日前，英伟达一下子解密了六颗芯片，引起了全球轰动。但其实早在去年年底，就有一则重磅消息在 AI芯片圈炸响：推理芯片初创公司 Groq 宣布，已与英伟达达成一项"非独家许可协议"。公告只有寥寥 数语，但随之而来的信息却迅速改变了这笔交易的分量——Groq 创始人兼 CEO Jonathan Ross、总裁 Sunny Madra 以及多名核心成员，将一并加入英伟达，参与授权技术的推进与规模化。 如果只看形式，这并不是一次收购；如果只看结果，它却几乎具备了收购的全部要素。技术被许可，团 队被吸纳，关键人物离场，Groq 虽然名义上继续运营，但其最具决定性的资产——技术路线与灵魂人 物——已然转移。这是一种典型的"收购式招聘"，也是英伟达近年来愈发娴熟的一种操作方式：在不触 碰监管红线的前提下，把潜在威胁纳入自己的体系之中。 更重要的是，这一步发生在一个极其敏感的时间点。AI 芯片的竞争，正在从"训练为王"转向"推理决 胜"。英伟达的 GPU 依旧牢牢统治着训练市场，但在推理端，AMD、定制 ASIC、云厂商自研芯片正在 快速逼近，成本与供应链多元化成为大客户最现实的诉求。Groq 的 LPU 正是为推理而生，主打 ...

公众号记得加星标⭐️，第一时间看推送不会错过。 日前，英伟达一下子解密了六颗芯片，引起了全球轰动。但其实早在去年年底，就有一则重 磅消息在AI芯片圈炸响：推理芯片初创公司 Groq 宣布，已与英伟达达成一项"非独家许可协 议"。公告只有寥寥数语，但随之而来的信息却迅速改变了这笔交易的分量——Groq 创始人 兼 CEO Jonathan Ross、总裁 Sunny Madra 以及多名核心成员，将一并加入英伟达，参与 授权技术的推进与规模化。 如果只看形式，这并不是一次收购；如果只看结果，它却几乎具备了收购的全部要素。技术被许可， 团队被吸纳，关键人物离场，Groq 虽然名义上继续运营，但其最具决定性的资产——技术路线与灵 魂人物——已然转移。这是一种典型的"收购式招聘"，也是英伟达近年来愈发娴熟的一种操作方式： 在不触碰监管红线的前提下，把潜在威胁纳入自己的体系之中。 更重要的是，这一步发生在一个极其敏感的时间点。AI 芯片的竞争，正在从"训练为王"转向"推理决 胜"。英伟达的 GPU 依旧牢牢统治着训练市场，但在推理端，AMD、定制 ASIC、云厂商自研芯片 正在快速逼近，成本与供应链多元化成为大客户最现 ...

证券研究报告 科技 英伟达吸收 Groq 定义 AI 下半场！ 华泰研究 2026 年 1 月 12 日│美国 专题研究 Groq 交易是英伟达迄今披露的最大一笔交易，规模明显高于其 2019 年以 69 亿美元收购 Mellanox。我们认为，Groq 所掌握的低时延推理核心 IP 在 战略层面的重要性，已与当年 Mellanox 的互连与网络技术处于同一量级。 该交易进一步凸显英伟达对确定性、Batch Size = 1 推理的前瞻性布局，契 合行业向 Agentic AI 演进的整体趋势。通过将 Groq 的确定性"反射式引擎" 深度整合至 CUDA 与 GPU 技术栈，英伟达正加速推动 Agentic 经济走向主 流，并在其已确立优势的 AI"上半场"基础上，逐步奠定低时延为核心特 征的"下半场"的技术与规则框架。 Acqui-hire 模式锁定 Groq 的 LPU 人才与核心 IP 英伟达对价约 200 亿美元获得 Groq 推理技术的授权、收购部分知识产权， 并引入 Groq 核心工程团队，包括创始人兼 CEO Jonathan Ross（原 TPU 架构师）与总裁 Sunny Madra。此次 ...

Core Viewpoint - Nvidia has agreed to acquire Groq, a high-performance AI accelerator chip design company, for $20 billion in cash, marking Nvidia's largest transaction to date, nearly tripling Groq's previous valuation of $6.9 billion within three months [3][7]. Group 1: Acquisition Details - The acquisition involves key Groq executives, including founder and CEO Jonathan Ross, joining Nvidia while Groq will continue to operate as an independent entity [4]. - Groq, founded in 2016 by former Google engineers, focuses on high-performance AI accelerator chip design, particularly for inference tasks [4][11]. - Nvidia's acquisition strategy is seen as a form of "acqui-hire," allowing the company to gain talent and technology while avoiding potential regulatory hurdles associated with traditional acquisitions [4][8]. Group 2: Financial Implications - Nvidia's offer includes generous compensation for Groq's shareholders, with approximately 85% of the payment made in cash upfront, and the remaining distributed over the next few years [9]. - Groq employees, approximately 600, will receive substantial financial incentives, with potential equity values estimated at $5 million per employee [4][9]. Group 3: Strategic Significance - The acquisition is viewed as a strategic move to strengthen Nvidia's competitive edge in the GPU market, especially as AI model focus shifts from training to inference, where traditional GPUs face limitations [4][12]. - Nvidia's purchase of Groq is compared to Microsoft's acquisition of GitHub, emphasizing its strategic importance in the AI landscape [11]. - The deal is expected to lock in customers, as AI labs now face the choice of either purchasing Nvidia GPUs or adopting Groq's LPU technology, thereby consolidating Nvidia's market position [12]. Group 4: Industry Trends - The AI chip market is evolving, with a clear divide between GPU-centric and non-GPU architectures, as companies like Google and Groq push for alternatives to traditional GPUs [14]. - The global AI chip market is projected to reach $413.8 billion by 2030, with non-GPU architectures expected to capture over 21% of the market share [15]. - In China, the trend towards non-GPU solutions is accelerating, with the market for non-GPU accelerated servers expected to approach 50% by 2029 [16].

