如果您希望可以时常见面，欢迎标星收藏哦~ 来源：内容 转载自快科技 。 全球首款L3级算力的AI汽车小鹏G7已于昨晚开启全球首秀预售。预售仅46分钟，小鹏G7小订就突 破了10000台。 今日，小鹏G7也在香港车博会上亮相，引发大量用户围观。车辆最核心的亮点就是首发搭载了小 鹏自研的图灵AI芯片，并且一次性用上了三颗，带来了行业首个L3级算力平台。 在媒体提问下，何小鹏现场揭秘G7的三颗图灵芯片位置在哪里。 何小鹏打开小鹏G7车门后介绍，芯片加上内存后，被包装在域控制器里面。在车辆中控扶手下方 的域控制器内，配备有两颗图灵芯片，是用来跑自动驾驶的；副驾手套箱里面，还有一颗图灵芯 片，是负责语音控制和整车大脑。 经过深度定制，图灵AI芯片算力可达到100%极致利用，不止AI智能汽车，它还将应用于小鹏AI机 器人，飞行汽车等物理AI硬件。 参考链接 https://news.mydrivers.com/1/1053/1053404.htm 点这里加关注，锁定更多原创内容 *免责声明：文章内容系作者个人观点，半导体芯闻转载仅为了传达一种不同的观点，不代表半导体芯闻对该 观点赞同或支持，如果有任何异议，欢迎联系我们。 ...

如果您希望可以时常见面，欢迎标星收藏哦~ 来源：内容 转载自快科技 。 全球首款L3级算力的AI汽车小鹏G7已于昨晚开启全球首秀预售。预售仅46分钟，小鹏G7小订就突 破了10000台。 参考链接 https://news.mydrivers.com/1/1053/1053404.htm 点这里加关注，锁定更多原创内容 今日，小鹏G7也在香港车博会上亮相，引发大量用户围观。车辆最核心的亮点就是首发搭载了小 鹏自研的图灵AI芯片，并且一次性用上了三颗，带来了行业首个L3级算力平台。 在媒体提问下，何小鹏现场揭秘G7的三颗图灵芯片位置在哪里。 何小鹏打开小鹏G7车门后介绍，芯片加上内存后，被包装在域控制器里面。在车辆中控扶手下方 的域控制器内，配备有两颗图灵芯片，是用来跑自动驾驶的；副驾手套箱里面，还有一颗图灵芯 片，是负责语音控制和整车大脑。 据了解，小鹏自研图灵AI芯片拥有40核处理器、最高30B可运行模拟参数、2xNPU自研神经网络 处理大脑、DSA集成神经网络特定领域架构、2个独立图像ISP，拥有超2000Tops全本地算力支 撑。 目前行业主流算力约在80～700TOPS，但小鹏G7直接把行业旗舰的算力 ...

Core Viewpoint - Ren Zhengfei's recent interview highlights that Huawei is not as formidable as perceived, especially in the context of U.S. sanctions on its Ascend chips, suggesting that the company is just one of many Chinese chip manufacturers [1][3][4] Group 1: U.S. Sanctions and Huawei's Position - The U.S. sanctions on Huawei's Ascend chips stem from concerns over China's potential advancement in AI technology, similar to the pressure faced due to Huawei's 5G leadership [3] - Ren Zhengfei believes that the U.S. has overestimated Huawei's significance, asserting that the company is merely a "shield" in the broader context of U.S.-China tech competition [4] - Despite the challenges, Huawei has maintained a low profile and focused on innovation rather than excessive publicity, which has contributed to its resilience over the past seven years [6] Group 2: Perceptions of Huawei's Technological Capability - There is a debate regarding the actual technological gap between China and the U.S. in AI chips, with Ren suggesting a significant difference, while South Korean media argue that the gap is not as pronounced [7] - Huawei's substantial R&D investment, totaling 1.249 trillion yuan over the past decade, has raised concerns among competitors like Samsung, indicating Huawei's commitment to innovation [8] - The company's ability to innovate under pressure, exemplified by the development of Kirin and Ascend chips, showcases its research strength despite external challenges [10] Group 3: Ren Zhengfei's Leadership and Outlook - Ren Zhengfei's optimistic attitude and resilience in the face of adversity have been pivotal for Huawei, allowing the company to thrive despite significant external pressures [12] - The changing global landscape necessitates a more respectful and equal approach towards China for potential collaboration and mutual benefit [12]

Core Insights - Xiaopeng Motors launched the G7 globally on June 11, showcasing its advanced capabilities in AI and autonomous driving technology [1] - The G7 features three Turing AI chips, providing the industry's first L3-level computing platform with over 2200 TOPS of computing power, which is 3 to 28 times more effective than other flagship models in the industry [1] - Despite achieving L3-level computing power, the G7 is not classified as an L3 vehicle, as it requires additional L3-level hardware redundancy and legal certification to meet the full criteria [1]

