Bank of America’s Vivek Arya summed up the 2026 setup in a single line: Nvidia is acting like it believes that and more. The chipmaker didn't want to wait for the next phase of the AI buildout to start, so it made a $20 billion deal with inference chip firm Groq. The agreement appears like an acquisition in terms of its economic impact, but it's not one on paper. Nvidia’s Groq agreement looks less like a partnership and more like a strategic capture.Photo by Woohae Cho on Getty Images Nvidia gives Groq ...

证券研究报告：宏观报告 发布时间：2025-12-29 研究所 分析师:李起 SAC 登记编号:S1340524110001 Email:liqi2@cnpsec.com 研究助理:高晓洁 SAC 登记编号:S1340124020001 Email:gaoxiaojie@cnpsec.com 近期研究报告 《工业企业利润延续放缓，结构性亮点 凸显》 - 2025.12.29 宏观观点 海外宏观周报：假期成交清淡，市场趋于平静 ⚫ 核心观点： | 1 海外宏观教据跟踪 | | --- | | 1.1 上周重要事件与数据跟踪 | | 1.2 本周重要数据与事件 | | 2 美联储观点跟踪 6 | | 3 风险提示 7 | 过去一年中，美元整体显著走弱。美元指数自年初高点约 110 持 续回落，跌至 98 附近。其背后的驱动因素是多方面的：一方面，美 国财政与贸易政策的不确定性上升，市场对财政可持续性的担忧加 剧，削弱了美元资产的长期吸引力；另一方面，美联储全年降息幅度 高于其他主要经济体，导致美国与主要发达经济体之间的利差明显收 窄。 展望后市，短期内，美元仍具备一定的结构性支撑。若美联储如 市场当前定价所隐含的 ...

Jay 发自 凹非寺 量子位 | 公众号 QbitAI 更多英伟达收购「TPU之父」创业公司Groq的细节曝光了—— 主打一个 黄仁勋BOSS直聘 。 Axios消息，老黄斥200亿美元买的不止是技术授权，还顺手给Groq的员工和股东也一起安顿了。 股东那边，给分红。按200亿美元的估值标准，比之前翻了约三倍。 员工这边，直接给Groq挖空， 90%的团队全给打包带走 ，人均到手500万美元。 啧啧，这哪是「人才」收购啊……恐怕连 人心 也一并带走了。 收购式招聘，Groq从上到下打包带走 平安夜老黄没有休息，一项200亿美元的创纪录芯片收购，轰动硅谷。 按照协议，Groq创始人兼CEO Jona t han Ross 、总裁 Sunny Madra 以及多名核心高管将加入英伟达。 但Groq并不会消失，公司将继续以独立实体运营，由原CFO Simon Edwards 出任CEO，云服务平台 GroqCloud 也照常对外提供服 务。 好一手「人才收购」，该拿的一样都没少，还规避了反垄断审查。 不过，Groq创始人和高管全跑路了，员工和股东咋办呢？ 敢情老黄早早就考虑到了，一句话总结： 人人都 有份 ，都不白 ...

GPU。 作者 / 梦晨 来源 / 量子位 （ID：QbitAI） 英伟达方面也明确表态：我们不是在收购Gr o q这家公司，我们只是获得技术授权，并将Gr o q的产品整合到未来的产品中。 看起来老黄也学会了"人才收购"这招：重金掏空一家公司的人才和核心资产，但又避免触发反垄断。 所以这2 0 0亿美元到底买了什么？ 席卷硅谷的"人才收购" 答案是 ：技术授权，加上一整支核心团队。 平安夜老黄没有休息，一项2 0 0亿美元创纪录芯片收购消息，轰动硅谷。 英伟达官宣 ：以2 0 0亿美元现金与AI芯片初创公司Gr o q达成交易。 消息一出迅速引发市场轰动，因为这是英伟达有史以来最大规模的一笔交易，远超2 0 1 9年收购Me l l a n o x的7 0亿美元。 但仅仅几小时后，画风突变。 英伟达和Gr o q双双发表声明，对交易性质进行了澄清，并非收购。Gr o q在官方博客中写道：我们与英伟达签订了一份非排他性技术 许可协议。 最先爆料的是Gr o q主要投资方Disr u p t i v e，其CEO透露英伟达已同意以2 0 0亿美元现金收购Gr o q的资产，交易进展非常迅速。 自2 0 1 6年 ...

