在前几天野村证券出了一份Global AI Trend Tracker 的分析，但报告中讲到的关于AI网络的内容和数 据很有意思。报告中对博通的Tomahawk 6进行了分析，也提到了Ethernet和InfiniBand的对比，其实 在大型数据中心里面InfiniBand的市场占有率早就超过了Ethernet，而且优势越来越大，只是之前并没 有单独去查过这个数据，InfiniBand当然就是英伟达，Ethernet则是以博通为代表的其他大厂的方案。 原始报告已经放到星球中。 Broadcom Tomahawk 6 Broadcom于2025年6月3日宣布开始出货Tomahawk 6交换芯片，这一消息在全球AI基础设施市场引起 了广泛关注。TH6采用3纳米工艺技术，配备200G SerDes（串行解串器），支持高达102.4Tbps的交 换容量，是当前主流以太网交换芯片（51.2Tbps）的两倍。这一技术突破不仅显著提升了网络传输带 宽和降低了延迟，还通过支持共封装光学（Co-packaged Optics, CPO）技术，实现了功耗、延迟、链 路波动和总体拥有成本（TCO）的全面优化。CPO技术通过将光学引 ...

Artificial intelligence (AI) continues to promise great returns for long-term investors in companies putting it to use in their businesses. Every industry is exploring how this technology can boost efficiency and innovation. This is expected to boost the global economy by trillions of dollars in the coming years.Here are two top tech stocks to buy that benefit from growing investment in AI services and hardware. 1. OracleOracle (ORCL -2.73%) stock recently surged to a new high following an impressive earnin ...

While some volatility has reentered the market due to rising hostilities in the Middle East, this can still be a good time to invest in some growth stocks for the long term.Let's look at 10 monster growth stocks to buy for the long haul.1. NvidiaNvidia (NVDA -1.14%) has been one of the market's biggest winners, as its graphics processing units (GPUs) have become the backbone of artificial intelligence (AI) infrastructure. That should continue as its CUDA software program has created a wide moat, helping it ...

Core Viewpoint - Artificial intelligence remains a leading investment theme in 2023, with several AI stocks showing significant growth potential, though some may be overvalued [1] Company Summaries - **Palantir**: Palantir is experiencing explosive growth with a strong client base but trades at a high valuation of 110 times sales, with only 39% revenue growth in the last quarter, suggesting it may not be a good investment [3] - **SoundHound AI**: SoundHound AI is a leader in audio recognition with a revenue increase of 151% in Q1 and projected growth of 97% for 2025, trading at 34 times sales, making it an attractive investment option [4][6] - **Nvidia**: Nvidia holds a 90% market share in data center GPUs, with a revenue increase of 69% year over year in Q1 and projected growth of 50% in Q2, positioning it as a top AI stock to buy [7][8] - **Taiwan Semiconductor Manufacturing (TSMC)**: TSMC is the leading contract chip manufacturer, expecting AI-related revenue to grow at a 45% CAGR and overall revenue at nearly 20% CAGR over the next five years, making it a strong buy [9][10] - **Alphabet**: Alphabet, the parent company of Google, is a key player in the AI space, trading at 18.5 times forward earnings, which is cheaper than the S&P 500, indicating long-term stability [11][12][13] - **Amazon**: Amazon's AWS segment accounted for 63% of its operating profits in Q1, benefiting from the rise of AI workloads, making it a strong investment choice [15] - **Broadcom**: Broadcom provides essential hardware for AI data centers, with increasing demand for its products as AI usage grows, positioning it as an excellent stock to buy [16][17] - **Meta Platforms**: Meta is leveraging AI to enhance its advertising services and develop AI devices, maintaining its dominance in social media, which supports its investment case [18][19] - **Tesla**: Tesla's focus on AI for autonomous vehicles and humanoid robots could yield significant returns, emphasizing the importance of maintaining its leadership in AI [20] - **Microsoft**: Microsoft is a key competitor in the AI arms race, with a significant stake in OpenAI and a strategy to facilitate existing generative AI models, ensuring long-term success [21]

Core Viewpoint - Artificial intelligence (AI) is positioned as a significant technological advancement with substantial investment potential, still in its early stages [1] Group 1: AI Investment Opportunities - Five top AI-related stocks identified for long-term investment include Nvidia, Broadcom, Taiwan Semiconductor Manufacturing, Palantir Technologies, and GitLab, each with unique niches and growth opportunities [2] Group 2: Nvidia - Nvidia's GPUs are essential for AI infrastructure, holding over 90% market share in the GPU space, driven by increasing demand from cloud computing and tech companies [4][5] - The company's CUDA software platform enhances its competitive edge, making it the preferred choice for AI developers [4] - Continued growth in AI infrastructure spending positions Nvidia as a clear winner in the sector [6] Group 3: Broadcom - Broadcom focuses on networking components and custom AI chip design, with AI networking revenue increasing by 70%, accounting for 40% of total AI revenue [8] - The potential market for custom AI chips is estimated between $60 billion to $90 billion, with significant demand expected from top customers [9] - Transitioning VMWare customers to subscription models and its VMware Cloud Foundation platform further strengthens Broadcom's position [10] Group 4: Taiwan Semiconductor Manufacturing - Taiwan Semiconductor Manufacturing is a leader in advanced semiconductor manufacturing, producing AI chips for major clients like Nvidia and Apple [12] - High-performance computing now constitutes 59% of its revenue, up from 46% a year ago, indicating strong growth driven by AI [13] - TSMC's advanced-node capacity and pricing power position it favorably in the AI infrastructure build-out [14] Group 5: Palantir Technologies - Palantir focuses on the applications and workflow layers of AI, developing an AI operating system that organizes data for various industries [16] - The company's AI Platform (AIP) has seen a 39% revenue increase last quarter, showcasing its growth potential [17] - Despite high valuation risks, Palantir's unique position in the AI space presents significant opportunities [18] Group 6: GitLab - GitLab leads in the DevSecOps space, enhancing software development productivity through AI-driven tools [20] - The company has achieved strong revenue growth of 25% to 40% over the past two years, with a 122% dollar-based net retention rate [21] - GitLab's technology is enhancing, rather than replacing, software developers, driving its growth trajectory [22]