萨克斯表示："我确实担心我们的恐惧让我们无法发现机遇，最终导致我们亲手扼杀眼前这些令人瞩目 的进步。"他还援引了各州立法机构通过的大量旨在监管人工智能的法案，并提到这些法案实际上给企 业在建设支撑AI的数据中心时带来了法律许可上的困难。 据介绍，特朗普此前已经撤销了拜登旨在促进竞争、保护消费者并确保AI不会被用于传播虚假信息的 一项行政命令。他还撤销了拜登所谓的"AI扩散准则"，该规则旨在限制部分国家进口美国AI芯片，防止 它们的AI计算能力快速提高。 "我们撤销了拜登的'扩散准则'，这个准则让'扩散'（diffusion）变成了一个负面词汇。但美国技术的扩 散其实应该是个正面词汇。"萨克斯说。 能够印证萨克斯表态的一点是，特朗普政府上个月刚刚和阿联酋共同宣布：后者将建设美国以外的最大 规模AI园区。此前拜登政府对向海湾国家出口AI芯片还多有限制。 据路透社6月10日报道，在华盛顿举行的亚马逊云科技峰会上，白宫人工智能和加密货币事务负责人戴 维·萨克斯表达了对美国的人工智能（AI）监管政策过于严格的担忧。萨克斯认为这可能会抑制美国AI 行业的增长并将关键市场拱手让给中国。他表示，如果5年后到处都是中国芯片，" ...

Core Viewpoint - The article highlights the significant opportunities for Chinese chip manufacturers, particularly Cambricon Technologies, due to the U.S. tightening restrictions on chip exports to China, including the ban on Nvidia's H20 chip [1][4]. Group 1: Company Overview - Cambricon Technologies has emerged as a leader in the AI chip sector, capitalizing on the market gap created by U.S. sanctions [3][4]. - The company plans to raise 4.98 billion yuan (approximately 0.7 billion USD) through a private placement to fund its AI chip and software development projects [1][4]. - Cambricon's revenue surged to 1.1 billion yuan in Q1, marking a year-on-year increase of 4,230%, and achieving profitability for two consecutive quarters [1][6]. Group 2: Financial Performance - The stock price of Cambricon has increased over 200% this year, with a staggering 387% rise projected for 2024, earning it the title of "stock king" in A-shares [3][4]. - The company's current forward P/E ratio stands at 267, significantly higher than Nvidia's 46 and domestic competitor Haiguang's 146 [5][6]. Group 3: Customer Dependency Risks - Cambricon's revenue is heavily reliant on a few major clients, with the top five clients contributing 95% of its revenue, raising concerns about future revenue stability [6][7]. - The company previously faced a similar risk when Huawei was its largest customer, leading to a drastic revenue drop when Huawei shifted to self-developed chips [7]. Group 4: Market Potential and Competition - The Chinese AI chip market is expected to grow significantly, with predictions indicating that domestic GPU manufacturers could achieve sales of 287 billion yuan by 2027, capturing 70% of the market [7]. - To mitigate customer concentration risks, Cambricon needs to diversify its client base and increase investment in new customer partnerships and R&D [7].

如果您希望可以时常见面，欢迎标星收藏哦~ 因为大模型的火热，AI芯片在过去几年的热度飙升。尤其是英伟达GPU，凭借其先天的软硬 件优势，该系列产品在过去几年里牢牢把持人工智能的训练甚至推理市场。但是，在中国市 场，却因为地缘政治的限制，导致我们不能获得业界先进的GPU或者其他领先的AI芯片。 最近以来，美国还频频升级政策，使得中国客户甚至买不到英伟达等厂商针对中国市场开发的芯 片。在这种态势下，如何打造适合中国市场需求的AI芯片及生态，成为了中国AI从业者的重中之 重。 近日，Gartner研究副总裁盛陵海（Roger Sheng）也以"国产AI芯片如何支持国内AI产业发展"为 题，分享了他对中国AI产业的发展看法。 中美AI的差异 盛陵海首先指出，因为所处的阶段不一样，中美在AI的发展战略上有所差异。 首 先 看 美 国 方 面 ， 因 为 有 最 先 进 的 芯 片 ， 也 是 最 早 推 动 AI 发 展 的 国 家 ， 还 有 OpenAI 、 谷 歌 （Transformer是它提出的）和大量的互联网巨头和软件巨头参与其中（他们想通过云服务来垄断 AI生态）。于是，盛陵海认为这是一种"集中式"的人工 ...

