Cycle on Jun 27, 2026

I am starting the research cycle by listing the open threads of interest and active watches for this topic to understand what has been flagged for follow-up or monitoring.

I am calling `read_notes` to review existing findings on this topic to see what has already been documented and avoid duplicate work.

I will pull the public-markets view for NVIDIA (ticker: NVDA) first, to ground the quantitative analysis in actual market data. This is an essential step for an investable topic.

I see that the public-markets view of NVDA shows a massive $48.59 billion in free cash flow in its latest quarter (ending April 30, 2026), with TTM Revenue of $253.49B, gross margin of 74.1%, and operating margin of 65.6%. However, the stock is down 7.8% over the past month and is at a 52-week position of 51.8%, representing a 16.9% decline from its 52-week high. This short-term pullback is a great context to explore. Let's do some web searches to see what the current news and discussions are regarding Nvidia's AI capex story, whether there's any sign of cooling, and how the market is reacting in late June 2026. I'll search for topics like "Nvidia AI capex", "hyperscaler capex 2026", and see if there are any updates or research reports. Let's search with `web_search`.

The results show some very interesting articles, including: 1. "Nvidia's Micron-Led Bounce Tests Confidence in the AI Capex Cycle" on Investing.com 2. "AI Capex 2026: The $690B Infrastructure Sprint - The Futurum Group" 3. "THE AI CAPEX EXPLOSION... How Artificial Intelligence Is Reshaping..." 4. "NVIDIA's $1 Trillion AI Chip Forecast at GTC 2026 Wasn't a Shock." on Medium Let's do some specific searches to fetch these pages and read them. I'll search for the Investing.com article first to read about the "Micron-Led Bounce" and "Confidence in the AI Capex Cycle". Let's run a search to get the exact URL or find more articles from late June 2026. Wait, I can search for the exact title of that Investing.com article to get its URL and fetch it. Let's do that.

Let's search for this article on Investing.com directly or a similar tech news site to fetch it. Let's do a web search with `query: "Nvidia's Micron-Led Bounce Tests Confidence in the AI Capex Cycle" site:investing.com`.

Great! I have the exact URL: `https://www.investing.com/analysis/nvidias-micronled-bounce-tests-confidence-in-the-ai-capex-cycle-200682817`. Let's fetch it using `web_fetch`.

This is an extraordinarily rich article with concrete late-June 2026 data. Let's list some key facts from this article: - **Date**: Published on June 25, 2026. - **Stock Price**: NVDA traded in a $196.58 to $201.67 band on June 25, 2026, roughly 15% below its all-time high of $236.54 set on May 14, 2026. Market value is near $4.85 trillion. - **Catalyst 1 (Micron's blow-out earnings)**: Micron reported fiscal Q3 2026 results (which was reported on June 24, 2026) with revenue of $41.46 billion and a record gross margin of 84.9% (stealing Nvidia's "margin king crown" since Nvidia's gross margin was 74.1%). Micron disclosed $100 billion in contracted memory revenue / backlog, which reset expectations. - **Catalyst 2 (Nvidia Shareholder Meeting)**: Jensen Huang spoke at the shareholder meeting (on June 24, 2026). He declared that AI has entered a true "profitability era", characterized the current shift as the largest industry reset in 60 years, and confirmed that the next-generation **Vera Rubin platform** (the successor to Blackwell) is moving into **full-scale production**. - **Financials**: - Fiscal 2026 (ended Jan 2026) revenue reached $215.94 billion (+65.47% YoY) and net income of $120.07 billion. - Q1 FY2027 (ended April 30, 2026) revenue was $81.61B (or ~$82B), up 85% YoY, with adjusted EPS of $1.87 beating the $1.77 forecast. Free cash flow was a record $48.59 billion (or ~$49B) in the quarter. - 2026 (calendar/fiscal?) free cash flow is forecast near $195.4 billion. - **Valuation**: NVDA forward P/E is near 19.3 (which is below the S&P 500 average!), and trailing P/E is around 30.6. - **Competition**: - Qualcomm announced a major push into data center chips at its investor day: Dragonfly C1000 CPU, projecting $15 billion in segment revenue by 2029, with Meta and Microsoft as launch customers. - OpenAI is making its own chip called "Jalapeño". - Intel-backed startup SambaNova is raising $800M to $1B for AI server chips. - AMD continues to gain ground and Broadcom's custom-silicon business is growing. - **Export Controls / Gray Market**: Smuggled AI chips are surging on China's black market, but Jensen Huang stated that US national security comes before commercial opportunity and called smuggled chips a "dead end". Let's do a web search for details on Qualcomm's investor day and the "Dragonfly C1000 CPU", as well as OpenAI's "Jalapeño" chip or other competitors to verify these details. I want to search for Qualcomm's database CPU / server chip announcements and OpenAI's Jalapeño chip. Let's do a search. This will give us excellent meat for a competitive threats or capex shift note. Let's run `web_search`.

