Core Insights - AlphaTON Capital Corp. has announced a strategic investment to acquire a high-performance compute cluster of over 1,000 Nvidia B200 GPUs, valued at approximately $68 million, positioning itself as a key compute provider for Telegram's Cocoon AI network [1][6][10] Transaction Structure and Financing - The acquisition will be financed through a combination of $30 million in equity capital from AlphaTON and $52.5 million in debt financing facilitated by Vertical Data, resulting in a total project capitalization of approximately $82.5 million [2][6] - The debt facility will be fully amortized over 36 months, maintaining debt service coverage ratios above 1.0x throughout the schedule [2] Strategic Infrastructure Partnership - AlphaTON has partnered with CUDO Compute, a leading European NVIDIA Cloud Partner, to deploy and operate the GPU cluster, ensuring production-grade uptime and scalable governance [3] - SNET Energy UK Ltd will provide expertise in sustainable power data center development, while Vertical Data will assist with GPU cluster deployment and hardware sourcing [4] Hardware Specifications and Capabilities - The compute cluster will feature over 1,000 Nvidia B200-class GPUs, designed for AI inference and training, with a target utilization rate of 90% [5][12] - The investment is projected to yield an internal rate of return (IRR) of 59.7%, with a projected net present value (NPV) of $59.6 million and a return on investment (ROI) of 615% over five years [7] Revenue Streams and Financial Projections - The cluster will generate revenue through dedicated workloads for Cocoon AI and external GPU rentals via CUDO Compute's marketplace [7] - Revenue projections account for conservative assumptions, including a 20% annual price compression in the first year and a 10% in subsequent years, with consistent EBITDA margins ranging from 64-73% over a five-year forecast [8] Strategic Rationale - AlphaTON aims to democratize AI with an "AI For Good" initiative, focusing on ethical access, user data sovereignty, and environmental responsibility [9][10] - The company emphasizes the importance of decentralized AI services, contrasting its approach with the centralized models of major tech companies [11] Deployment Timeline and Commercial Milestones - The transaction is expected to close in Q4 2025, with hardware delivery and performance validation planned for Q1 2026, followed by full implementation of Cocoon AI workloads in Q2 2026 [13][15] Market Context and Competitive Positioning - The investment aligns with a significant growth forecast in AI infrastructure, with data center AI workload capacity projected to grow 3.5x from 2025 to 2030 [16] - The U.S. government's support for AI infrastructure, including a 10% equity stake in Intel, underscores the sector's importance [17]

摩根士丹利科技对话：乔·摩尔谈 NVDA 季度后表 现,AVGO vs NVDA vs AMD 偏好鉴于 GOOGL 热 情,MU SNDK 最新亚洲调查+EPS 实力 20251125 摘要 英伟达库存增加主要用于扩充机架，客户库存并不充足，DSO 处于历史 平均水平，无需过度担忧。公司正进行战略性投资，以利用资产负债表 发展业务和巩固合作关系，长期来看将受益于人工智能项目，如 OpenAI 的 1,000 兆瓦项目。 尽管新产品周期面临挑战，但对 2026 年 Blackwell 到 Ruben 的产品 交接持乐观态度。亚洲市场信息显示产品上市情况良好，需求旺盛，预 计 Vera Rubin 需求量巨大，市场信心有望恢复。 谷歌 Gemini 3 的良好表现可能提振 TPU 业务，预计 2025 年谷歌与 TPU 相关业务收入将保持中等水平并有所增长。博通处理器业务实现三 位数增长，英伟达、AMD 等公司预计实现同比翻番或接近这个水平。 内存领域需求强劲，来自云计算、AI 及传统服务器 DDR5 内存的需求均 显著增长。供给严重不足导致价格大幅上涨，部分产品价格涨幅达 40%-70%，甚至临时采购价格较最 ...

AI 发展为何离不开金属软磁粉芯 20251125 摘要 英伟达 AI 算力卡功耗持续提升，从 2022 年安培架构的 700 瓦增长到 2025 年 Blackwell Ultra 架构 GB300 芯片的 1,400 瓦，对电源分配 网络（PDN）提出更高要求，推动垂直堆叠设计和模组式电感的需求。 ASIC 芯片在面积、能耗和成本方面优于通用 GPU，海外 AI 巨头如英伟 达、英特尔、亚马逊和 Meta 正积极布局自研 ASIC，Meta 预计 2025 年四季度推出超越英伟达规格的 AI ASIC 芯片。 垂直堆叠设计通过近距离耦合减小 PDN 损耗，但要求内部元件小型化， 模组式电感以其高功率密度和更小尺寸成为关键解决方案，可提供高达 60%的功率密度。 金属软磁粉芯在高功率、大电流 AI 服务器中优势明显，其高饱和磁通密 度（0.9~1.6 特斯拉）允许更小体积，并能满足大电流对温度稳定性的 要求，避免磁芯饱和。 供应链调研显示，尽管英伟达目前在 AI 服务器市场占据主导地位，但谷 歌、亚马逊等公司正在加速部署自研 ASIC 芯片，预计 2025 年底至 2026 年间，海外 AI 巨头 ASIC ...

