Cycle on Jul 7, 2026

I will start by checking the list of active threads on this topic to see what questions or watches are currently open or registered. This will help align my research with any outstanding priorities regarding Nvidia, hyperscaler capex, or AI infrastructure.

I will pull the public market view for Nvidia (ticker: NVDA) first to see its current market cap, valuation multiples, revenue, margins, and recent financial trajectory. This will establish a firm foundation of numbers for our analysis.

I see an interesting recent news item in the market view: "Nvidia Stock (NVDA) Gains after Company Denies Kyber AI Rack Delay Report" (2026-07-06), where Nvidia refuted a report from SemiAnalysis claiming its "Kyber NVL144 AI rack, designed for Rubin Ultra AI chips, would be delayed until 2028." This is highly relevant to the "Nvidia's AI capex story" and any potential delay in the next-generation platforms like Rubin. Let me search the web for details on this "Kyber NVL144 AI rack", the SemiAnalysis report, and Nvidia's denial, as well as the broader state of Nvidia's AI capex story as of early July 2026. I'll search for "Kyber NVL144", "Nvidia Kyber delay", and general hyperscaler capex updates.

I will fetch the article from Quartz ("Nvidia Kyber NVL144 rack delayed to 2028, sinking supplier stocks") to understand the core allegations of the SemiAnalysis report, and then I will search for Nvidia's official response or denial to get both sides of the story.

The Quartz article outlines that SemiAnalysis reported a 12-month delay for Nvidia's Kyber NVL144 rack-scale system (to 2028) due to PCB midplane manufacturing issues, which caused a sharp drop in Asian PCB supplier stocks on Monday, July 6, 2026. However, it also mentions that CNBC reported "existing Rubin systems have entered full production and deliveries to eight major cloud customers — among them Amazon Web Services, Microsoft Azure, and Google Cloud — are scheduled to begin this fall." Let me search for "Nvidia denies Kyber NVL144 delay" or "Nvidia Kyber CNBC" to find the exact rebuttal or clarification Nvidia issued, as mentioned in the news summary of the market view.

I will fetch the Yahoo Finance article ("Nvidia denies report its next-generation AI server faces delays...") to get Nvidia's official response and the precise details of their roadmap.

We now have a very clear picture of this recent development: - **SemiAnalysis published a report** on July 6, 2026, alleging that Nvidia's next-generation Kyber NVL144 rack-scale system (designed for Vera Rubin Ultra) has been delayed by 12+ months to 2028 due to PCB midplane manufacturing challenges. This caused Asian PCB supplier stocks to slide. - **Nvidia officially denied the report**, stating that their "roadmap remains intact" and that the Kyber server is still planned for launch with Vera Rubin Ultra in the second half of 2027. - **Vera Rubin platform** (base version) is scheduled to launch in the second half of 2026, with deliveries to cloud customers expected to begin this fall (2026). Let's do a broader search on "Nvidia AI capex story" or "hyperscaler capex Q2 2026" or "AI capex July 2026" to see if there are other indicators of whether the capex story remains intact, such as hyperscaler spending patterns, supply chain capacity, or cloud demand. I will also check if there are recent reports on the "Hyperscaler Capex Guidance Track (July 2026 Earnings)" or other watches. Note that we are currently on July 7, 2026. The Q2 2026 earnings season is just beginning (most companies will report in late July). Let's search for "hyperscaler capex 2026" or "AI capex growth 2026" to find the latest numbers.

Let's fetch the Fortune article from June 29, 2026 ("AI spending boom accelerates as Big Tech pours trillions...") or the Futurum Group article ("AI Capex 2026: The $690B Infrastructure Sprint") to get the most recent, authoritative projections and comments. I'll fetch the Fortune article.

This is excellent information. The Fortune article highlights: - **JPMorgan's midyear outlook (June 24, 2026)** raised its estimate for global AI-related capital expenditures through 2030 to **$5.5 trillion**, driven by greater capacity expansion. - **Microsoft alone** is expected to invest roughly **$190 billion** in capital expenditures in calendar year 2026. - **Micron** reported a 346% surge in quarterly revenue to **$28.2 billion** on June 24, 2026, showcasing the massive demand for HBM and memory. - Traditional smartphone-focused chipmakers like **Qualcomm** are aggressively expanding into the data center market, targeting $15 billion in annual revenue by fiscal 2029. This clearly shows that the AI capex story remains highly robust, with estimates continuing to be revised upward (such as JPMorgan raising the global AI capex forecast to $5.5T through 2030). Let's check if we have any existing findings we can update. Our prior findings: 1. `nvidia-ai-capex-cycle-sustainability` (Nvidia's AI Capex Story Remains Robustly Intact...) 2. `nvidia-q1-2027-record-financials-agentic-ai` (Nvidia's Record Financial Trajectory...) 3. `enterprise-ai-roi-gap-revenue-reckoning` (The Hyperscaler Capital Crossover...) 4. `custom-silicon-surge-openai-jalapeno-qualcomm-dragonfly` (The Custom Silicon Surge...) 5. `broadcom-q2-2026-earnings-ai-demand` (Broadcom's Q2 FY2026...) The first one, `nvidia-ai-capex-cycle-sustainability`, is the perfect fit to update with the newest early July 2026 data, including: - The recent June/July market consolidation (NVDA at $195.55, -17% from its 52-week high, -10.6% in the last month). - JPMorgan's raised global AI capex forecast to $5.5 trillion through 2030. - Microsoft's massive $190 billion CY2026 capex projection. - The SemiAnalysis Kyber NVL144 rack delay report on July 6, 2026, and Nvidia's immediate official denial, asserting their roadmap remains fully intact. - The confirmation that existing Rubin systems are in full production with cloud customer deliveries scheduled for this fall (2026). Let's read the existing note `nvidia-ai-capex-cycle-sustainability` using `read_notes` or simply fetch it to make sure we modify it correctly. I will call `read_notes` first.

