Cycle on Jul 4, 2026

I am starting this research cycle by listing the active threads on this topic to see what specific questions or watches are currently open, and then I will query Nvidia's public markets view to inspect its latest financials and valuation, followed by web searches for recent news on AI capex and Nvidia's supply chain/demand indicators.

I will now fetch the public markets view for Nvidia (NVDA) using the `market_view` tool to check its current valuation, revenue trajectory, margins, and recent news. This will provide a quantitative foundation for our analysis of whether Nvidia's AI capex story remains intact.

I see some very interesting notes in the recent news, particularly: 1. "AAOI Stock Falls 17% but Still Outpaces Nvidia by 200% - Applied Optoelectronics (NASDAQ:AAOI)" on July 3, 2026, which mentions: "dropped 17% due to comments from Meta Platforms CEO Mark Zuckerberg regarding AI infrastructure spending, which triggered a broad sell-off in photonics stocks." 2. "Tesla Joins Uber in Capping Employee AI Spending - Tesla (NASDAQ:TSLA), Uber Technologies (NYSE:UBER)" on July 3, 2026. I need to search the web to understand Mark Zuckerberg's recent comments regarding AI infrastructure spending, as well as general updates on Nvidia's AI capex demand indicators, hyperscaler capex plans (Meta, Microsoft, Alphabet, Amazon), and the overall state of the AI capex cycle. Let's run some targeted web searches.

Ah, let's fetch the CNBC article `https://www.cnbc.com/2026/07/01/meta-stock-cloud-ai-compute.html` and the AOL article `https://www.aol.com/articles/money-pit-zuckerberg-just-exposed-175856000.html` to find out what exactly Mark Zuckerberg said that triggered this. This seems to be a major development in the AI capex story where Meta is looking to monetize or lease out its excess AI compute or is changing its strategy. Let's fetch these.

This is a massive development. Let's look at the AOL article or do a search for the AOL article to understand why it states "Zuckerberg Just Exposed Why Hyperscaler AI Spending...". Let's search and fetch it.

Let's fetch the MSN article to see what that is about, since it's a recent article: `https://www.msn.com/en-us/money/markets/money-pit-zuckerberg-just-exposed-why-hyperscaler-ai-spending-keeps-going-up/ar-AA23hYJv?gemSnapshotKey=GM08777656-snapshot-2&uxmode=ruby&apiversion=v2&domshim=1&noservercache=1&noservertelemetry=1&batchservertelemetry=1&renderwebcomponents=1&wcseo=1` (or let's just do a web search about "Meta" "cloud push" or "excess AI compute" to understand the broader market reaction). Let's fetch the MSN article first.

This is a crucial insight: Meta's capex was raised to $125B-$145B, and a major reason was actually component pricing (like HBM memory and Nvidia chips) and not just physical site expansion. Furthermore, the July 1, 2026 news that Meta is launching a cloud business to sell excess AI compute capacity is a massive shift. Let's do a search on how the market is reacting to this and what other hyperscalers are doing. Let's search for "Meta cloud business" "excess AI compute" and "Nvidia" to see if there's any commentary on how this affects Nvidia's demand or the neoclouds (CoreWeave, Nebius, etc.).

Let's look into the Reuters article `https://www.reuters.com/business/meta-sell-excess-ai-computing-capacity-via-cloud-business-bloomberg-news-reports-2026-07-01/` or the Bloomberg article using a web fetch, or search for other articles discussing this, to see what the direct implications are for Nvidia and the neoclouds (CoreWeave, Nebius, etc.). Let's search for "Meta cloud business" "CoreWeave" OR "neocloud" OR "Nvidia" to get the market commentary.

Let's fetch the Barron's article `https://www.barrons.com/articles/meta-stock-ai-cloud-coreweave-nebius-1e35955b` and the Seeking Alpha article `https://seekingalpha.com/article/4919343-coreweave-stock-meta-compute-scare-a-long-term-buying-opportunity` if we can, or search for "Meta Is Spending $48 Billion With CoreWeave and Nebius" to see what that is about. Let's do a search for that exact text.