This is incredible! The news is extremely fresh from late June 2026. Let's see: 1. "OpenAI and Broadcom's AI chip has a name: Jalapeño" - Fortune article from June 25, 2026. 2. "OpenAI and Broadcom unveil Jalapeño AI Inference chip" - Techzine article. 3. "OpenAI unveils its first custom chip, built by Broadcom" - Hacker News thread (id: 48663324) from around June 24-25, 2026. 4. "OpenAI, Broadcom unveil first AI inference chip" - Constellation Research, which mentions "Qualcomm Dragonfly C1000 CPU, Qualcomm High Bandwidth..." Let's read these sources. First, let's fetch the Techzine article or Fortune article, and then let's fetch the Constellation Research page to understand both Jalapeño and Qualcomm's Dragonfly C1000 CPU. Let's fetch the Techzine or Fortune article. I'll search for the Techzine article url: `https://www.techzine.eu/news/infrastructure/142460/openai-and-broadcom-unveil-jalapeno-ai-inference-chip/`. Let's fetch it.

This is a massive development in the AI chip landscape. Let's see: - OpenAI and Broadcom unveiled OpenAI's first in-house AI chip, **Jalapeño**, which is an Intelligence Processor (accelerator) designed from the ground up for LLM inference. - Broadcom CEO Hock Tan claims it is "just as good" as Nvidia's Blackwell GPUs and Google's TPUs for LLM inference. - Developed in only **nine months** from initial design to production-ready chip, the fastest ASIC development cycle ever, aided by OpenAI's own AI models (AI designing chips). - It is manufactured with Broadcom providing interconnect and networking IP (Tomahawk network silicon), and Celestica acting as the ODM (Original Design Manufacturer) integrating the chips into server racks for deployment in Microsoft Azure data centers. - Hock Tan confirmed the road map includes rolling out **gigawatt-scale data centers starting in 2026** with Microsoft and other partners. - OpenAI filed a confidential S-1 in early June 2026 for an IPO, with an expected market value soaring toward $1 trillion. - The chip is optimized for LLM inference, whereas Nvidia's Blackwell is a generalist. This highlights a shift from general-purpose GPUs to custom ASICs for inference. Let's search for the Constellation Research article or other sources about Qualcomm's Dragonfly C1000 CPU and data center push to see what that is about. Let's do a search.

This is also a major development! Let's see: - Qualcomm unveiled its first server CPU, the **Dragonfly C1000**, at its 2026 Investor Day (held around June 24, 2026). - It is a purpose-built data center CPU designed for "agentic, general-purpose, and AI head node workloads". - Qualcomm signed **Meta** as a major customer, with Meta deploying it when it starts production in 2028. - The Dragonfly C1000 features a multi-chiplet design, claiming single-core leadership, delivering up to 250+ cores and 5 GHz by 2028. - Qualcomm also announced Qualcomm High-Bandwidth Computing (HBC) technology and the **Dragonfly AI300** inference accelerator, projecting $15 billion in segment revenue by 2029. Let's fetch the CNBC article `https://www.cnbc.com/2026/06/24/qualcomm-data-center-cpu-meta.html` and the Qualcomm press release or a tech blog to get exact details and quotes. Let's do a `web_fetch` for the CNBC article.