以下文章来源于RimeData 来觅数据 ，作者来觅研究院 RimeData 来觅数据 . 全面的一级市场数据平台 导读： 2025年6月，北美ASIC两大巨头博通、Marvell相继发布财报，不约而同的表示光电芯片是下一阶段的研发重点。随着摩尔定律逐渐 逼近极限，电子芯片面临的 "功耗墙" 以及数据传输速率瓶颈愈发显著，光电芯片凭借其高速率、低功耗的特性，为提升计算能力与数据传 输效率带来新希望。光电芯片优势在哪？AI时代如何赋能算力？目前国内产业进展如何？赛道内投融趋势有无变化？本文尝试分析和探讨。 01 光电芯片飞速导入推理算力 AI 算力需求呈现出爆发式增长态势。从早期简单的神经网络模型到如今参数规模庞大的 GPT 等大语言模型，模型复杂度呈指数级上升。以 GPT-3为例， 其拥有1750亿个参数，而GPT- 4.5参数量约为1万亿。与此同时，为提升模型准确性与泛化能力，训练数据集规模也在不断膨胀，如用于图像识别训练的 ImageNet 数据集，包含数百万张各类图像。新兴的 AI 应用场景，如自动驾驶需实时处理大量传感器数据以做出决策，对算力的实时性与吞吐量提出了严 苛要求。以市场规模来看，全球 AI ...

In a fast-growing segment like artificial intelligence (AI), it's not always easy to spot bargains; however, they can be found. One of the best ways to find bargains in a hot sector is to look beyond stocks' price-to-earnings (P/E) ratios and instead look at their price/earnings-to-growth (PEG) ratios, as this metric takes into consideration their earnings growth.Stocks with PEGs under 1 are generally considered undervalued, and based on this metric, five of the best values in the AI space are Advanced Micr ...

J P M O R G A N North America Equity Research 15 June 2025 Semiconductors AVGO and MRVL Continue to Dominate ASICs - Both Secure Next-Gen 2nm AI ASIC Programs; Overall Advanced Custom Chip (ASIC) Market Opportunity of $30B in CY25 and 30%+ Growth Outlook We are increasing our high-end custom ASIC chip market outlook to $30B (from our prior outlook of ~$25B ) growing at a 30% CAGR driven by the continued aggressive adoption of AI (training and inference), which is driving accelerating demand for custom chip ...

点击蓝字，关注我们 摩根大通指出，随着人工智能(AI)的兴起，共封装光学技术(CPO)技术及相关公司正面临新 的发展机遇。 摩根大通指出，随着人工智能(AI)的兴起，共封装光学技术(CPO)技术及相关公司正面临新的发展 机遇。CPO是一种将光学器件与硅芯片在同一封装基板上进行异构集成的先进技术，旨在解决下 一代在带宽、功耗和成本方面的挑战。 以Samik Chatterjee为首的分析师团队表示，CPO多年来一直是行业讨论的热点。虽然该技术潜在 优势显著，但由于存在技术门槛高、市场需求有限(例如现有技术足以应对当前应用)等问题，许 多专家曾预测该技术商业化仍需较长时间。 分析师认为，CPO的采用速度可能会比最初预期更快。越来越多的供应商正在加大对CPO的投资 力度。然而，尽管投资者对行业剧变有所担忧，分析师指出目前光学供应链中的关键厂商仍将在 CPO发展中发挥重要作用，因此市场不会遭遇严重扰动。 分析师指出， 将从CPO中受益的公司包括：博通(AVGO.US)、Coherent(COHR.US)、康宁 (GLW.US) 、 Fabrinet(FN.US) 、 Lumentum(LITE.US) 、 迈 威 尔 ...

不过，分析师指出，随着人工智能的发展及相关基础设施的部署，行业对CPO的态度已发生转变。现有 技术在高数据速率与功耗之间难以平衡的问题日益凸显，CPO因此成为焦点。 分析师指出，CPO相较于传统互联方式具有明显优势，尤其在性能和功耗方面表现更优，而传统技术正 逐步接近物理极限。尽管CPO具备上述优势，分析师也指出其仍面临如散热管理、可靠性和可维护性方 面的挑战。然而，在这些领域的持续进展，以及业界对传统方案物理极限的关注，正推动行业转向 CPO。分析师表示，业内专家预测CPO市场将在2027年开始显著增长，2028年市场规模将超过10亿美 元，并在2030年突破50亿美元。 智通财经APP获悉，摩根大通指出，随着人工智能(AI)的兴起，共封装光学技术(CPO)技术及相关公司 正面临新的发展机遇。CPO是一种将光学器件与硅芯片在同一封装基板上进行异构集成的先进技术，旨 在解决下一代在带宽、功耗和成本方面的挑战。 分析师认为，CPO的采用速度可能会比最初预期更快。越来越多的供应商正在加大对CPO的投资力度。 然而，尽管投资者对行业剧变有所担忧，分析师指出目前光学供应链中的关键厂商仍将在CPO发展中发 挥重要作用， ...