如果您希望可以时常见面，欢迎标星收藏哦~ 黄仁勋当时正处于事业巅峰。 他创办的公司英伟达（Nvidia）的名字源于拉丁语中"嫉妒"一词。但上个月被问及此事时，黄仁勋 却想不出自己有什么值得嫉妒的。"我的生活非常美好，"他在接受《时代》杂志75分钟采访的尾 声说道，随后列举了一系列令他感恩的事情：幸福的婚姻、已成年的子女，以及当天早些时候超声 波检查结果一切正常的两只爱犬。 当然，还有他的职业生涯：掌管着全球市值最高的公司，市值约4.3万亿美元。"我们正在打造世界 上最具影响力的技术，"黄仁勋说道，他指的是驱动人工智能革命的芯片。"对任何人来说，这都是 梦想成真。" 作为世界第八大富豪，黄仁勋享受生活并不令人意外，但当我们在11月下旬与他面谈，准备撰写 《时代》杂志2025年度人物报道时，他或许还有个特别的理由如此开心。他与特朗普总统长达一 年的友谊正在升温，更重要的是，开始带来回报。多年来，美国的出口管制措施阻碍了黄仁勋的芯 片出口到中国、沙特阿拉伯和阿联酋等利润丰厚的市场，而如今，特朗普似乎即将满足他的愿望。 就在我们采访前三天，黄仁勋在华盛顿会见了特朗普和沙特王储穆罕默德·本·萨勒曼（MBS），此 事变得 ...

2025年12月，全球芯片巨头英伟达传出重磅消息：其自主研发的AI芯片定位验证技术已完成内部私下 演示，该技术将以可选软件形式率先搭载于最新Blackwell架构芯片，通过独特的通信时延分析实现芯 片所属国家定位，精度比肩主流互联网定位服务，剑指日益猖獗的高端AI芯片走私乱象，为全球芯片 出口管控体系增添关键技术屏障。 这一技术的诞生并非偶然，而是全球AI芯片领域监管与走私博弈升级的直接产物。近年来，美国对高 端AI芯片实施严格出口管制，将英伟达H100、A100等顶级产品纳入限售清单，但巨大的市场需求催生 了成熟的走私产业链。据业内估算，仅2025年第二季度，就有价值约10亿美元的被禁英伟达芯片通过东 南亚中转、伪装交易等方式流入中国市场，深圳华强北甚至形成了专门的走私芯片维修地下产业，十几 家企业专注于修复因超负荷运行而故障的高端GPU。这种现状不仅让美国出口管制政策形同虚设，也引 发其政府与国会的强烈关注，推动英伟达加速研发芯片流向追踪技术。 从技术原理来看，该定位功能跳出了传统GPS定位的思维框架，依托英伟达GPU的机密计算能力构建核 心架构。其关键在于捕捉芯片与英伟达官方服务器通信过程中的时延差异— ...

华夏时报记者石飞月北京报道 "向商业客户供应H200是一种值得肯定的举措。"12月9日，在被问及H200被美国政府解禁一事时，英伟达方面这样回复《华夏时报》记者。当日，环球网 引述外媒报道称，美国总统特朗普在社交媒体上发文宣布，美国政府将允许英伟达向中国出售其H200人工智能芯片。 受此利好消息影响，当地时间12月8日，英伟达股价在盘后涨2.34%，目前其市值为4.51万亿美元，仍居全球第一。然而这次解禁是有条件的，美国政府将 从英伟达销售所得收益中抽取分成。在业内人士看来，H200如果能够回归中国市场，英伟达还会在整个产业中占据重要地位，但市场份额或很难再回到 曾经的高峰。 有条件解禁高端芯片 其实在上个月，就有消息传出，称特朗普政府正考虑批准向中国出口英伟达的H200人工智能芯片。H200芯片被称为英伟达已面世芯片中"性能第二 强"的，英伟达更领先的芯片是Blackwell和即将面世的Rubin。 由于这几年对华出口芯片政策的不断收紧，英伟达的AI芯片接连被迫"舍弃"中国市场，从一开始的A100、H100，再到"阉割版"的A800和H800，今年轮到 了进一步的"阉割版"H20。H200发布于2023年1 ...

由于这几年对华出口芯片政策的不断收紧，英伟达的AI芯片接连被迫"舍弃"中国市场，从一开始的 A100、H100，再到"阉割版"的A800和H800，今年轮到了进一步的"阉割版"H20。H200发布于2023年11 月，可以说从来未被美国政府允许过在中国市场流通。 在H20重返中国市场前景尚不明朗之际，美国政府的解禁考虑却直接指向了性能更高的H200，其背后的 政策逻辑令人深思。 不过，在看到特朗普列出的"条件"时，一切似乎又变得"合理"。特朗普表示，美方将从相关芯片出口中 收取25%的分成。他还称，美国商务部正在敲定相关安排细节，同样的安排也将适用于超微半导体公 司、英特尔等其他人工智能芯片公司。 对于英伟达来说，被美国政府抽取25%的分成，必然是影响其自身收入的，但如果没有这样的妥协，或 许英伟达还将继续缺席中国市场。《华夏时报》记者就此采访了英伟达方面，对方表示暂无更多细节可 分享。 在芯片禁令之前，来自中国市场的收入一度占到英伟达总收入的五分之一，而今年10月，黄仁勋指出， 英伟达在中国的市场份额从95%降到0%（高端AI芯片），该公司已100%离开了中国市场："我们（美 国）实施的政策，导致美国失去了 ...