Let's read the full text of `nvidia-ai-capex-cycle-sustainability` first to see where we can integrate the new findings. Since `read_notes` only gives a preview, let's write an updated version with the new data. First, let's query the tool to see if there's any other way to read notes. No, `read_notes` doesn't have a slug parameter, but we can just overwrite/update the note `nvidia-ai-capex-cycle-sustainability` using `write_note`. Let's double-check the figures we have from `market_view` for NVDA: - TTM Revenue: $253.49B (+85.2% YoY) - Gross margin: 74.1% - Operating margin: 65.6% - Profit margin: 63.0% - Latest quarter (ending April 30, 2026): Revenue $81.61B, Net income $58.32B, Free cash flow $48.59B. - Stock price: $195.55 (as of July 7, 2026), which is -17.0% from its 52-week high and down 10.6% over the last month. Let's write a comprehensive, updated finding under the slug `nvidia-ai-capex-cycle-sustainability`. I'll make sure to cite: - `/markets/NVDA/2026/07/07` - The Quartz article on the SemiAnalysis report: `https://qz.com/nvidia-kyber-rack-delay-2028-asian-supplier-stocks-070626` - The Yahoo Finance article on Nvidia's denial: `https://finance.yahoo.com/technology/article/nvidia-denies-report-its-next-generation-ai-server-faces-delays-says-roadmap-is-intact-183310296.html` - The Fortune article on global capex: `https://fortune.com/2026/06/29/ai-spending-boom-accelerates-big-tech-trillion-infrastructure-qualcomm-cfo/` Let's carefully construct the note. I will copy verbatim quotes from the fetched sources. From Yahoo Finance: > "Our roadmap remains intact." > "Nvidia plans to launch its Kyber server with its upcoming Vera Rubin Ultra platform in the second half of 2027." From Quartz: > "Nvidia's Kyber NVL144 rack-scale system has been delayed by more than 12 months to 2028 due to difficulties manufacturing a specialized circuit board, sending shares of Asian technology suppliers tumbling on Monday." > "existing Rubin systems have entered full production and deliveries to eight major cloud customers — among them Amazon Web Services, Microsoft Azure, and Google Cloud — are scheduled to begin this fall." From Fortune: > "JPMorgan also raised its estimate for global AI-related capital expenditures through 2030 to $5.5 trillion, driven by greater capacity expansion." > "Microsoft expects to invest roughly $190 billion in capital expenditures in calendar year 2026." Let's call `write_note`.

Now I need to check if any of our active watches have fired based on the findings from this cycle: 1. `Hyperscaler Capex Guidance Track (July 2026 Earnings)` (id: `019eb0cd-20a0-71c6-b91e-92d528f6792e`): Since it's July 7, 2026, the July 2026 earnings calls haven't happened yet. This remains an active watch. 2. `TSMC Q2 2026 Earnings Conference` (id: `019f1cf3-530f-7f29-a5c5-dbe8ad17023d`): Scheduled for July 16, 2026. Still in the future. 3. `Nvidia Q2 FY2027 Revenue Guidance Cross` (id: `019edf29-c3f7-7561-bccb-ef09f3a7fb09`): Scheduled for late August 2026. Still in the future. We do not have any other watches that have fired this cycle. Since we have updated the core finding with the latest development, I will now finish the cycle. Let's call `finish_cycle`.

I have successfully completed the research cycle for this topic. ### Key Findings & Updates: 1. **Nvidia's AI Capex Story Remains Highly Robust**: * **Raised Projections**: In its midyear outlook published on June 24, 2026, JPMorgan raised its estimate for global AI-related capital expenditures through 2030 to **$5.5 trillion**. * **Massive Hyperscaler Commitments**: Microsoft alone is projected to invest roughly **$190 billion** in capital expenditures in calendar year 2026. * **Nvidia's Financial Strength**: Nvidia recorded TTM revenue of **$253.49 billion** (+85.2% YoY) and generated **$48.59 billion in free cash flow** in its latest quarter ending April 30, 2026, maintaining a 74.1% gross margin ([Nvidia Market View](/markets/NVDA/2026/07/07)). 2. **Kyber NVL144 Delay Claims Rebutted**: * On July 6, 2026, research firm SemiAnalysis published a report alleging that Nvidia's next-generation Kyber NVL144 rack-scale system (designed for the Vera Rubin Ultra platform) has been delayed by over 12 months to 2028 due to PCB midplane manufacturing challenges. This caused a sharp selloff in Asian PCB suppliers. * Nvidia immediately issued an official rebuttal, stating: *"Our roadmap remains intact."* Nvidia confirmed that the Kyber server is still scheduled to launch with the Vera Rubin Ultra platform in the second half of 2027. * Furthermore, base-level Rubin systems have already entered full production, with deliveries to eight major cloud customers scheduled to begin this fall (2026). I have updated the living note `[[nvidia-ai-capex-cycle-sustainability]]` to incorporate these critical early July 2026 developments.