Summary of Key Points from the Conference Call Company: NVIDIA Core Insights and Arguments - **Share Repurchases and Shareholder Value**: Since 2018, NVIDIA reported approximately $205 billion in net income and $188 billion in free cash flow, while repurchasing $91 billion in shares. The claim that repurchases did not create shareholder value is countered by the assertion that repurchases were made below intrinsic value, boosting EPS by 5% and adding over $200 billion in market capitalization at constant P/E [4][5][7] - **Accounts Receivable Growth**: NVIDIA's Days Sales Outstanding (DSO) of 53 days is slightly above the historical average of 52 days, indicating that the company is not struggling to collect payments from customers. The claim of increasing DSO is refuted by stating that overdue accounts receivable are negligible [8][9] - **Inventory Growth and Demand**: A 32% quarter-over-quarter increase in inventory is interpreted as weak demand. However, NVIDIA argues that inventory includes raw materials and work-in-progress, and is consistent with historical trends in anticipation of future growth [10][11][12] - **Cash Flow Conversion**: NVIDIA generated $23.8 billion in cash from operations in Q3, with a free cash flow conversion rate competitive with peers. The claim of low-quality earnings is dismissed as NVIDIA's free cash flow has averaged 98% of GAAP net income since FY18 [13][15][16] - **Circular Financing**: Allegations of NVIDIA participating in a $610 billion circular financing scheme are countered by stating that strategic investments represent a small share of revenue and are transparent in financial reporting [17][18][19] - **AI Startups Financial Viability**: Concerns about AI startups losing money are acknowledged, but NVIDIA emphasizes that these companies are valued based on future growth potential rather than current earnings, with significant market opportunities anticipated [21][23] - **Historical Accounting Fraud Comparisons**: NVIDIA refutes claims that its practices resemble historical accounting frauds, asserting that its business model is sound and transparent, with no use of special purpose entities to hide debt [24][25][26][27] - **Gross Margin Decline**: A slight decline in gross margin is attributed to channel incentives and warranty reserves, with the company clarifying that margins increased sequentially [29][30] - **Depreciation Practices**: Claims that NVIDIA's depreciation is understated are countered by stating that useful life estimates are consistent with peers, and older GPUs continue to generate strong margins [31][32][33] - **Insider Selling**: Claims of significant insider selling are dismissed, clarifying that the individuals mentioned are not insiders and their investment decisions are not commented on [34][36] - **Bitcoin Price and AI Sector Stress**: The correlation between Bitcoin prices and NVIDIA's stock is noted, but NVIDIA denies any connection between cryptocurrency prices and its accounting practices [37][38] - **Threats to Hyperscale Data Centers**: Concerns about decentralized alternatives to hyperscale data centers are addressed by emphasizing the advantages of large-scale operations in AI performance and efficiency [39][40] - **SEC Scrutiny**: NVIDIA states that it is not aware of any SEC investigations and maintains compliance with all regulations [41] Other Important Points - **Future Revenue Guidance**: NVIDIA anticipates a sequential revenue increase of $8 billion to $65 billion, indicating confidence in demand despite inventory growth [12] - **Warranty Costs**: Warranty costs have increased due to the complexity of new products, which is properly accounted for in financial statements [35] - **Market Potential of AI**: The potential economic impact of AI is highlighted, with estimates suggesting a significant market opportunity that justifies high valuations for AI startups [23] This summary encapsulates the key points discussed in the conference call, providing insights into NVIDIA's financial health, market strategies, and responses to various claims and concerns.

Flash | 25 Nov 2025 09:39:29 ET │ 10 pages NVIDIA Corp (NVDA.O) Updated Thoughts on TPU Competition CITI'S TAKE Bloomberg reported today that Meta is in talks to use Google's TPU chips in 2027 following prior news that Meta might use Google's GCP cloud services next year (see our note from early September). We believe Custom AI accelerators, including the Google TPU and Amazon Trainium, continue to scale as these platforms increasingly support both internal AI workloads and external customers. Custom XPU ex ...