This is an incredible set of news! Let's analyze what's going on here. First, we have two major competitive announcements targeting Nvidia's dominance in late June 2026: 1. **OpenAI and Broadcom** introduced **Jalapeño**, OpenAI's first custom AI chip, on June 24, 2026. It is designed for LLM inference (specialist vs. generalist), developed in just 9 months (fastest ever ASIC cycle), and manufactured with Broadcom's interconnect and networking IP. Hock Tan of Broadcom noted plans for "gigawatt-scale" deployments starting in 2026 with Microsoft, and that the chip is "just as good" as Blackwell for inference. This demonstrates that the custom ASIC trend is accelerating, with major AI developers like OpenAI moving away from Nvidia's general-purpose GPUs for their inference workloads. 2. **Qualcomm** unveiled its **Dragonfly C1000** data center CPU on June 24, 2026, at its Investor Day. It signed **Meta** as a major customer for a 2028 deployment. Qualcomm is targeting $15B in data center sales by 2029 and almost doubled its non-handset revenue forecast for 2029 to $40B (from $22B). Qualcomm also acquired **Modular** (maker of Mojo and MAX, which compiles and runs AI models across different hardware), positioning it as a direct competitor to Nvidia's CUDA software moat. Wait! There's another fascinating article linked in the CNBC piece: - "Oracle stock has worst week since 2001 dot-com bust as AI financing concerns escalate" Let's search for this Oracle article to see if there are broader macro/financing concerns that might impact the AI capex story. This is crucial for evaluating if "Nvidia's AI capex story is still intact" or if there's a financing crack forming. Let's do a search for the Oracle article URL: `https://www.cnbc.com/2026/06/26/oracle-stock-ends-worst-week-since-2001-as-investors-dwell-on-finances.html`.

This is a highly significant development! Let's examine the Oracle situation: - **Oracle's stock fell 19%** in the week of June 22–26, 2026, the steepest drop since the August 2001 dot-com bust. - **The Core Issue**: To build out massive AI data centers (primarily for OpenAI, such as the Stargate data center in Abilene, Texas), Oracle is burning massive cash and raising record amounts of debt. - Capital expenditures surged **162% to nearly $56 billion** in fiscal year 2026 (ended May 2026). - Oracle generated **negative free cash flow of almost $24 billion** in FY2026. - It was sitting on **$130 billion in debt** at the end of May 2026. - To fund the buildout in FY2027, Oracle plans to raise **$40 billion** through debt and equity financing (including a $20 billion share sale), on top of $43 billion in debt sales and $5 billion in equity sales in FY2026. - **Market Impact**: This is a direct real-world example of the "Hyperscaler Capital Crossover" (which we have a finding for under `[[enterprise-ai-roi-gap-revenue-reckoning]]`). It shows that the massive capital requirements of AI are stressing corporate balance sheets, causing severe investor anxiety, and leading to massive sell-offs (Oracle down 55% from its September 2025 peak). - **Nvidia Connection**: Oracle is buying massive quantities of Nvidia GPUs to fulfill these commitments. If cloud builders like Oracle are forced to slow down their capex because of financing constraints, equity dilution, and debt-market resistance, this represents a major future headwind for Nvidia's revenue growth. This is a crucial piece of the "Is Nvidia's AI capex story still intact?" puzzle. Let's do a search on other hyperscalers or the broader market to see if there are other signs of capital crossover or capex stress. Let's look for "hyperscaler capex Q2 2026" or "AI capex financing concerns". Let's run a `web_search`.

Let's try a broader search. Let's search for `hyperscaler capex Microsoft Google Meta Amazon 2026` or `AI capex financing concerns Microsoft Meta Google`. Let's do a `web_search`.

