国芯网[原:中国半导体论坛] 振兴国产半导体产业！ 不拘中国、 放眼世界 ！ 关注 世界半导体论坛 ↓ ↓ ↓ 11月24日消息，近日，据外媒最新报道称，美国政府正考虑允许英伟达对华出售H200芯片。 报道援引知情人士消息称，负责监管美国出口管制的美国商务部正就改变对华出口限制一事进行审查，并称相关计划可能会发生变动。 目前，美国商务部暂未对此作出回应，英伟达尚未就此直接置评。 两年前发布的H200芯片相比于其前代产品H100芯片拥有更多的高带宽内存，使其能够更快速地处理数据。据估计，H200芯片的性能是英 伟达H20芯片的两倍。 对于黄仁勋来说，他们迫切希望重新能够对中国厂商出手英伟达芯片，因为除了能够获得利润外，还能够在一定程度上抑制华为等国产厂 商的发展势头。 半导体公众号推荐 半导体论坛百万微信群 加群步骤： 第一步：扫描下方二维码，关注国芯网微信公众号。 第二步：在公众号里面回复"加群"，按照提示操作即可。 爆料｜投稿｜合作｜社群 文章内容整理自网络，如有侵权请联系沟通 "由于美国出口限制，英伟达对华芯片销售陷入停滞，预计未来两个季度在华销售额将为零。伤害中国的事情，往往也可能伤害美国，甚至 会更严 ...

据环球网22日援引外媒报道称，特朗普政府正考虑批准向中国出口美国芯片制造商英伟达的H200人工 智能芯片。 报道表示，两年前发布的H200芯片相比于其前代产品H100芯片拥有更多的高带宽内存，使其能够更快 速地处理数据。据估计，H200芯片的性能是英伟达H20芯片的两倍。 具体数据上，H200是英伟达第一款使用HBM3e内存的芯片，以每秒4.8TB的速度提供141GB的内存；与 A100相比，H200内存容量约提升两倍，带宽约增加了2.4倍。此外，英伟达随后推出的H200 NVL是基 于Hopper架构的PCIe版GPU。与H100 NVL相比，H200 NVL的内存容量提升了1.5倍，带宽提升了1.2 倍，能够在数小时内完成大型语言模型的微调，并提供高达1.7倍的推理性能提升。 另一方面，谷歌连续发布新模型相对于ChatGPT展现出了巨大的进步，高盛等对此分析表示，"这一点 已显而易见，但市场对此却反应迟钝。"谷歌供应链对OpenAI链的强烈替代走势。 华泰证券也认为谷歌TPU是目前唯一能与英伟达GPU匹敌的AI加速器，同时依托TensorFlow、OpenXLA 等学习框架与TPU构建软硬一体的AI生态， ...

当下，一场电力饥渴正席卷全球AI产业。2025年，OpenAI在德州阿比林规划4.5GW电力接入，相当于五座核电站的发电量；微软GPU集群因电力不足 而"吃灰"；北弗吉尼亚数据中心的项目排队"等电"长达数年……这些现象指向一个残酷现实：AI竞赛的核心正从算力芯片转向电力基建。 从"缺芯片"变为"缺电" "我们现在面临的最大问题，不再是算力资源过剩，而是电力——以及能否够快地在有电力的地方完成设施建置。"微软首席执行官萨提亚·纳德拉在最近 与OpenAI首席执行官萨姆·奥特曼对话的播客节目中坦言。他补充道："如果做不到这点，我们可能会有一堆芯片躺在仓库里却插不上电。" 缺电，让微软高价购买的显卡在仓库里吃灰 回顾过去几年，AI的能耗增长速度令人震惊。据国际能源署数据，2022年全球数据中心总耗电量约460太瓦时，占全球用电量的2%；到2026年，这一数据 将超过1000太瓦时，大约是整个日本2022年全年的用电量。 "一年多以前短缺的是芯片，接下来短缺的将是电力。"而就在一年之前，埃隆·马斯克在"博世互联世界2024"大会上的警告言犹在耳。几乎在同一时间， 萨姆·奥特曼在达沃斯论坛上也表达了类似担忧，承认人工智 ...

Group 1 - Elon Musk stated that the advent of Starship opens the path for large-scale deployment of solar-powered AI satellites, which he believes is the only way to achieve 1 terawatt (1TW) of AI computing power annually [1] - The energy supply shortage is a critical bottleneck for AI data center construction, with FTI Consulting predicting that energy demand for data centers in the U.S. will nearly double by 2027 [3] - The construction of data centers in space is becoming a viable option for many Silicon Valley tech companies due to power limitations on Earth [4] Group 2 - StarCloud successfully launched a satellite equipped with NVIDIA H100 chips and Google Gemini models, which will provide 100 times the GPU computing power compared to previous space computing facilities [4] - StarCloud plans to build a 5-gigawatt orbital data center powered by large solar and cooling battery panels, which will not require water for cooling or rely on batteries [4] - Google is considering a project named "Project Suncatcher" to establish the first space data center, planning to launch prototype satellites by 2027 in collaboration with Planet [5] Group 3 - The first space computing satellite constellation was successfully launched by Guoxing Aerospace and Zhijiang Laboratory, consisting of 12 satellites capable of processing data in orbit [7] - The U.S. has launched over 10,000 satellites, with significant contributions to the ongoing conflict in Ukraine, and there are expectations for accelerated launch progress in China in the coming years [7]