不确定因素依然存在。12月3日，英伟达首席执行官黄仁勋在与特朗普会面后曾表示，如果美国放松对 AI芯片销售的限制，他也无法判断中国是否会接受其公司的H200芯片。 H200芯片发布于2023年11月，在FP16（16位浮点数）精度下的峰值理论算力达1979 TFLOPS（1 TFLOPS等于每秒一万亿次浮点运算），而H20芯片同精度下的算力仅为148 TFLOPS。在H200芯片放开 前，作为"缩水版"的H20是美国政府允许对华出口的最尖端AI芯片，不过后来陷入芯片"后门"安全漏洞 的争议。 获准重启H200芯片对华销售之后，英伟达中国方面向南都记者回应称："向商业客户供应H200是一种值 得肯定的举措。"英伟达发言人公开表示："赞赏特朗普总统允许美国芯片产业参与竞争，从而支持美国 的高薪工作岗位和制造业。政府允许英伟达向经商务部审查的商业客户提供H200，是一种深思熟虑的 平衡，这对美国来说是一件好事。" 实际上，黄仁勋今年以来一直在游说特朗普政府和美国国会，希望放宽对华芯片出口。他在10月初一场 论坛上表示，英伟达在华AI芯片市场份额已从95%降至0，美国对华芯片管制不仅"伤害中国"，"往往也 会伤害美国 ...

【文/观察者网 阮佳琪】 他坦言，"我们必须承认，华为是全球范围内极具实力的科技企业之一。我们与该公司存在竞争关系， 他们实力强劲、反应敏捷且行动迅速。" 在周三的活动中，黄仁勋还提到，中国在AI竞赛中相对美国的另一大优势是能源供应充足。他说，"中 国的能源总量是美国的两倍，而美国经济规模却大于中国，这在我看来完全不合常理。" 基于此，他呼吁美国"必须利用一切可用的能源形式"，并强调美国"应该努力鼓励和加速"核能发展。 急，很急，特别急——这就是英伟达首席执行官黄仁勋近来的真实写照。为了搞定对华芯片出口，最近 一段时间他都在忙活着向各方强调事态"十万火急"。 据《日经亚洲》4日报道，当地时间周三，在华盛顿智库美国战略与国际问题研究中心（CSIS）举办的 一场活动上，黄仁勋第N+1次发出警告称，如果美国企业放任华为等中国竞争对手抢占市场，中国将很 快寻求向全球输出其人工智能（AI）技术。 活动上，黄仁勋表示，美国限制英伟达芯片对华出口的举措，"实质上已拱手让出全球第二大人工智能 市场"，这将为华为等中国本土企业的技术成熟提供发展空间，最终使其具备在全球层面与美国企业抗 衡的实力。 "中国市场的地位无可替代，我们 ...

Core Viewpoints - Michael Burry emphasizes that the primary indicator of a bubble is supply-side greed, which leads to over-expansion and ultimately market crashes, rather than demand shortages or profit deficiencies [7][11][12] - The current AI boom mirrors the 1990s internet bubble, with significant investments in AI infrastructure that may not align with actual demand [12][13] Group 1: Historical Analysis of Bubbles - The internet bubble of the 1990s was driven by excessive capital investment in data transmission infrastructure, leading to a supply-demand imbalance [7][8] - Major companies like AT&T and MCI invested heavily in infrastructure, but the actual demand for broadband was not met, resulting in a significant market crash by 2002 [8][11] - Similar patterns of over-investment leading to market corrections have been observed in the real estate bubble of the 2000s and the shale oil revolution of the 2010s [11] Group 2: Current AI Landscape - Major tech companies plan to invest nearly $3 trillion in AI infrastructure over the next three years, raising concerns about potential overcapacity [12] - OpenAI's projected spending of $1.4 trillion over eight years, with revenues not even close to covering this expenditure, highlights the unsustainable nature of current valuations [12] - The rapid pace of technological advancement in AI, particularly with companies like NVIDIA, raises questions about the longevity and economic viability of older chip models [22][23] Group 3: Financial Practices and Risks - Burry points out that major tech firms are extending the depreciation periods of their assets, which artificially inflates reported profits [20][21] - This accounting practice can lead to significant risks, as seen in the case of Baidu, which had to write down substantial asset values after extending depreciation periods [25] - The rapid obsolescence of technology, particularly in data centers, poses a risk of "zombie assets" that may not generate expected returns [24] Group 4: Clarifications on Misinterpretations - Burry clarifies that his positions in options against companies like Palantir and NVIDIA have been misrepresented in the media, emphasizing the importance of accurate reporting [26] - He distinguishes between criticizing accounting practices and directly accusing companies of fraud, asserting that his concerns are about industry-wide practices rather than specific companies [26]