Let's fetch `https://www.fool.com/investing/2026/04/21/meta-is-spending-48-billion-with-coreweave-nebius/` to get the context of this $48 billion contract. This is incredibly detailed and highly relevant to the neoclouds and the overall Nvidia GPU demand story.

This is incredibly rich context. Let's synthesize: 1. Meta has committed $48B total with neoclouds: $21B with CoreWeave (`CRWV`) and up to $27B with Nebius (`NBIS`) to secure access to Nvidia GPUs. 2. Yet on July 1, 2026, Bloomberg and CNBC reported that Meta is building out a *cloud business* itself to sell *excess AI compute capacity*. 3. This announcement caused neocloud stocks (CoreWeave and Nebius) to plunge ~12% on Wednesday, July 1, 2026, because Meta, their massive customer, is now standing up as a potential competitor to sell excess compute or models hosted on its infrastructure. 4. This "excess capacity" indicates that Meta is either overprovisioned or looking to monetize its massive capex ($125B-$145B guided for 2026). 5. Let's look for more information on Nvidia's demand. If Meta has "excess" capacity, does it mean the AI capex boom is peaking, or is Nvidia's order book still extremely robust? Let's search for Nvidia's recent demand signals, Blackwell and Rubin shipments, or general analyst views on Nvidia's capex story in June/July 2026. Let's do a search.

Ah, let's broaden the query to: `Nvidia Blackwell Rubin demand June OR July 2026` or `Nvidia AI capex demand June OR July 2026`. Let's search.

Let's fetch the TrendForce article on Rubin delays and Blackwell shipments: `https://iconnect007.com/article/149537/rubin-faces-delays-blackwell-to-drive-70-of-nvidia-highend-gpu-shipments-in-2026/149534/design`. This is an excellent industry source on Nvidia's product transition and demand dynamics.

Let's do a web search on other big tech earnings or capex news from mid-2026. Let's see if there are any other specific updates on Microsoft, Amazon, or Google capex, or if Nvidia's overall order book is still viewed as intact. Let's search.

Let's fetch the Futurum Group article: `https://futurumgroup.com/insights/ai-capex-2026-the-690b-infrastructure-sprint/` to see their latest projections and analysis of AI capex in 2026. This will give us a strong independent industry perspective.