This is excellent data! Let's analyze the capex spending patterns of the major hyperscalers: - **Total Hyperscaler Capex in 2026**: - Yahoo Finance / Apollo Academy / CNBC reports that the four major hyperscalers (Alphabet, Amazon, Meta, Microsoft) plus Oracle are spending massive amounts, with estimates ranging from **$600 billion to $725 billion in 2026**. - Let's look at specific company projections: - **Amazon**: ~$200 billion projected for 2026. - **Microsoft**: ~$190 billion projected for 2026. - **Alphabet (Google)**: ~$180 billion to $190 billion projected for 2026. - **Meta**: ~$115 billion to $145 billion projected for 2026. - **Oracle**: ~$56 billion in FY2026, planning further expansion in FY2027. - **Economic Scale**: This aggregate spending is approaching **2% of US GDP** (historically unprecedented for a single technology buildout). - **Financial Strain**: The CNBC article "Tech AI spending approaches $700 billion in 2026, cash taking big hit" and the Oracle crash show that these companies are burning through cash and taking on heavy debt, which is beginning to cause major anxiety among equity investors. Let's look at the active watches we listed earlier: 1. `019eb0cd-20a0-71c6-b91e-92d528f6792e` - **Hyperscaler Capex Guidance Track (July 2026 Earnings)**: This watch is for the upcoming earnings calls in July 2026. Since the current date is June 27, 2026, these earnings have not happened yet. This remains an ongoing watch. 2. `019edf29-c3f7-7561-bccb-ef09f3a7fb09` - **Nvidia Q2 FY2027 Revenue Guidance Cross**: This is for the August 2026 report (guiding $91.0 billion). It has not happened yet. 3. `019ee44e-b944-74a7-bc73-b87fb1bf1ac5` - **Aggregate Hyperscaler Free Cash Flow Crossover (Q3 2026)**: Expected around October 2026. This has not happened yet. 4. `019f0335-ed19-7036-b3c4-d445abcf96c7` - **First Commercial Shipments of Nvidia Vera Rubin Platform**: Scheduled for Q3 calendar 2026 (fire date Sept 30, 2026). However, in Jensen Huang's shareholder meeting on June 24, 2026, he confirmed that the next-generation Vera Rubin architecture is **moving into full-scale production**. This is a massive milestone! It doesn't fully resolve the watch (which is for "commercial shipments or cloud deployments" in Q3), but it's a major step forward. Let's make sure we update our note on the Vera Rubin platform. Wait, let's examine the note `[[nvidia-ai-capex-cycle-sustainability]]`. The preview says: "In mid-2026, the artificial intelligence capital expenditure boom continues to defy skepticism, transitioning rapidly from the Blackwell generation to the newly announced Vera Rubin platform…" Let's see if we should refine this note or create a new note, or update existing ones. Wait, let's look at the slugs: 1. `nvidia-ai-capex-cycle-sustainability`: "The AI Capex Sustainability Debate: Is Nvidia's Hyper-Growth Runway Still Intact?" - This fits perfectly with our topic! We should update this note to include: - The latest stock performance of Nvidia (down 7.8% over the past month, trading near $200, 15% below its May 14 peak of $236.54, but supported by a $4.82T market cap and record $48.59B quarterly free cash flow). - Jensen Huang's declarations at the June 24, 2026 shareholder meeting: AI has entered a "profitability era", and the Vera Rubin architecture is moving into "full-scale production". - Micron's blowout fiscal Q3 2026 earnings on June 24, 2026 (revenue of $41.46 billion, gross margin of 84.9%, and $100 billion backlog), which acted as a massive sector catalyst to lift Nvidia and restore confidence in the capex cycle. 2. `enterprise-ai-roi-gap-revenue-reckoning`: "The Hyperscaler Capital Crossover: AI Capex Outstrips Cash Flow, Triggering Historic Debt and Equity Waves" - This fits perfectly with the Oracle news! Oracle's massive 19% stock crash in late June 2026 (its worst week since the 2001 dot-com bust), driven by $130 billion in debt, $24 billion in negative free cash flow, and its plans to raise $40 billion in FY2027 to fund its AI buildout for OpenAI. - We should update this note to include the Oracle crash as the most acute real-world manifestation of the capital crossover. 3. `nvidia-q1-2027-record-financials-agentic-ai`: "Nvidia's Record Financial Trajectory: S&P Upgrade to 'AA', $25B Jumbo Bond Sale, and Relentless Blackwell Demand" - We can also update this or reference it. 4. Let's write a new finding or update an existing finding on custom silicon/ASIC competition, since OpenAI's **Jalapeño** chip (built with Broadcom and Celestica) and Qualcomm's **Dragonfly C1000** data center CPU represent a pivotal shift in the AI chip competitive landscape. - Wait, let's see if we have an existing finding on custom silicon. - Let's check the list of notes we read: - `enterprise-ai-roi-gap-revenue-reckoning` - `micron-ai-memory-pricing-power` - `nvidia-ai-capex-cycle-sustainability` - `apollo-blackstone-anthropic-35b-tpu-debt-deal` (this tracks the $35B AI XPV platform for Anthropic's buildout, which uses custom TPUs from Broadcom) - `hyperscaler-capex-surge-2026` - `nebius-group-ai-cloud-infrastructure` - `nvidia-q1-2027-record-financials-agentic-ai` - `broadcom-q2-2026-earnings-ai-demand` - `apple-outsources-siri-to-nvidia-blackwell` - `nvidia-groq-20b-acquihire-inference` - We don't have a specific note for the custom silicon threat or the custom