Unicorns and Cockroaches: Blessed Fraud 独⻆兽与蟑螂：神圣骗局 Part 2 of The Heretic's Guide to AI's Stars 《异端 AI 指南》第⼆部分 NOV 26, 2025 2025 年 11 ⽉ 26 ⽇ ∙ PAID ∙付费内容 MICHAEL BURRY 迈克尔·伯⾥ 分享 Perusing the 10-Ks of the companies that are making and spending the most money is always fun. 翻阅那些烧钱最多的公司的 10-K 年报总是件趣事。 In the last stages of a proper mania, where everyone fawns lovingly over new spending, perusing 10-Ks can seem a pointless old-world endeavor, one or more steps below brushing youngsters off my lawn. Yet, it is w ...

Core Insights - SoftBank's recent divestment from Nvidia, selling its entire stake for approximately $5.8 billion, raises questions among investors about potential warning signs [2][3][10] - The decision to sell is primarily driven by SoftBank's strategic shift towards investing in OpenAI, committing around $30 billion to support AI model development and infrastructure projects [4][6] - SoftBank's exit from Nvidia allows it to reallocate funds towards younger ventures and capitalize on Nvidia's significant stock appreciation over the past five years, where it has increased by over 1,280% [7][10] Investment Strategy - Investors are advised to consider their own goals and risk tolerance rather than directly following SoftBank's investment moves, as their objectives may differ significantly [9][10] - The sale of Nvidia shares, while substantial, represents a small fraction of SoftBank's overall valuation of over $4.3 trillion, suggesting that it should not be interpreted as a definitive market signal [10] - Investors holding Nvidia shares may want to reassess their positions, especially in light of Nvidia's recent earnings report showing a 62% increase in total revenue and a 66% increase in data center revenue [11] Investment Approach - A dollar-cost averaging strategy is recommended for those interested in investing in Nvidia, allowing for consistent investment regardless of market fluctuations [12][14] - Nvidia is recognized as a strong company with long-term potential, and investors should remain focused on its fundamentals rather than short-term price movements [15]

Core Insights - Google is considering selling its tensor processing units (TPUs) directly, marking a shift in its AI hardware strategy [1][11] - The company aims to capture a portion of Nvidia's market share, with discussions of potential multi-billion-dollar deals with customers like Meta Platforms [4][5] - Google's TPUs are designed for efficiency, being application-specific integrated circuits (ASICs), which could appeal to companies building large AI data centers [2][6] Company Strategy - Google's TPUs were initially developed to enhance its own services and later offered to Google Cloud customers for AI workloads [2] - The latest Ironwood TPUs are reported to be twice as power-efficient as previous models and 30 times more efficient than the first TPUs released in 2018 [6] - Google Cloud executives see an opportunity to capture 10% of Nvidia's annual revenue, translating to billions in new revenue [5] Market Competition - Nvidia currently dominates the AI accelerator market, but faces competition from tech giants like Google, Amazon, and Microsoft, which are developing their own AI chips [3] - The competition is expected to gradually erode Nvidia's market dominance, with AMD also making inroads [8] - Google's TPUs present a long-term risk to Nvidia, as they could attract customers prioritizing energy efficiency over raw performance [10][11] Customer Adoption Challenges - The different architecture of Google's TPUs compared to Nvidia's GPUs may pose challenges for customers already invested in Nvidia's ecosystem [7] - Large tech companies like Meta have the resources to transition to TPUs if the benefits justify the switch [7] - Despite potential threats, Nvidia's current cloud GPUs are sold out, indicating continued strong demand for its products [9]

Nvidia Corporation (NASDAQ: NVDA ) released its third quarter FY 2025 earnings on Nov. 19, 2025, and the results exceeded expectations . Revenue came in at $57 billion, outpacing estimates by 3.48%, while diluted EPS cameRick is a Wall Street Journal best-selling author with over 20 years of experience trading stocks and options. The most authoritative publications, including Good Morning America, Washington Post, Yahoo Finance, MSN, Business Insider, NBC, FOX, CBS, and ABC News, cover his work. His passion ...

Core Insights - Nvidia Corporation reported third quarter FY 2025 earnings on November 19, 2025, with results surpassing expectations [1] - Revenue reached $57 billion, exceeding estimates by 3.48% [1] - Diluted EPS figures were also reported, indicating strong financial performance [1]