Group 1 - Qi Xiangdong, chairman of Qi Anxin, highlighted the severe internal competition in the cybersecurity industry, stating that software development can take years and cost billions, yet the costs remain invisible during bidding processes, leading to destructive price wars [1][2] - Qi called for government intervention to improve the business environment for the software industry and promote healthy development [1] Group 2 - Wang Ziru expressed gratitude towards Dong Mingzhu and Lei Jun for their encouragement and support at different times, which helped him maintain his ideals [3] Group 3 - Xizi Elevator announced the passing of its chairman and general manager, Liu Wenchao, who died at the age of 54 [4] Group 4 - Amazon announced layoffs in its book department, affecting fewer than 100 employees, as part of efforts to improve efficiency and streamline operations [5] Group 5 - Indonesia is evaluating the feasibility of purchasing Chinese-made J-10 fighter jets to enhance its air force's modernization while considering cost-effectiveness [6] - The performance of the J-10 in the India-Pakistan conflict is a significant factor in Indonesia's consideration [6] Group 6 - XPeng Motors launched the "Chasing Light Panorama" AR-HUD system, developed in collaboration with Huawei, featuring an 87-inch display and supporting multiple driving scenarios [7] - The system aims to enhance driving safety and user experience through advanced imaging technology [7] Group 7 - Elon Musk announced that Tesla's next-generation AI chip, Dojo 2, will be launched later this year, indicating progress in AI training technology [8] Group 8 - Trump Media & Technology Group submitted an application for a Bitcoin ETF, aiming to list it on the New York Stock Exchange, which will directly hold Bitcoin and reflect its price performance [9] - The ETF's approval process is expected to conclude by early next year [9] Group 9 - Semiconductor Manufacturing International Corporation (SMIC) announced plans to sell a 14.832% stake in its Ningbo subsidiary to Hunan Guoke Microelectronics [10] Group 10 - The National Health Commission of China announced plans to promote painless childbirth services across medical institutions, aiming for comprehensive availability by 2027 [12][13]

Core Viewpoint - The article discusses the emerging trend of ASIC (Application-Specific Integrated Circuit) chips in the AI chip market, highlighting NVIDIA's introduction of NVLink Fusion and the competitive landscape between NVIDIA and the UALink alliance [1][3][4]. Group 1: NVIDIA's Position and Strategy - NVIDIA has entered the ASIC market with NVLink Fusion, allowing various processors from companies like MediaTek and Qualcomm to connect to NVIDIA's AI data centers [1]. - NVIDIA's CEO Jensen Huang stated that 90% of ASIC projects are likely to fail, emphasizing the competitive edge of NVIDIA's ASICs over others [3]. - NVIDIA's gross margin remains high at 60%, indicating strong profitability and competitive positioning in the ASIC market [4]. Group 2: Market Growth and Future Projections - The market for cloud-based ASIC chips is projected to exceed $40 billion by 2028, as noted by MediaTek's chairman [3]. - Major cloud service providers are developing their own ASIC chips to create a competitive advantage, with companies like Google, Microsoft, Meta, AWS, and Apple actively involved in this trend [6]. Group 3: Competitive Landscape - The UALink alliance, which includes major companies like Alibaba, AWS, AMD, Apple, Cisco, Google, HPE, Meta, Intel, Microsoft, and Synopsys, is positioning itself against NVIDIA by promoting an open standard for AI chip interconnects [4]. - The competition between NVIDIA's NVLink and UALink is creating two distinct ecosystems in the cloud ASIC chip market [4]. Group 4: Supply Chain and Development - The development of ASIC chips heavily relies on the Taiwanese semiconductor supply chain, with companies like TSMC and others involved in the production process [7]. - The number of AI ASIC servers is expected to surpass 50% of the overall market by 2028, indicating a significant shift towards ASIC technology in AI applications [7].

Group 1 - Meta has signed a 20-year power purchase agreement with Constellation Energy to meet the electricity demands driven by the AI boom, starting from June 2027, purchasing approximately 1.1 GW of power from a nuclear reactor [2] - This marks Meta's first formal entry into the nuclear energy sector, as the company aims to increase its nuclear power capacity in the U.S. by 1 to 4 GW [2] Group 2 - Broadcom has officially launched the Tomahawk 6 series switch chips, achieving a record industry capacity of 102.4 Tbps per chip, which is double the bandwidth of existing Ethernet switch products [3] - The Tomahawk 6 is designed for the AI era, offering unprecedented performance and efficiency, and is already being integrated by several partners into next-generation switching platforms [3] Group 3 - OpenAI has opened the memory feature of ChatGPT to free users, allowing them to save and reference past conversations, although in a limited capacity compared to paid users [4] - This move is aimed at enhancing user engagement and maintaining subscription appeal through more robust memory capabilities for paid users [4] Group 4 - Nvidia has regained its position as the world's most valuable publicly traded company, surpassing Microsoft with a market capitalization of $3.45 trillion, following a 69% year-over-year revenue growth to $44.06 billion in the first fiscal quarter [5] - Nvidia's strong performance highlights the immense potential of the AI chip market, which may drive investment enthusiasm across related industries [5] Group 5 - Elon Musk has criticized Trump's "Big and Beautiful" budget proposal, claiming it will significantly increase the federal budget deficit to $2.5 trillion and burden American citizens with unsustainable debt [6][7] - Musk's strong opposition reflects internal divisions in U.S. fiscal policy and its potential impact on the economy and markets [6][7]