通过此次上市，沐曦股份将借助资本市场的力量，加速从技术验证向大规模商业化落地的跨越。根据此 前招股书披露，所募集资金将主要投入三大项目：新型高性能通用GPU研发及产业化项目、新一代人工 智能推理GPU研发及产业化项目以及面向前沿领域及新兴应用场景的高性能GPU技术研发项目。此举旨 在推动公司GPU产品线迭代升级，并强化前沿技术的战略储备。 《2025年中国人工智能计算力发展评估报告》认为，2025年，中国智能算力规模将达1037.3EFLOPS， 增长43%，远高于通用算力增幅；中国信通院预测，到2028年，中国智能算力规模将达2782EFLOPS， 未来五年复合增长率超过40%。摩尔线程、沐曦股份等GPU企业的上市，恰逢其时地填补了这一市场真 空。 2025年12月17日，上海证券交易所科创板迎来一个历史性的时刻。坤元资产FOF生态圈核心伙伴——沐 曦股份(688802.SH)正式登陆科创板，成为A股上市的第二家国产GPU领军企业。上市首日开盘价700 元，股票价格高开568.83%，全天收涨692.95%，市值一举突破3320亿元人民币大关。按盘中最高价计 算，沐曦股份也凭借近40万元的单签盈利成为近10 ...

高文介绍，"中国算力网"研究计划的与初衷源于国家发改委的"东数西算"工程，该工程旨在将西部丰富 的能源优势与东部密集的计算需求相结合，实现全国范围内的算力资源优化配置。 高文建议要高度关注美国正在布局的"星际之门计划"、"创世纪任务"等国家级战略工程，意图建立全球 性的"算力霸权"。他表示，加快建设"中国算力网"不仅是经济发展的需要，更是保障国家在数字时代拥 有"算力主权"、避免在未来的竞争中受制于人的关键举措。 责任编辑：郝欣煜 专题：2025年大湾区交易所科技大会 11月28日至29日，深交所联合港交所、广期所，围绕"迈向人工智能+时代"主题，共同举办2025年大湾 区交易所科技大会。鹏城实验室主任、北京大学博雅讲席教授高文出席并做"中国算力网计划与鹏城脑 海大模型"主题演讲。 ...

Core Viewpoint - The AI chip market is experiencing significant shifts as Google accelerates the commercialization of its self-developed AI chip, TPU, potentially impacting NVIDIA's dominance in the GPU market [1][4]. Group 1: Google's TPU Development - Google has been developing TPU since 2013, initially for internal AI workloads and Google Cloud services, but is now pushing for external commercialization, with Meta considering deploying TPU in its data centers by 2027 [4]. - The potential contract with Meta could be worth several billion dollars, indicating a significant market opportunity for Google [4]. - Google’s strategy aligns with its long-term goal of integrating hardware and software, especially as the costs of training large models rise dramatically [4]. Group 2: NVIDIA's Market Position - NVIDIA currently holds over 90% of the AI chip market share, but faces increasing competition from companies like Google [4]. - In response to the competitive landscape, NVIDIA emphasizes its "one generation ahead" advantage and the versatility of its GPUs, which are seen as irreplaceable in current AI innovations [5]. - Despite the challenges posed by self-developed chips, NVIDIA continues to supply GPUs to Google, indicating a complex relationship between the two companies [5]. Group 3: Industry Trends - The trend towards self-developed AI chips is not limited to Google; other tech giants like AWS and Microsoft are also advancing their own chip technologies [6][7]. - The industry is moving towards a heterogeneous architecture, where companies are diversifying their chip supply strategies rather than relying solely on one type of architecture [7]. - The collaboration between companies like Anthropic with both NVIDIA and Google highlights a shift towards a multi-supplier strategy in AI infrastructure [7]. Group 4: Market Reactions - Following news of Google's TPU commercialization, NVIDIA's stock experienced significant volatility, reflecting market concerns about its future share and profitability in the AI infrastructure space [8]. - The evolving landscape suggests a transition from hardware competition to system-level competition, with changes in software frameworks and energy efficiency influencing the AI chip market [8].

Core Viewpoint - The AI chip market is experiencing significant shifts, with Google accelerating the commercialization of its self-developed AI chip, TPU, which may disrupt NVIDIA's dominance in the GPU market [2][6][10] Group 1: Google's Strategy - Google is pushing its TPU chip towards external clients, with Meta considering deploying TPU in its data centers as early as 2027, potentially involving contracts worth billions [6] - The move aligns with Google's long-term strategy of "soft and hard integration" and aims to reduce costs associated with large model training [6] - Google's latest TPU versions, including TPU v7 and Gemini 3, are designed to enhance its technological capabilities in the era of large models [6] Group 2: NVIDIA's Response - NVIDIA has responded to the competitive threat by emphasizing its leadership in the GPU market and the unique advantages of its products, claiming to be the only platform capable of running all AI models [4][7] - Despite the rise of TPU, NVIDIA maintains that its GPUs remain irreplaceable due to their versatility and compatibility across various AI applications [7] - NVIDIA's stock has been volatile in response to Google's advancements, indicating market concerns about its future share and profitability in AI infrastructure [10] Group 3: Industry Trends - The trend of major tech companies developing their own AI chips is growing, with AWS and Microsoft also advancing their proprietary chip technologies [9] - The industry is shifting from a GPU-centric model to a heterogeneous architecture involving multiple suppliers, as companies seek to diversify their computing resources [9] - The collaboration between companies like Anthropic with both NVIDIA and Google highlights a preference for a multi-route procurement strategy, indicating a move away from reliance on a single chip architecture [9]