Core Viewpoint - Nvidia is facing renewed scrutiny regarding its AI valuation and potential bubble, prompted by significant shareholder sell-offs and questions about its accounting practices. In response, the company has issued a detailed internal memo addressing at least twelve key allegations from critics, including prominent figures like Michael Burry [1][2]. Group 1: Shareholder Concerns and Accounting Practices - Nvidia's internal memo counters Michael Burry's claims regarding stock buybacks, clarifying that the actual buyback amount since 2018 is $91 billion, not the $112.5 billion Burry cited, which included miscalculated RSU tax payments [1]. - The company firmly denies comparisons to historical accounting fraud cases like Enron and WorldCom, asserting that its business fundamentals are strong and its financial reporting is transparent and complete [1]. Group 2: Depreciation and Competition - Nvidia addresses concerns about long customer depreciation periods, stating that its customers typically depreciate GPUs over 4 to 6 years, reflecting actual equipment lifespan and utilization. The A100 chip, released in 2020, continues to operate at high utilization rates and will generate significant economic value beyond 2 to 3 years [2]. - The company is also responding to competitive pressures from Google's self-developed TPU chips, emphasizing that its GPUs outperform Google's ASIC chips in versatility, flexibility, and performance. Nvidia maintains that its platform is the only solution capable of running all AI models across various environments [2]. Group 3: Market Position and Future Outlook - Despite the competitive landscape, analysts believe Nvidia retains over 90% of the global AI chip market share, with short-term challenges primarily related to market sentiment and valuation fluctuations [3]. - Nvidia's CEO Jensen Huang acknowledges the competition from TPUs and maintains communication with Google DeepMind's CEO, emphasizing that the theory of "scaling laws" supports ongoing demand for high-performance GPUs and systems, which underpins Nvidia's long-term growth prospects [3].

Group 1 - The resurgence of the "AI bubble" narrative is causing significant corrections in AI-related growth stocks across US, A-share, and Hong Kong markets, indicating a potential disconnect between price and value [1] - Analysts suggest that the assessment of whether an "AI bubble" exists depends on the extent of price deviation from value and whether investments exceed demand and capacity [1] - Leading AI companies are beginning to generate substantial revenue, and the current investment intensity in AI remains reasonable [1] Group 2 - Michael Burry, a well-known investor, argues that the current AI hype mirrors the late 1990s internet bubble, with Nvidia at the center of this "bubble" due to its significant market capitalization of approximately $4.44 trillion as of November 24 [2] - Burry highlights that the current AI boom is driven by high-profit tech giants, similar to the role played by Microsoft, Intel, Dell, and Cisco during the internet bubble, with these companies planning to invest nearly $3 trillion in AI infrastructure over the next three years [2][3] - The unsustainable high capital expenditures by tech giants for data centers and chip purchases are not matched by actual revenue from downstream applications, raising concerns about the viability of these investments [3] Group 3 - The total global spending on AI data centers and chips is projected to reach $2.9 trillion by 2028, with tech giants expected to contribute $1.4 trillion, while the remaining gap may be filled through debt financing [3] - The potential emergence of financial derivatives in future fundraising efforts raises concerns reminiscent of the risks seen during the subprime mortgage crisis [3] - Nvidia has responded to bubble concerns by stating that its strategic investments represent a small portion of its revenue and that the capital raised in global markets is minimal compared to the total [4] Group 4 - Nvidia's CFO emphasized the long lifespan of its chips, asserting that older models like the A100 are still operating at full capacity, countering claims about the sustainability of its products [5] - Analysts argue that not all long-term investments are bubbles, and the true measure of a bubble lies in whether investments exceed demand and capacity [5] - The discussion around AI's potential is centered on two main aspects: internal demand for cost reduction and productivity enhancement, and external demand from new application scenarios, with the latter still lacking breakthrough developments [6]

Group 1 - The U.S. government is considering whether to allow NVIDIA to sell advanced AI chips to China, with President Trump seeking advice from various advisors [1][2] - There is a conflict between promoting economic expansion and maintaining national security, as highlighted by U.S. Commerce Secretary Raimondo [1] - NVIDIA has been actively lobbying the Trump administration to regain access to the Chinese market, with CEO Jensen Huang emphasizing the need for a compelling reason to export [1][2] Group 2 - Recent discussions among U.S. officials have focused on which chips could be sold to China, with some viewing the H200 processor as a compromise compared to more advanced options like Blackwell [2] - If the Trump administration approves the sale, it would represent a significant easing of trade restrictions aimed at curbing China's AI capabilities [3] - The H200 chip, which has double the performance of its predecessor H100, is currently the most advanced model approved for export to China, although demand for it has waned [3] Group 3 - Since 2022, the U.S. government has implemented strict export controls to suppress China's chip development, prohibiting NVIDIA from exporting high-end GPUs for AI applications [3] - Earlier this year, NVIDIA faced restrictions on exporting the H20 chip until it agreed to pay 15% of its sales in China to the U.S. government [3] - Chinese companies are actively working to develop domestic AI chips to replace NVIDIA's market share, with major firms like Huawei and Alibaba increasing their investment in chip R&D [4]