ASIC shift, though `broadcom-q2-2026-earnings-ai-demand` and `apollo-blackstone-anthropic-35b-tpu-debt-deal` touch on custom chips. - Let's create a new note specifically for this: `custom-silicon-surge-openai-jalapeno-qualcomm-dragonfly` or similar, or we can update `broadcom-q2-2026-earnings-ai-demand` or create a fresh note on the competitive threats to Nvidia's CUDA moat and hardware dominance. Let's create a fresh note titled: "The Custom Silicon Surge: OpenAI's Jalapeño and Qualcomm's Dragonfly C1000 Challenge Nvidia's Moat" with slug `custom-silicon-surge-openai-jalapeno-qualcomm-dragonfly`. Wait! Let's double check if there are any other notes we should read. Let's run a `read_notes` query with a limit of 50 to see if there are any other notes that might be relevant. We did that and got 10 notes. That is the complete list of notes for this topic. Let's read `nvidia-ai-capex-cycle-sustainability` fully so we can update it properly without losing prior context. I can fetch its content. But wait! `read_notes` only returned previews. How do we read the full body of a note? Ah, the `read_notes` tool lists notes, but wait, there isn't a direct "read_note" tool. Oh, wait, is there? No, the tools are: - `list_threads` - `record_thread` - `resolve_thread` - `record_watch` - `record_event` - `web_search` - `web_fetch` - `market_view` - `write_note` - `read_notes` - `finish_cycle` Ah! `read_notes` returns the slug, title, preview, updated_at, and source_count. To get the full body of a note, let's see if we can call `read_notes` or if we have to write it based on the preview and our updates. Wait, let's look at the `read_notes` parameter: it takes `topic_id` and `limit`. It returns the preview. Let's check if there is any other way to read the full body. No, there's no other tool. That's fine! We can write the updated note with the same slug. The system description says: "reusing a slug updates the existing note and snapshots the prior content into note_revisions. Slug is path-segment safe; reusing a slug updates the existing note... When this cycle advances a subject you already have a finding for, call write_note with that SAME slug to revise it in place". Let's draft our updates for the existing notes first. ### Note 1: `nvidia-ai-capex-cycle-sustainability` **Title**: The AI Capex Sustainability Debate: Is Nvidia's Hyper-Growth Runway Still Intact? **Content**: We will synthesize: 1. **Nvidia's Stock Pullback and Valuation Paradox**: The stock has pulled back 7.8% over the past month to $195.74 (as of June 26, 2026), about 15% below its May 14 all-time high of $236.54, with its market cap sitting at $4.82 trillion. Despite the absolute scale, NVDA trades at a forward P/E of just 19.3 (below the S&P 500), reflecting the massive earnings growth expectations. 2. **The Micron Catalyst**: On June 24, 2026, Micron's fiscal Q3 blowout (revenue of $41.46B, gross margin of 84.9%, and $100B backlog) reignited sector-wide confidence, acting as a massive bounce catalyst for Nvidia. 3. **Jensen Huang's Shareholder Meeting Declarations**: On June 24, 2026, CEO Jensen Huang declared that AI has entered a true "profitability era," with autonomous agents driving compute demand for decades. Crucially, he confirmed that the next-generation **Vera Rubin platform** (successor to Blackwell) has officially moved into **full-scale production**. 4. **Financial Performance**: Nvidia's Q1 FY2027 (ended April 30, 2026) delivered revenue of $81.61 billion (+85.2% YoY), gross margin of 74.1%, and a record $48.59 billion in free cash flow. Calendar 2026 FCF is projected to reach $195.4 billion, supporting its $4.82T valuation on fundamentals. Let's write this update to `nvidia-ai-capex-cycle-sustainability`. Wait, let's review the guidelines for writing notes: - No YAML frontmatter — the title is the first `# ` heading and the slug is the slug parameter you pass. - Name and link the actual thing (the story, article, paper, or page that was posted/published) — not just "the discussion". - Carry 1-2 REPRESENTATIVE VERBATIM QUOTES, copied exactly, each kept with the link it came from. - Then say what it means. - To link another finding, use its slug as a wikilink, e.g., `[[enterprise-ai-roi-gap-revenue-reckoning]]`. Let's fetch the exact quotes from the Investing.com article to include in `nvidia-ai-capex-cycle-sustainability`: Quote 1: > "CEO Jensen Huang used the company's shareholder meeting to declare that AI has entered a true profitability era and confirmed the next-generation Vera Rubin architecture is moving into full-scale production, reframing the stock's narrative just as the AI trade needed reassurance." Link: `https://www.investing.com/analysis/nvidias-micronled-bounce-tests-confidence-in-the-ai-capex-cycle-200682817` Quote 2: > "The Thursday advance was borrowed from Micron... Micron's fiscal third-quarter results, with revenue of $41.46 billion and a record gross margin of 84.9%, reignited confidence across the entire AI semiconductor chain after a week in which the PHLX Semiconductor Index tumbled 8% on Tuesday and Nvidia itself shed more than 4%." Link: `https://www.investing.com/analysis/nvidias-micronled-bounce-tests-confidence-in-the-ai-capex-cycle-200682817` Let's write the note `nvidia-ai-capex-cycle-sustainability` now. I will state my intent before calling the tool.