Core Insights - The software ETF (515230) rose by 1.1% in early trading on November 17, indicating positive market sentiment towards the software sector [1] - China's space computing constellation is accelerating its construction, with the "Three-Body Computing Constellation" led by Zhijiang Laboratory, aiming to complete a layout of over 50 computing satellites by 2025 to build an integrated space-ground computing network [1] - NVIDIA's H100 chip is set to support space applications, boasting computational power 100 times greater than previous space-based computers, with future tests planned for AI processing applications [1] - SpaceX plans to expedite its space computing layout, with the "Starship" expected to achieve an annual deployment of 1 terawatt of AI computing power, and the capacity of V3 satellites projected to increase tenfold to 1 Tbps [1] - The global acceleration of space AI development is expected to create new business models, with satellite manufacturing and space-based AI applications becoming areas of significant interest [1] Industry Overview - The software ETF (515230) tracks the software index (H30202), which selects listed companies involved in system software, application software development, and related services, reflecting the overall performance of the software industry [1] - The software index primarily focuses on the information technology sector, characterized by significant growth and innovation, serving as an important barometer for observing the development dynamics of the software industry [1]

Core Insights - Major companies plan to invest $1 trillion in AI data centers over the next five years, with a focus on depreciation as a key financial consideration [2] - The lifespan of AI GPUs is uncertain, with companies like Google, Oracle, and Microsoft estimating a maximum lifespan of six years, but potentially shorter [2][4] - Investors are concerned about the depreciation period, as longer asset lifespans lead to smaller impacts on profits [2] Depreciation Challenges - AI GPUs are relatively new, with NVIDIA's first AI-specific processor launched around 2018, and the current AI boom starting in late 2022 [4] - NVIDIA's data center revenue surged from $15 billion to $115 billion in the fiscal year ending January 2023 [4] - There is no historical reference for the lifespan of GPUs, making it difficult for companies to estimate depreciation accurately [4][5] Market Reactions - CoreWeave has set a six-year depreciation cycle for GPUs, indicating a data-driven approach to asset valuation [4][5] - Despite high demand for NVIDIA's A100 and H100 chips, CoreWeave's stock fell 16% after earnings guidance was affected by third-party data center developer delays [5][6] - The stock of Oracle has also dropped 34% since reaching a historical high in September [6] Skepticism in the Market - Short-seller Michael Burry has expressed doubts about the longevity of AI chips, suggesting that companies may be overstating their lifespan and underestimating depreciation costs [6] - Burry believes that the actual lifespan of server equipment is around two to three years, which could inflate reported earnings [6] Technological Advancements - AI chips may depreciate within six years due to wear and tear or obsolescence from newer models [8] - NVIDIA's CEO has indicated that older chip models will lose significant value as new models are released [8] - Amazon has shortened the expected lifespan of some servers from six years to five years due to rapid technological advancements [8][9] Strategic Procurement - Microsoft is diversifying its AI chip procurement to avoid over-investment in any single generation of processors [9] - The rapid iteration of technology in the AI sector complicates depreciation estimates, requiring careful financial forecasting [9]

Core Viewpoint - The depreciation period of GPUs is a critical issue affecting corporate profits and investment returns, especially as major tech companies plan to invest $1 trillion in AI data centers over the next five years [3][5]. Depreciation Challenges - The actual lifespan of GPUs is under scrutiny, with estimates ranging from two to six years, leading to concerns about inflated earnings by companies like Microsoft, Google, and Oracle [3][6]. - The lack of historical data on GPU usage complicates depreciation assessments, making it difficult for investors and lenders to gauge the value of these assets [5][6]. Market Reactions - Concerns about AI spending have already impacted stock prices, with CoreWeave's shares dropping 57% from their June peak and Oracle's stock falling 34% from its September high last year [3]. - CoreWeave has adopted a six-year depreciation cycle for its infrastructure, but its stock fell 16% following earnings reports due to delays from third-party data center developers [6][3]. Technological Impact - Rapid technological advancements are pressuring the depreciation of AI chips, with new models being released annually, which may render older models obsolete more quickly [7][8]. - Companies like Amazon have shortened the expected lifespan of some servers from six years to five years due to the accelerated pace of technological development in AI and machine learning [7]. Corporate Strategies - Microsoft is diversifying its AI chip procurement to avoid over-investment in any single generation of processors, acknowledging the rapid pace of innovation [8][9]. - Depreciation estimates are influenced by various factors, including technological obsolescence and maintenance, requiring companies to justify their assumptions to auditors [9].