This is an incredible analysis of the overall AI capex landscape in early/mid-2026. Let's look at the key details: - The 5 largest US cloud and AI infrastructure providers (Microsoft, Alphabet, Amazon, Meta, and Oracle) have collectively committed to spending between $660B and $690B (or even up to $725B according to some newer estimates in mid-2026) on capex in 2026, which is nearly double 2025 levels. - Specifics: - Amazon: $200B capex plan for 2026 (mostly data centers). - Alphabet: $175B–$185B. - Meta: $115B–$135B (revised up to $125B–$145B as of late Q1/mid-2026, with Susan Li confirming $125B–$145B). - Microsoft: $120B+ for fiscal 2026. - Oracle: $50B (targeting $40B in debt/equity in FY2027 to fund it). - The "Stargate" project: JV between OpenAI, SoftBank, Oracle, and MGX targets $500B by 2029. - Revenues: OpenAI ARR ended 2025 at $20B; Anthropic run rate surpassed $9B in January 2026. - The Meta Cloud Push: On July 1, 2026, Meta announced it is building a cloud business to sell excess AI compute capacity (CNBC, Bloomberg). This has caused neocloud companies CoreWeave and Nebius to plunge ~12% because Meta is a major customer of both ($21B with CoreWeave and up to $27B with Nebius) and is now positioning itself to sell excess compute. - Despite this, Nvidia's demand remains exceptionally high, with Jensen Huang noting demand for Blackwell and upcoming Rubin chips could reach $1 trillion by 2027. - TrendForce notes that the Blackwell platform will account for over 70% of shipments in 2026, while the Rubin series faces delays (due to HBM4 validation, CX9 interconnect transition, power consumption, and liquid cooling optimization), lowering its share from 29% to 22%. Let's check if there are any other recent articles or events to review. I will write two comprehensive findings update notes: 1. **`nvidia-record-q1-results-and-trillion-dollar-order-book`**: To update Nvidia's AI capex story with the latest numbers (from the market view and the Futurum report), the Blackwell/Rubin shipment mix (70%+ Blackwell in 2026, Rubin delays), and Jensen Huang's updated $1 trillion chip demand projection through 2027. 2. **`gpu-specialized-clouds-and-strategic-co-investment`**: To track the dramatic shift in the neocloud market sparked by Meta's July 1, 2026 announcement of its new cloud business to sell excess AI compute capacity, which sent CoreWeave and Nebius shares down ~12%, and contextualize this against Meta's existing $48 billion in commitments to them ($21B with CoreWeave, up to $27B with Nebius). Wait, let's also check if there are any other watches we can resolve or record. Let's see: - For `gpu-specialized-clouds-and-strategic-co-investment`: CoreWeave's capex and Nebius' Q2 earnings are still upcoming. - Let's check `019ed2f6-f136-7051-a5b0-fa3b26ec02e0` (Nvidia Q2 FY2027 Earnings Release expected late August 2026) -> keep open. - Let's check `019ef1e2-810a-7234-ab4a-1680b5c05e86` (NY Governor Hochul action on Data Center Moratorium Bill, fire date July 15, 2026) -> keep open. Let's write the updated find notes first. Let's make sure we ground the notes in the concrete evidence we retrieved this cycle, with the exact links and quotes. Let's double check if there are any other specific articles we can fetch to get a direct quote from the Meta cloud push. The CNBC article: `https://www.cnbc.com/2026/07/01/meta-stock-cloud-ai-compute.html` Quote: > "Shares of Meta closed up nearly 9% on Wednesday following news that the company is building out a new cloud business that could help recoup some of the billions of dollars it’s poured into artificial intelligence infrastructure." > "Meta will sell its excess computing power to outside customers, CNBC’s Jim Cramer confirmed. Bloomberg was first to report the news." > "Shares of neocloud companies CoreWeave and Nebius Group both plunged following the Meta news, sinking about 12% each." The Motley Fool article on the $48B deals: `https://www.fool.com/investing/2026/04/21/meta-is-spending-48-billion-with-coreweave-nebius/` Quote: > "Meta recently committed to spending $21 billion with CoreWeave just after agreeing to spend up to $27 billion with Nebius." > "Both Nebius and CoreWeave look to sign long-term contracts with big AI companies like Meta and OpenAI." The TrendForce article on Blackwell/Rubin shipments: `https://iconnect007.com/article/149537/rubin-faces-delays-blackwell-to-drive-70-of-nvidia-highend-gpu-shipments-in-2026/149534/design` Quote: > "This downward revision is primarily due to shipment delays facing the Rubin series. In addition to the time required for HBM4 validation, challenges include transitioning network interconnects from CX8 to CX9, managing significantly higher power consumption, and optimizing performance under more advanced liquid cooling solutions. As a result, Rubin’s share of NVIDIA’s high-end GPU shipments is expected to decline from 29% to 22%." > "By comparison, the more mature Blackwell platform is estimated to account for over 70% of shipments, led by the GB300/B300 series." The Futurum Group article on Hyperscaler capex: `https://futurumgroup.com/insights/ai-capex-2026-the-690b-infrastructure-sprint/` Quote: > "The five largest US cloud and AI infrastructure providers – Microsoft, Alphabet, Amazon, Meta, and Oracle – have collectively committed to spending between $660 billion and $690 billion on capital expenditure in 2026, nearly doubling 2025 levels." > "This includes Amazon with a projected $200 billion in capex for 2026 (most, but not all, for data centers), Alphabet at $175-185 billion, Meta at $115-135 billion, Microsoft tracking toward $120 billion or more, and Oracle targeting $50 billion." The MSN article on Meta's capex being driven by component costs: `https://www.msn.com/en-us/money/markets/money-pit-zuckerberg-just-exposed-why-hyperscaler-ai-spending-keeps-going-up/ar-AA23hYJv?gemSnapshotKey=GM08777656-snapshot-2&uxmode=ruby&apiversion=v2&domshim=1&noservercache=1&noservertelemetry=1&batchservertelemetry=1&renderwebcomponents=1&wcseo=1` Quote: > "Meta Platforms' Chief Financial Officer, Susan Li, said in the earnings call that they anticipate 2026 capital expenditures '...to be in the range of $125-145 billion, increased from our prior range of $115-135 billion.'" > "Mark Zuckerberg gave that rationale early in the call, saying, '...we are increasing our infrastructure capex forecast for this year. Most of that is due to higher component costs, particularly memory pricing'." Let's write these two updated notes. I will call `write_note` for each. I will reuse the slugs `nvidia-record-q1-results-and-trillion-dollar-order-book` and `gpu-specialized-clouds-and-strategic-co-investment`. Let's do this now.