I will now write the updated note for `nvidia-ai-capex-cycle-sustainability` to incorporate the late June 2026 developments, including Nvidia's recent stock performance, Jensen Huang's shareholder meeting confirmations, and Micron's massive earnings beat.

I will now write the updated note for `enterprise-ai-roi-gap-revenue-reckoning` to capture the dramatic Oracle stock crash in late June 2026. This is the most acute real-world manifestation of the "Hyperscaler Capital Crossover" where massive AI capex outstrips corporate cash flow, forcing historic debt issuance and threatening balance sheets. I will state my intent before calling the tool.

I will now write a fresh note with slug `custom-silicon-surge-openai-jalapeno-qualcomm-dragonfly` to document the major late June 2026 competitive threats to Nvidia's AI chip dominance. This includes OpenAI and Broadcom's joint unveiling of **Jalapeño**, OpenAI's first in-house AI inference chip, and Qualcomm's unveiling of the **Dragonfly C1000** data center CPU with Meta signed as a major customer. I will state my intent before calling the tool.

I will now call `finish_cycle` to complete this research cycle, as I have updated the relevant notes with the latest findings from late June 2026. I will state my intent before making the tool call.

### Executive Summary of the Research Cycle This research cycle evaluates whether Nvidia's AI capital expenditure story remains intact in late June 2026. The findings confirm that while near-term demand indicators and physical supply-chain backlogs continue to support Nvidia’s hyper-growth runway, structural cracks and long-term competitive threats are accelerating rapidly. #### Key Findings 1. **Nvidia De-Risks Its Next Product Cycle Amid Sector Bounce** * **Vera Rubin Production**: At Nvidia’s annual shareholder meeting on June 24, 2026, CEO Jensen Huang confirmed that the next-generation **Vera Rubin platform** (the successor to Blackwell) has officially moved into **full-scale production**, de-risking its next major revenue leg. Huang declared that AI has entered a true “profitability era.” * **Micron's Blowout Catalyst**: Confidence in the capex cycle was strongly reinforced on June 24, 2026, by Micron's fiscal Q3 blowout (revenue of $41.46B, record 84.9% gross margin, and a $100 billion backlog), which acted as a massive sector-wide catalyst to lift Nvidia. * **Valuation Paradox**: Despite a $4.82 trillion market cap, Nvidia's explosive earnings growth has compressed its valuation, leaving it trading at a forward P/E of just 19.3—below the S&P 500 average. 2. **The Capital Crossover Triggers Severe Balance Sheet Strain** * **Oracle's 19% Crash**: The financial stress of the AI buildout was acutely demonstrated when **Oracle (ORCL) suffered a 19% stock crash** (its worst week since the 2001 dot-com bust). To fund its massive AI data center commitments for OpenAI (such as the 1GW "Stargate" project), Oracle’s capex exploded 162% to $56 billion in FY2026, leading to a massive **negative free cash flow of $24 billion** and $130 billion in accumulated debt. * **Macro Capex Scale**: The aggregate capex planned by the top four hyperscalers plus Oracle is projected to reach **$600B to $725B in 2026** (approaching 2% of US GDP), putting severe strain on corporate cash generation and raising debt concerns that could eventually force a deceleration in hardware purchases. 3. **The Custom Silicon Surge Directly Attacks Nvidia's Moat** * **OpenAI's Jalapeño Chip**: On June 24, 2026, OpenAI and Broadcom unveiled **Jalapeño**, OpenAI's first in-house AI inference chip. Built in just nine months (the fastest high-performance ASIC cycle in history) using OpenAI’s own models, the chip is designed for specialized LLM inference and will power gigawatt-scale data centers starting in 2026. * **Qualcomm's Dragonfly C1000 CPU**: Qualcomm unveiled its first server CPU, signing **Meta** as a launch customer for 2028. Qualcomm also acquired **Modular** (an open-source equivalent to Nvidia's CUDA), directly targeting Nvidia's software moat and positioning for $15 billion in data center sales by 2029. These findings show that while Nvidia's immediate financial runway and product pipeline remain secure, the broader AI buildout is entering a more challenging phase characterized by severe financial strain on cloud builders and an aggressive shift toward custom, power-efficient, and lower-cost ASICs for inference workloads.