Core Insights - The competition in artificial intelligence (AI) is shifting towards infrastructure, with unprecedented capital flowing into computing power foundations. Anthropic announced a $50 billion investment to build an AI infrastructure network across the U.S. [1] - Despite the significant investment from Anthropic, it pales in comparison to competitors like OpenAI, which plans to invest approximately $1.4 trillion over the next eight years, and Meta, which will invest $600 billion in the U.S. infrastructure and employment sectors over the next three years [1][5] - A Morgan Stanley report predicts that global investments in AI and data center infrastructure will reach $5 trillion, indicating a fierce competition for computing power supremacy [1][5] Company-Specific Developments - Anthropic, founded in 2021 by former OpenAI researchers, aims to enhance its infrastructure to support rapid business growth and long-term R&D needs. The company has seen a nearly sevenfold increase in large clients contributing over $100,000 annually [3][4] - The $50 billion investment will be executed in partnership with Fluidstack, a UK-based AI cloud platform, and is part of Anthropic's strategy to become a key player in the U.S. AI infrastructure sector [3][4] - Anthropic's previous funding round raised $13 billion, leading to a post-money valuation of approximately $183 billion [3] Industry Trends - The current investment surge in AI infrastructure mirrors the dot-com bubble of the early 2000s, characterized by overly optimistic capital flows and valuations detached from fundamentals. However, tech giants today have healthier cash flows, providing them with more room for error [6][7] - Major tech companies, including Amazon, Google, Microsoft, and Meta, have committed to substantial AI investments, with Amazon projecting a total investment of $125 billion by 2025 and Google increasing its capital expenditure to between $91 billion and $93 billion for the same year [4][5] - Concerns about sustainability and potential bubble risks are rising, particularly regarding the U.S.'s ability to meet the electricity demands of AI data centers, which could lead to a power shortfall of up to 20% by 2028 [6][7]

Core Insights - The competition in artificial intelligence (AI) is shifting towards infrastructure, with unprecedented capital flowing into computing power foundations. Anthropic announced a $50 billion investment to build a nationwide AI infrastructure network in the U.S. [1] - Despite the significant investment from Anthropic, it pales in comparison to competitors like OpenAI and Meta, which have announced plans to invest $1.4 trillion and $600 billion respectively in AI infrastructure [1][4] - A Morgan Stanley report predicts that global investment in AI and data center infrastructure could reach $5 trillion, indicating a fierce race for computing power supremacy among tech giants [1][4] Investment Details - Anthropic, founded in 2021, has raised $13 billion in its Series F funding round, with a post-money valuation of approximately $183 billion. The $50 billion infrastructure investment will be in collaboration with Fluidstack, a UK-based AI cloud platform [2] - The new data centers are expected to support Anthropic's rapid business growth and long-term R&D needs, positioning the company as a key player in the U.S. AI infrastructure sector [2][3] - Anthropic's client base has grown significantly, with over 30,000 enterprise customers, and the number of high-revenue clients has surged nearly sevenfold in the past year [3] Competitive Landscape - The investment trend in AI infrastructure is a reflection of the broader competitive landscape, with major players like OpenAI, Google, Microsoft, and Meta also committing substantial resources to AI [3][4] - Amazon plans to invest $125 billion by 2025, while Google has raised its capital expenditure forecast to between $91 billion and $93 billion for the same year [4] Concerns and Challenges - The rapid expansion of AI infrastructure raises concerns about sustainability and potential market bubbles, particularly regarding the U.S.'s ability to meet the electricity demands of these data centers [5][6] - Microsoft has highlighted a significant power shortage risk, estimating that the U.S. could face a 20% electricity shortfall by 2028 due to the high energy consumption of AI data centers [5][6] - Despite the aggressive capital expenditures, many tech companies, including OpenAI, are still operating at a loss, raising questions about the long-term viability of these investments [6]