Core Viewpoint - Nvidia's investment of $20 billion in acquiring Groq's Language Processing Unit (LPU) technology highlights the rising importance of SRAM in the AI, server, and high-performance computing (HPC) sectors, shifting the focus from mere capacity to speed, latency, and energy consumption [1][5]. Group 1: SRAM and HBM Comparison - SRAM (Static Random Access Memory) is characterized by high speed and low latency, commonly used within CPUs, GPUs, and AI chips. It is volatile, meaning data is lost when power is off, and it does not require refreshing, making it suitable for immediate data processing [3][4]. - HBM (High Bandwidth Memory) is an advanced type of DRAM that utilizes 3D stacking and through-silicon vias (TSV) to connect multiple memory layers to logic chips, offering high bandwidth (up to several TB/s) and lower power consumption compared to traditional DRAM, but with higher costs and complexity [4][6]. Group 2: Shift in Market Demand - The focus in AI development has shifted from computational power to real-time inference capabilities, driven by applications such as voice assistants, translation, customer service, and autonomous systems, where high latency is a critical concern [6]. - Nvidia's acquisition of Groq's technology is not just about enhancing AI accelerator capabilities but is fundamentally linked to SRAM's strengths in providing extremely low-latency memory access, which is essential for real-time AI applications [5][6].

Core Viewpoint - Nvidia has made significant acquisitions in a short period, spending over $20 billion to acquire Groq and AI21 Labs, aiming to strengthen its position in the AI market and counter competition from companies like Google and Broadcom [1][2][27]. Group 1: Acquisitions and Investments - Nvidia's recent acquisitions include Groq, which was acquired for $20 billion, and AI21 Labs, estimated to cost between $2-3 billion, along with the acquisition of Enfabrica for $900 million [2][3][21]. - The acquisition of Groq not only brought in the LPU technology but also 90% of Groq's employees, enhancing Nvidia's talent pool [6][23]. - AI21 Labs, valued at $1.4 billion, is a hub for top AI PhDs, further bolstering Nvidia's capabilities in AI architecture [7][10]. Group 2: Market Position and Strategy - Nvidia holds over 90% of the AI training market share, but the inference market is becoming increasingly fragmented, with custom ASIC chips capturing 37% of the deployment share [4]. - The company aims to address this fragmentation by acquiring talent and technology, positioning itself to compete effectively against Google’s TPU and other competitors [5][27]. - The combination of Groq's LPU and AI21's Jamba architecture is expected to enhance Nvidia's inference capabilities, allowing for significant improvements in processing efficiency [16][26]. Group 3: Talent Acquisition and Technology Integration - Nvidia's strategy includes not just acquiring companies but also securing their talent, as seen with the recruitment of 200 top AI PhDs from AI21 Labs [12][17]. - The Jamba architecture from AI21 is particularly suited for memory-constrained inference chips, which aligns with Nvidia's needs in the evolving AI landscape [16][28]. - The integration of these acquisitions is designed to create a closed loop of hardware, network, and architecture, solidifying Nvidia's competitive edge in the AI market [26].

Investment Rating - The report maintains a rating of "Outperform" for the industry, indicating an expected performance that exceeds the market average [2]. Core Insights - The report emphasizes the dominance of AI inference in driving innovation, particularly in areas related to operational expenditure (Opex) such as storage, power, ASIC, and supernodes [31]. - The acquisition of 35% of Limin Da by Lingyi Zhi Zao for 875 million RMB is highlighted, positioning the company to leverage advanced thermal management technologies in the AI sector [25]. - The report identifies significant growth opportunities in the domestic storage industry, particularly with the anticipated expansion of NAND and DRAM production in the coming year [32]. Summary by Sections Market Review - The Shanghai Composite Index rose by 1.88%, while the Shenzhen Component Index increased by 3.53%, and the ChiNext Index saw a rise of 3.9%. The Shenwan Electronics Index increased by 4.96%, ranking 4th among 31 sectors, with a year-to-date increase of 48.12% [12][18]. Weekly Focus - Lingyi Zhi Zao's acquisition of Limin Da is noted for its strategic alignment with AI computing and thermal management solutions [25]. - NVIDIA's non-exclusive licensing agreement with Groq is discussed, highlighting its potential to enhance NVIDIA's position in high-performance computing and AI chips [26]. Weekly Insights - The report forecasts a significant increase in demand for storage solutions driven by advancements in products from Yangtze Memory Technologies and Changxin Memory Technologies, suggesting a focus on the domestic storage supply chain [31]. - The report also highlights the importance of power supply innovations, recommending attention to both generation and consumption technologies [33]. - ASIC technology is expected to gain market share, with a focus on key domestic and international cloud service providers [33]. - The report anticipates growth in supernode technologies, including high-speed interconnects and liquid cooling solutions [33].