此案由主导Groq最近一轮融资的创投机构Disruptive执行长Alex Davis向外透露，并指出双方洽谈时间相 当短。此时距Groq于9月以约69亿美元估值完成7.5亿美元募资，仅约3个月。当时投资人包括黑石、 Neuberger Berman，以及Samsung Electronics、思科等企业。 根据Groq公告，该公司已与英伟达签署一项非独家技术授权协议，授权其AI推论相关技术，但未揭露 具体金额。随交易完成，Groq创办人兼执行长罗斯、总裁Sunny Madra，以及其他资深领导人将加入英 伟达，协助扩大相关技术的应用与规模。 据媒体报道，英伟达（Nvidia）同意以现金200亿美元，取得新创公司Groq的部分资产，将透过非独家 技术授权取得Groq的推论芯片技术，同时延揽该公司创办人、前Google芯片工程师Jonathan Ross在内的 多名核心高层加入。这成为英伟达历来金额最高的一笔资产与技术交易。 英伟达执行长黄仁勋在致员工信中指出，英伟达计划将Groq的低延迟处理器整合至英伟达AI工厂架 构，以支援更广泛的即时与推论工作负载。他并强调，英伟达并未收购Groq整家公司，而是引进人才 ...

AI（人工智能）芯片龙头英伟达拿下潜在竞争对手、芯片初创公司Groq的核心班底。 此外，分析指出，美国科技巨头正在投入巨资，通过各种类型的许可协议聘请顶尖AI人才，从而降低 触发反垄断审查的可能性。 据外媒报道，罗斯曾表示，Groq设计带有嵌入式内存，这意味着其芯片的生产和部署速度比GPU更 快，功耗也更低，而GPU通常能耗很高，且对于训练模型更为必要。 英伟达方面表示，Groq的低延迟芯片对输入的响应速度极快，将为英伟达的产品带来新的能力，帮助 其开拓新的市场领域。 今年7月，谷歌以24亿美元引进AI编程公司Windsurf的人才和技术。今年6月，Meta以143亿美元的价格 收购了AI数据标注公司Scale AI 49%的股份，并让后者的创始人兼CEO亚历山大·王（Alexandr Wang） 加入了Meta，领导Meta的AI部门。苹果、微软和亚马逊也都在近两年进行过类似的交易。 在一封内部邮件中，英伟达CEO黄仁勋写道："我们计划将Groq的低延迟处理器整合到英伟达的AI工厂 架构中，以服务更广泛的AI推理和实时工作负载......虽然我们正在吸纳Groq的优秀人才并获得其知识产 权许可，但我们并没有 ...

但仅仅几小时后，画风突变。 英伟达和Groq双双发表声明，对交易性质进行了澄清，并非收购。Groq在官方博客中写道： 转自：量子位 | 公众号 QbitAI 平安夜老黄没有休息，一项 200亿美元 创纪录芯片收购消息，轰动硅谷。 英伟达官宣：以200亿美元现金与AI芯片初创公司Groq达成交易。 消息一出迅速引发市场轰动，因为这是英伟达有史以来最大规模的一笔交易， 远超2019年收购Mellanox的70亿美元 。 我们与英伟达签订了一份非排他性技术许可协议。 英伟达方面也明确表态： 我们不是在收购Groq这家公司，我们只是获得技术授权，并将Groq的产品整合到未来的产品中。 看起来老黄也学会了"人才收购"这招：重金掏空一家公司的人才和核心资产，但又避免触发反垄断。 所以这200亿美元到底买了什么？ 席卷硅谷的"人才收购" 答案是：技术授权，加上一整支核心团队。 最先爆料的是Groq主要投资方Disruptive，其CEO透露英伟达已同意以200亿美元现金收购Groq的资产，交易进展非常迅速。 自2016年Groq成立以来，Disruptive已经往这家公司砸了超过5亿美元。 根据交易协议，Groq 创始人兼 ...