Core Viewpoint - The GPU industry is experiencing rapid growth, with new players like Muxi emerging to challenge established companies like Cambricon. Muxi plans to go public on the Sci-Tech Innovation Board, aiming to provide autonomous computing power for China's AI industry [1][2]. Group 1: Company Overview - Muxi was founded in September 2020 in Shanghai by key former AMD employees, including Chen Weiliang, who has extensive experience in GPU production [3][4]. - The company has raised over 2 billion yuan in funding, with significant investments from various institutions, and reported revenue of 53.02 million yuan in 2023 [5]. - Muxi's revenue is projected to grow explosively, reaching approximately 743 million yuan in 2024, with a business model that includes direct sales and distribution [5]. Group 2: Market Dynamics - The GPU market is driven by the demand for computing power, particularly in AI training and scientific simulations, with the rise of generative AI further expanding market needs [2][8]. - The GPU industry has evolved through three phases, with the current phase characterized by the integration of cloud and edge computing and multi-GPU interconnect technology [9]. - Despite the dominance of global players like NVIDIA and AMD, domestic companies are narrowing the gap, with Muxi showing the highest revenue growth rate in the industry at 4074.52% from 2022 to 2024 [9][11]. Group 3: Opportunities and Challenges - The domestic GPU market presents significant opportunities for new players, particularly in areas like domestic substitution, intelligent computing center construction, and AI for Science [12]. - However, challenges include rapid algorithm iteration, high capital requirements, and the complexity of building an ecosystem that integrates chips, software, and applications [14][15].

Core Viewpoint - The global competition in artificial intelligence (AI) is increasingly focused on the physical foundation of computing power, leading to a silent war over "Compute Sovereignty" [2][3][4]. Group 1: Understanding Compute Sovereignty - Compute sovereignty is a complex issue that must be deconstructed into three levels: the location of AI computing resources, the nationality of the companies owning these data centers, and the origin of the AI accelerators (chips) powering them [2][3]. - A survey of nine leading public cloud service providers reveals a highly uneven global distribution of computing power, with only 33 countries hosting critical AI infrastructure, indicating a significant gap between "compute-rich" and "compute-poor" nations [3][4]. Group 2: Territorial Illusions and Economic Considerations - The concept of territorial sovereignty in computing power is primarily about having physical AI data centers within a country's borders, which is seen as essential for ensuring supply security and regulatory oversight [4][5]. - The report highlights that while attracting foreign tech giants to build data centers can bring economic benefits, the environmental and resource costs may outweigh these advantages, especially for countries lacking competitive energy and climate conditions [5]. Group 3: Supplier Loyalty and Geopolitical Strategies - Merely having data centers does not equate to true sovereignty; the nationality of AI cloud service providers introduces a layer of complexity due to overlapping legal jurisdictions [6][7]. - Countries face strategic choices between two approaches: "Aligning" with a single foreign superpower's digital infrastructure or "Hedging" by diversifying suppliers to mitigate risks [8][9]. Group 4: The Chip Dependency - The report identifies a critical dependency on AI accelerators, with U.S. companies like NVIDIA dominating 80% to 95% of the global market, leading to a situation where most countries rely on U.S. technology for their AI capabilities [10][11]. - Countries like the EU and China are striving for "strategic autonomy" in chip production, but achieving this is a long-term and costly endeavor [12][13]. Group 5: Conclusion on Sovereignty - The report concludes that compute sovereignty is not a straightforward goal but a complex framework filled with trade-offs, where a nation may achieve sovereignty in one area while remaining dependent in another [13].

Group 1: AI Sovereignty and Infrastructure - European countries, particularly France and the UK, face a significant shortfall in AI computing power, with Europe accounting for 20% of global AI demand but only 3%-5% of supply capacity, leading to heavy reliance on US and Chinese technology [1][3][4] - The French President emphasized the need for Europe to establish its own computing and chip manufacturing capabilities to reduce external dependencies and achieve "computing sovereignty" [3][4] - France and the UK announced plans to significantly expand their computing infrastructure, with the UK aiming for a 20-fold increase in public computing capacity by 2030 [1][4] Group 2: Talent Retention and Ecosystem Development - There is a pressing issue of talent retention in Europe, with many AI professionals being attracted to other regions; creating an environment conducive to research and innovation is crucial [1][8][9] - France is implementing measures to retain AI talent, including allowing researchers to engage in entrepreneurial activities while remaining in academia and modifying intellectual property laws to facilitate technology transfer [9][34] - The importance of a supportive ecosystem that includes collaboration between public and private sectors, as well as startups, is highlighted as essential for fostering innovation [9][34] Group 3: Technological Leadership and Open Source Strategy - DeepMind's CEO warned that to have a voice in global AI governance, countries must maintain technological leadership, emphasizing that those who can train models and deploy systems hold the real power [5][6][7] - Mistral AI's open-source strategy aims to democratize access to AI models, allowing more researchers to participate in innovation and reducing the dominance of a few large companies [10][11] - The open-source approach is seen as a way for Europe to establish its influence in the global AI ecosystem and create a counterbalance to the US and China [11] Group 4: Global Collaboration and Future Outlook - The discussion emphasized the need for a global approach to AI innovation, with collaboration across borders being essential to address challenges in various sectors, including energy and life sciences [42][43] - The importance of maintaining a competitive edge in computing power and reducing reliance on external sources, particularly in chip manufacturing, is underscored [44][45] - The upcoming AI summits are viewed as critical opportunities for fostering international dialogue and collaboration in the AI space [48][54]