Core Insights - The article discusses the significant financial implications of determining the depreciation period for GPUs as major tech companies plan to invest $1 trillion in AI data centers over the next five years [1] - The depreciation period directly affects financial performance, with longer periods allowing companies to spread costs over more years, thus reducing profit impact [1][4] - Concerns about AI spending are reflected in stock price declines for companies like CoreWeave and Oracle, indicating investor skepticism about over-investment in AI [1] Depreciation Challenges - Estimating GPU depreciation is complicated due to a lack of historical usage data, as the first AI processors from NVIDIA were launched around 2018, and the current AI boom began in late 2022 [4] - CoreWeave has adopted a six-year depreciation cycle for its infrastructure, while its CEO emphasizes a data-driven approach to assess GPU lifespan [5] - Market opinions vary, with some suggesting actual GPU lifespan may be as short as two to three years, leading to concerns about inflated earnings projections by major tech firms [5] Technological Pressure - The rapid pace of technological advancement is a key factor in GPU depreciation, with new models potentially rendering older ones obsolete within a short timeframe [6][7] - NVIDIA has shifted to an annual release cycle for new AI chips, increasing the risk of older models losing value quickly [7] - Amazon has reduced the estimated lifespan of some servers from six years to five due to accelerated technological development in AI and machine learning [7] Strategic Responses from Tech Giants - Microsoft is diversifying its AI chip procurement strategy to avoid over-investment in any single generation of processors, learning from NVIDIA's rapid product cycles [8] - Depreciation estimates in fast-evolving industries like technology require careful consideration of various factors, including technological obsolescence and historical lifespan data [8]

Core Insights - The article discusses the emerging trend of space computing, highlighting the competitive landscape among global tech giants to establish capabilities in this area [1][2][3] - China's advancements in space computing are noted, with a complete chain from technology development to commercial application already in place [1][4] - The article raises critical questions regarding the high costs of construction and operation, the potential for market saturation, and the timeline for commercial viability in China's space computing sector [1][5] Industry Developments - StarCloud, a US startup, launched a test satellite equipped with NVIDIA H100 chips and Google's Gemini model, aiming to create a gigawatt-level distributed data center in space [2] - The European Union has initiated a "Space Data Center Plan" as part of its Horizon Green Transition strategy, focusing on building low-carbon computing clusters in space [2] - Madari Space in the Middle East plans to launch its first orbital data center by 2026, with a goal of deploying 8,000 space nodes by 2028 [2] Technological Trends - The demand for new computing capabilities in the AI era is driving the shift towards space computing, which is seen as a next-generation "green high-density computing platform" [3] - The natural conditions of space, such as low temperatures and long-term solar energy supply, provide ideal conditions for zero-carbon computing [3] - The concept of "space-based computing constellations" is emerging, with AI-driven management and coordination becoming essential for large satellite networks [3][4] Commercialization Pathways - The commercial path for space computing is becoming clearer, with applications in various sectors such as AI, research, and emerging industries like low-altitude economy and digital consumption [7] - The business model is shifting from "selling computing power" to "selling services," indicating a more integrated approach to customer needs [8] - The complete industry chain for space computing includes satellite manufacturing, communication devices, chip modules, AI algorithms, and ground access terminals [8] Challenges and Opportunities - The Chinese space computing industry faces structural challenges, including a high number of satellite constellation plans with low implementation rates [11] - The need for coordination among launch capabilities, manufacturing capacity, and orbital resource management is critical for the development of satellite constellations [11] - Despite initial high costs and long cycles, there is optimism about the long-term advantages of space computing as industrialization progresses and costs decrease [12]

Core Points - The Biden administration has informed federal agencies that Nvidia will not be allowed to sell its latest simplified AI chip, the B30A, to China [5] - The B30A chip is a compliant version that meets current U.S. export restrictions, but its performance is lower than Nvidia's flagship H100 [5] - Nvidia has provided samples of the B30A chip to several Chinese customers, indicating a potential demand for its capabilities in training large language models [5] - Nvidia's spokesperson stated that the company has zero market share in the competitive Chinese data center computing market and has not included this market in its performance guidance [5] - Nvidia is reportedly working on modifying the B30A design in hopes that the U.S. government will reconsider its stance [5] Summary by Sections Section 1: Government Regulations - The Biden administration has prohibited Nvidia from selling the B30A chip to China, reflecting ongoing tensions and regulatory scrutiny in the tech sector [5] Section 2: Product Details - The B30A chip is designed to comply with U.S. export restrictions and is intended for use in training large language models, which is a significant requirement for many Chinese enterprises [5] Section 3: Market Position - Nvidia's spokesperson highlighted the company's lack of market presence in China, emphasizing that it has not factored this market into its financial outlook [5] Section 4: Future Developments - Nvidia is in the process of redesigning the B30A chip, aiming for a potential policy shift from the U.S. government regarding its export to China [5]