Core Viewpoint - Groq has entered a non-exclusive licensing agreement with Nvidia for its inference technology, aiming to enhance high-performance AI inference globally, with the deal valued at $20.6 billion in cash [1][4]. Group 1: Agreement Details - The licensing agreement focuses on expanding access to high-performance, low-cost inference capabilities, with key Groq executives joining Nvidia to advance the technology [2][6]. - Groq will remain an independent company, with Simon Edwards appointed as the new CEO, while GroqCloud services will continue to operate without disruption [3][6]. Group 2: Financial Context - The deal follows Groq's recent funding rounds, including a $750 million raise earlier this year, which significantly increased the company's valuation amid the AI boom [5]. - Groq has received over $500 million in investments since its founding in 2016, indicating strong investor confidence [4][6]. Group 3: Strategic Implications - The collaboration combines Groq's specialized inference speed with Nvidia's extensive ecosystem, positioning both companies to leverage their strengths in the AI hardware market [5][6]. - The agreement reflects a mutual focus on expanding access to efficient inference solutions, highlighting the growing demand in the AI sector [6].

Core Insights - IBM has partnered with Groq to integrate Groq's inference technology, GroqCloud, into its watsonx Orchestrate platform, enhancing AI model performance and cost efficiency [1][8] - The collaboration aims to address challenges in speed, cost, and reliability of AI adoption across various sectors, including healthcare, finance, government, retail, and manufacturing [3][8] - The Agentic AI market is projected to grow at a compound annual growth rate of 43.61% from 2025 to 2030, presenting a significant opportunity for IBM's watsonx platform [4] Company Developments - IBM's integration of Groq's Language Processing Units (LPUs) is expected to provide over five times faster and more cost-efficient inference compared to traditional GPU systems [1] - The partnership will enable IBM watsonx clients to leverage high-speed inference, facilitating the deployment of agentic AI for various applications such as customer care and productivity enhancement [2][8] - IBM plans to enhance its Red Hat open source vLLM technology and Granite models through the integration of Groq's LPU architecture [3] Market Context - Other tech firms, such as ServiceNow and Microsoft, are also focusing on AI deployment, with ServiceNow enhancing its AI platform and Microsoft developing Azure AI Foundry [5][6] - IBM's stock has increased by 23% over the past year, while the industry has seen a growth of 48.6% [7] - The Zacks Consensus Estimate for IBM's earnings for 2025 has risen over the past 30 days, indicating positive market sentiment [12]

Group 1 - The core viewpoint of the news highlights the performance of the Sci-Tech Innovation Artificial Intelligence ETF (589010), which opened lower and is currently trading at 1.399 yuan, down 1.41%, with active trading volume of approximately 9.4 million yuan [1] - Among the 30 constituent stocks, only 4 are up while 26 are down, indicating a bearish trend in the short term, with notable declines in stocks like Haitai Ruisheng and Jingchen Technology [1] - The ETF is currently in a technical adjustment phase, but there has been significant net inflow of funds over the past five days, suggesting strong investment interest [1] Group 2 - Dongwu Securities emphasizes the scarcity of factors such as the ceiling of the AI industry, monetization potential, growth prospects, and industry chain friendliness [2] - The high-frequency iteration of AI computing power, with a "year-on-year + hardware-software synergy" approach, is expected to refresh unit computing costs within 12-18 months, creating new demand and redefining pricing before a price decline occurs [2] - The Sci-Tech Innovation Artificial Intelligence ETF closely tracks the Shanghai Stock Exchange Sci-Tech Innovation Board AI Index, covering high-quality enterprises across the entire industry chain, benefiting from high R&D investment and policy support [2]