I will now write the second note regarding the neocloud providers (CoreWeave and Nebius) and Meta's dramatic shift to sell its excess AI compute capacity in a new cloud business, which directly impacts their business model and caused their stocks to plunge. I will reuse the slug `gpu-specialized-clouds-and-strategic-co-investment`. Let's do this now.

I have updated both notes with the critical new evidence from this cycle. Let's review if there are any other threads we can resolve or watches we should register. We don't have any threads to resolve right now as the existing threads are forward-looking watches (e.g., Nvidia Q2 earnings in late August 2026, Alphabet Q2 earnings in late July, etc.). I will now call `finish_cycle` to complete this research cycle.

I have successfully completed this research cycle on Nvidia's AI capex story. ### Summary of Findings: 1. **Nvidia's AI Capex Story Remains Exceptionally Robust**: * **Massive Hyperscaler Capex**: The five largest US infrastructure providers (Amazon, Alphabet, Microsoft, Meta, and Oracle) have collectively committed to spending between **$660B and $690B** in 2026, nearly doubling 2025 levels. * **Nvidia Blackwell Prominence**: Due to delays in Nvidia's next-generation "Vera Rubin" platform (arising from HBM4 validation, CX9 interconnect transitions, and liquid cooling optimization), the mature **Blackwell platform** is now projected to drive **over 70%** of high-end GPU shipments in 2026. Nvidia's order book remains incredibly strong, with CEO Jensen Huang forecasting up to $1 trillion in demand for Blackwell and Rubin chips through 2027. * **Financial Performance**: Nvidia's TTM revenue has grown to **$253.49B (+85.2% YoY)** with a **74.1% gross margin** and **$50.34B in operating cash flow** in its latest reported quarter, demonstrating that hyperscaler capex is converting directly into highly profitable cash flows for the chip designer. 2. **Competitive Disruption in the Neocloud Market**: * **Meta's Cloud Pivot**: On July 1, 2026, Meta announced plans to launch its own cloud business to sell **excess AI compute capacity** and hosted models to outside customers. * **Market Impact**: This announcement sent specialized GPU cloud providers **CoreWeave (`CRWV`)** and **Nebius Group (`NBIS`)** shares tumbling **~12%** in a single day. * **Overprovisioning Under the Spotlight**: The pivot is highly disruptive because Meta is a primary financial backer of these neoclouds, having previously committed **$48 billion** in long-term contracts ($21B with CoreWeave and up to $27B with Nebius). Meta's entry as a direct competitor to monetize its excess capacity exposes the commodity risk of dedicated GPU clouds as hyperscalers look to recoup their massive infrastructure investments. Both findings have been updated under their stable slugs: * `nvidia-record-q1-results-and-trillion-dollar-order-book` * `gpu-specialized-clouds-and-strategic-co-investment`