broadcom covering approximately 3.5 gigawatts of capacity starting in 2027. - Broadcom is co-creating the "AI XPV platform" with Apollo, Blackstone, and others to deploy over 20 gigawatts of compute capacity through 2028. This ties perfectly into the overall narrative: - The physical demand for AI chip infrastructure is insatiable, and the major players (Nvidia, Broadcom) are seeing spectacular, triple-digit growth in AI hardware sales. - However, the public markets have priced these stocks to absolute perfection (NVDA at a $4.97T market cap, AVGO at a $1.83T market cap), so any minor detail—such as a Q3 AI target that is slightly below whisper numbers ($16B vs $17.2B), gross margin dilution from a shift to hardware mix, or macro factors like the Fed rate worries and the Iran war oil shocks—triggers sharp stock pullbacks. - This creates a massive divergence between the *financial stock performance* (which is undergoing a correction) and the *underlying capex fundamentals* (which are accelerating, as proven by the finalized $35B Anthropic TPU debt deal, Broadcom's $100B+ AI runway, and Nvidia's Vera Rubin platform entering full production). Let's review the prior findings and see how we can refine or add to them: - `hyperscaler-capex-surge-2026`: This finding can be updated to incorporate the $1 trillion capex forecast for 2027, Jensen Huang and Colette Kress's $3 to $4 trillion annual capex prediction by the end of the decade, and the finalized $35 billion private credit Anthropic-Google TPU debt deal. Wait, let's look at the prior findings list again. The prior findings include: - `hyperscaler-capex-surge-2026` (Hyperscaler Capex Surge: The $700 Billion+ AI Infrastructure Sprint and Private Credit Funding Shift) - `nvidia-q1-2027-record-financials-agentic-ai` (Nvidia's Record Q1 FY2027: Parabolic Demand and the "Blackwell Cliff" Dissolution) - `nvidia-groq-20b-acquihire-inference` (Nvidia's $20 Billion Groq Acquihire: Securing the Agentic Inference Market) - `broadcom-q2-2026-earnings-ai-demand` (Broadcom's Q2 FY2026: Reaffirming a $100 Billion AI Runway Amid a Sharp Market Correction) - `apple-outsources-siri-to-nvidia-blackwell` (Apple Outsources Siri Cloud Workloads to Nvidia Blackwell GPUs) Let's write two distinct, high-quality, evidence-backed findings to address the brief ("Is Nvidia's AI capex story still intact?"): 1. **Finding 1: Nvidia's Next-Gen "Vera Rubin" Platform Enters Full Production Amid a $4 Trillion Capex Vision** (This can refine `nvidia-q1-2027-record-financials-agentic-ai` or be a new finding, but let's see if we should refine `nvidia-q1-2027-record-financials-agentic-ai` or create a new one. Wait, the brief is "Is Nvidia's AI capex story still intact?" which is directly related to `nvidia-q1-2027-record-financials-agentic-ai` or `hyperscaler-capex-surge-2026`. Let's refine `nvidia-q1-2027-record-financials-agentic-ai` since it is about Nvidia's financial performance and the demand trajectory.) Wait, let's read `nvidia-q1-2027-record-financials-agentic-ai` first to see what's in there. Since we don't have the full body of it, let's call `read_notes` to see what is already written. Let's do that.

Let's write a new finding or update `nvidia-q1-2027-record-financials-agentic-ai`. Wait, the brief asks: "Is Nvidia's AI capex story still intact?" Let's update `nvidia-q1-2027-record-financials-agentic-ai` because it is about Nvidia's financial results for Q1 FY2027 (ended April 2026, reported May 2026) and how it addresses the capex story. Wait, let's also update `broadcom-q2-2026-earnings-ai-demand` or `hyperscaler-capex-surge-2026` to reflect the newly finalized $35 billion private credit deal which closed on June 5, 2026. Let's look at the thread `019ea15d-1286-7d6e-9dcc-200f0309cec7` (Closing of the $36 Billion Anthropic-Google TPU Debt Deal). The thread says: "Track the official closing and final terms of the historic $36 billion private-credit debt package arranged by Apollo and Blackstone to buy Google TPUs for Anthropic, as well as Broadcom's $35 billion custom chip financing platform." This is a perfect fit for a dedicated finding about the $35 billion private credit deal. We can write a new finding for this with a fresh slug (e.g. `apollo-blackstone-anthropic-35b-tpu-debt-deal`), and then link it from other notes. Wait! Let's write the finding for the Anthropic deal. Let's name it: "Apollo and Blackstone Finalize Historic $35 Billion private-credit Debt Deal for Anthropic's Google TPU Buildout" Wait, let's write this finding first. Let's make sure it contains: - The exact details of the $35B deal (tranches, coupons, Broadcom's residual value support, Atlas SP Partners' $800M equity). - Hock Tan's comments on the "AI XPV platform" to deploy 20 gigawatts of compute by 2028. - The context of Anthropic's confidential IPO filing and Series H round ($65 billion raised, valuing it at $965 billion post-money). - Citing the source: `/markets/AVGO/2026/06/08` (since Broadcom backstops it and Hock Tan commented on it) and the web URL `https://cio.economictimes.indiatimes.com/news/investments/apollo-wraps-up-35-billion-debt-to-buy-ai-chips-for-anthropic/131577366` and `https://cryptobriefing.com/apollo-blackstone-anthropic-35b-debt-deal/`. Let's do that! Let's call `write_note` for the $35B Anthropic deal. Then we will write/update the Nvidia capex finding. Let's compose the note body.

I will now write a comprehensive update to the Nvidia finding `nvidia-q1-2027-record-financials-agentic-ai` to incorporate the latest June 2026 developments, including the "Vera Rubin" platform entering full production, the $3-4 trillion capex forecast, and the recent market correction. This directly answers the brief on whether Nvidia's capex story is still intact.

I will now resolve the open watch thread for the closing of the $36 billion Anthropic-Google TPU debt deal, as the deal has been officially finalized at $35 billion on June 5, 2026. This is a massive milestone that I have fully documented in the note `apollo-blackstone-anthropic-35b-tpu-debt-deal`.