一幅算力版图浮现 。 作者/刘博 报道/投资界PEdaily 这样一幕令人惊叹。 随着摩尔线程、沐曦相继上市，科创板短短半个月内一举诞生两个3 0 0 0亿市值I PO，创投圈为之沸腾。在此之前，寒武纪、海光信息 市值均一度突破6 0 0 0亿元。至此，中国"算力四贵"合计市值已超1 . 6万亿元。 鲜少有人注意到，梳理寒武纪、海光信息、摩尔线程、沐曦背后，意外浮现着一个共同的身影——联想创投。复盘下来，能做到独中 四元的投资机构，似乎就只有联想创投一家。 这是中国算力崛起叙事当中的一个惊喜片段。感叹于此，我们尝试去还原这段几近被忽略的投资往事。 出 手 寒 武 纪 始 于 一 个 下 午 。 投 资 界 曾 记 录 这 段 往 事 ： 2 0 1 7 年 ， 联 想 创 投 总 裁 、 创 始 合 伙 人 贺 志 强 在 中 国 科 学 院 第 一 次 见 到 陈 天 石，"话不多，典型的技术天才气质。" 经过几次交流，贺志强发现陈天石很有商业头脑，但内部投资成员还是觉得估值贵，一度犹 豫要不要把项目带到投决会。 寒武纪的价值在于，它是当时中国市场上极少数从底层指令集和芯片架构层面，为人工智能专门设计和优化 ...

据IPO早知道消息，随着日前摩尔线程和沐曦的先后登陆科创板，AI算力赛道可谓迎来一波"上市 潮"。 值得注意的， 在摩尔线程和沐曦背后，均出现了联想创投的身影，其也是为数不多同时投中这两 家"国产GPU新锐"的企业。 事实上， 早在AI算力成为风口之前，联想创投就已经开始基于CVC视角系统布局 ——2016年前 后，当多数VC还在观望，联想创投成为寒武纪A轮投资方，并连续四轮跟投，直至其上市。2021 年，国产GPU尚在艰难的概念验证期，联想创投又出现在沐曦的A轮和摩尔线程的Pre-A+轮融资名 单里；同一年，联想创投还投资了海光。 这意味着， 手握四张"AI算力王牌"的联想创投，成为唯一一家独中四元的投资机构。 这四家公司在技术路径和产品定位上形成互补，恰好对应了AI算力需求的多元化趋势。 "投资半导体周期长，很难赚钱"——这在近十年前几乎是行业共识。但联想创投选择从产业视角出 发，沿着集团"端-边-云-网-智"的技术路线图，在算力基础设施层面展开全栈布局：寒武纪主打AI 专用芯片，瞄准算法效率；海光走x86兼容路线，强调生态平稳过渡；摩尔线程做全功能GPU，覆盖 图形与计算；沐曦聚焦数据中心GPU，攻坚高 ...

Core Insights - Nvidia is experiencing unprecedented cash accumulation due to explosive growth in AI chip demand, with free cash flow projected to soar from $3.8 billion in FY2023 to $96.5 billion in the current fiscal year, marking a staggering increase [2] - The company's AI-specific chip business is witnessing a compound annual growth rate (CAGR) of 194%, contributing significantly to this cash flow surge [2] - Nvidia's revenue for Q3 FY2026 is expected to grow by 62% year-over-year, and it is projected to generate approximately $850 billion in cash over the next four years, even if actual amounts are only half of expectations [5][8] Financial Performance - Nvidia's free cash flow has seen a historic increase, unmatched by any tech company since 1990, with only Apple in 2001 coming close after the iPod launch [5] - The company’s total cash flow from FY2020 to FY2023 was $21 billion, which is significantly lower than the anticipated cash flow for the upcoming years [8] Strategic Investments - Nvidia has announced strategic investments, including $10 billion into Anthropic and a commitment of $100 billion to OpenAI, raising speculation about whether these investments are aimed at stimulating downstream demand [11] - The company is also investing in emerging cloud service providers like CoreWeave, which are key customers for its chips, and is engaging in innovative partnerships to strengthen ecosystem ties [11] Cash Management Strategy - Nvidia's CEO, Jensen Huang, indicated a focus on using cash to drive business growth, with $36 billion in stock buybacks planned for the first three quarters of FY2026, a significant increase from $10 billion in FY2023 [11] - The company is concentrating on its core business through strategic investments, contrasting with other tech giants that are diversifying into various sectors [12] Market Impact - Nvidia's increasing cash reserves and investment strategies are expected to significantly influence the global AI industry landscape, with the balance between short-term returns and long-term strategy being crucial for maintaining its leadership in the AI era [12]

Core Viewpoint - The forum highlighted the rapid development of China's artificial intelligence (AI) industry, with significant growth projections for the AI large model market, which is expected to reach 16.5 billion yuan in 2024 and 62.4 billion yuan by 2028, reflecting a compound annual growth rate of 40% [2]. Group 1: Industry Growth and Trends - The AI industry in China is experiencing robust growth, with the large model market projected to expand significantly in the coming years [2]. - The Chinese government is adopting a cautious yet inclusive approach towards the large model industry, with a surge in related policies expected from 2024 [2]. - Breakthroughs in AI-specific chips and quantum computing are anticipated to enhance computing hardware, leading to more efficient, low-power, and environmentally friendly solutions [2]. Group 2: Technological Innovations - Innovations in embodied intelligence perception and control technologies are emerging, particularly in sensor technology, which is expected to see advancements in high-resolution, low-power, and multi-modal capabilities [2]. - The development of soft robotics is poised to create new opportunities in various fields, including medical rehabilitation and disaster rescue, driven by new soft material research and applications [2]. Group 3: Challenges Facing the Industry - The AI industry faces significant challenges, including bottlenecks in data, algorithms, and computing power [3][5]. - There is a lack of high-quality professional datasets, difficulties in data sharing, and an inadequate data governance framework [5]. - Ethical and safety concerns regarding large models present global governance challenges, alongside a growing talent supply-demand imbalance due to increasing application scenarios [3][5].