传媒 2025 年 12 月 31 日 2026 海外 AI 前瞻：模型和算力 投资要点： 一、模型端：Gemini VS OpenAI VS Claude 行 业 动 态 跟 近期谷歌发布 Gemini 3 Pro、Nana Banana Pro 以及 Deepthink 后， 市场关注较高，对 OpenAI 和 Claude 都产生了一定的影响。OpenAI 后来推出了 GPT 5.2 和 Image1.5 进行竞争，而 Claude 发布了 Opus 4.5 进行竞争，三家竞争有助于大模型能力的提升。 行 业 研 究 二、算力篇：英伟达 VS 谷歌 TPU 踪 根据腾讯科技，Gemini 3 通过谷歌自研的 TPU 进行训练，TPUv7 则沿用 3D 环面（3D Torus）架构，在实际场景中展现出更优的 TCO 表 现。同时根据新智元，谷歌正在推进一项代号为 TorchTPU 的战略行 动，核心是让全球最主流的 AI 框架 PyTorch 在自家 TPU 芯片上跑 得更顺畅。这将进一步拓展谷歌 TPU 的使用场景，进而带动其与英伟 达 GPU 的竞争。 三、产能篇：台积电 VS 三星 12 月 24 ...

Core Viewpoint - Nvidia's recent $20 billion non-exclusive licensing agreement with Gro for inference technology is a strategic move to enhance its capabilities in the AI race, particularly in the area of inference, which is becoming increasingly important in the industry [1][5][10] Group 1: Deal Structure and Implications - The deal is framed as a strategic acquisition but is more accurately described as an architectural admission, allowing Nvidia to capture Grock's capabilities without the regulatory challenges of a full acquisition [3][6] - Nvidia's focus on inference economics is crucial, as inference costs are significantly higher than training costs, with inference being 15 times more expensive over a model's lifetime [4][5] - The licensing agreement allows Nvidia to integrate necessary technology while navigating regulatory constraints, as they cannot acquire every competitor outright [8][12] Group 2: Competitive Landscape - The deal is seen as a defensive strategy to maintain Nvidia's competitive edge against companies like Google, which has a vertically integrated approach with its TPU chip [7][8] - Nvidia's move to secure Grock's technology is a reaction to the growing dominance of inference in AI applications, indicating that the company is adapting to market changes [6][9] - Concerns have been raised about the potential anti-competitive nature of such deals, as they may limit competition and hinder the growth of startups like Grock [10][12] Group 3: Market Reaction and Future Outlook - Nvidia's stock performance reflects the boom in training models, but the new deal positions the company favorably for the anticipated growth in inference applications [9][10] - The deal is viewed as a protective measure for Nvidia's future, ensuring that the company remains relevant as the focus shifts from training to inference in AI technologies [10]

Core Viewpoint - Nvidia's acquisition of Groq for $20 billion is not just about technology but also involves significant compensation for Groq's employees and shareholders, effectively a "talent acquisition" strategy [2][10][19]. Group 1: Acquisition Details - Nvidia's acquisition includes not only technology rights but also a commitment to Groq's employees and shareholders, with a valuation that has tripled from previous estimates [3][19]. - 90% of Groq's team will be integrated into Nvidia, with each employee receiving an average of $5 million [4][20]. - Groq will continue to operate as an independent entity, with its cloud service platform GroqCloud remaining active [8]. Group 2: Employee and Shareholder Compensation - Employees will receive cash for vested shares and Nvidia stock for unvested shares, with a significant portion of the compensation being accelerated [11][12]. - Employees who have been with Groq for less than a year will still receive some compensation, as Nvidia waived the typical vesting cliff [15][16]. - Shareholders, including major investors like Disruptive and Blackstone, will receive dividends based on the $20 billion valuation [17][19]. Group 3: Market Context and Implications - The acquisition reflects a broader trend where companies prefer "acquisition-style hiring" to avoid antitrust scrutiny while gaining access to key technologies and talent [21][22]. - Nvidia's financial strength, with $60.6 billion in cash and short-term investments, enables such large-scale acquisitions [32]. - The deal signifies Nvidia's recognition of the need to adapt to changing AI technology landscapes, particularly in inference capabilities [44][45].