刚发布第三代机器人Figure 03、收获全网关注的Figure，很快又因CEO的言论再度引发热议？ 事情是这样婶儿的…… 有网友发推文表示：Figure CEO Brett Adcock最近接受了《时代》采访，《时代》称Figure 03仍存在重大问题，尚未适用于日常工作。 当被问及还需要什么时，Brett回答：数据。 要是认真琢磨他这话，会发现他最终想表达的是，scale是让机器人技术能在所有工作领域派上用场的关键，而这也是大型超大规模企业（big hyperscalers）能从中获利的又一领域。 一石激起千层浪。 有网友直言这话说得没毛病，"数据是新的石油"。 但也有不少反对的声音，有网友反驳道： 我不同意。不过他当然更有见地。在我看来，问题在于缺乏合理的架构和算力。人类执行简单任务并不需要海量数据。 争议声中，有网友还认为： 显而易见，这正是他们回避进行现场演示的原因。但无可否认，Figure在人形机器人的自主能力上正进行着开创性的工作。 最终这场争论引起了Brett Adcock本人的关注。 没想到Brett Adcock亲自来反驳了，直言"事实并非如此"。 我们向《时代周刊》演示的绝大多数操作策略 ...

西风 发自 凹非寺 量子位 | 公众号 QbitAI 刚发布第三代机器人Figure 03、收获全网关注的Figure，很快又因CEO的言论再度引发热议？ 事情是这样婶儿的…… 有网友发推文表示：Figure CEO Brett Adcock最近接受了《时代》采访，《时代》称Figure 03仍存在重大问题，尚未适用于日常工作。 当被问及还需要什么时，Brett回答： 数据 。 要是认真琢磨他这话，会发现他最终想表达的是，scale是让机器人技术能在所有工作领域派上用场的关键，而这也是大型超大规模企业 （big hyperscalers）能从中获利的又一领域。 一石激起千层浪。 有网友直言这话说得没毛病，"数据是新的石油"。 但也有不少反对的声音，有网友反驳道： 我不同意。不过他当然更有见地。在我看来，问题在于缺乏合理的架构和算力。人类执行简单任务并不需要海量数据。 争议声中，有网友还认为： 显而易见，这正是他们回避进行现场演示的原因。但无可否认，Figure在人形机器人的自主能力上正进行着开创性的工作。 最终这场争论引起了Brett Adcock本人的关注。 没想到Brett Adcock亲自来反驳了，直言 ...

点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近30个 方向 学习 路线 以下文章来源于Vehicle ，作者Pirate Jack Vehicle . 汽车行业先进个人与团队关注 Vehicle， 一起智能、出海、成长 作者 | Pirate Jack 来源 | Vehicle 原文链接： Waymo自动驾驶最新探索实践：世界模型、长尾问题、最重要的东西 >>自动驾驶前沿信息获取 → 自动驾驶之心知识星球 本文只做学术分享，如有侵权，联系删文 辅助驾驶/自动驾驶产业应该最终都会交叉，因为他们算法软件的底层逻辑是一样的。我们之前文章《 IAA 2025 慕尼黑车展：中国汽车军团们，从"贸易出口"到"势 不可挡" 》也分享到了，中国在辅助驾驶开拓算法的公司不管Momenta、元戎、大疆都也和其他专做L4的公司一样在欧洲和中东各地掘金L4。 本文是基于Waymo 感知算法负责人 Wu Chen女士在今年CVPR上演讲内容总结，再根据自身经验分析自动驾驶算法、发展现状以及对于自动驾驶到底最重要的东西 是什么。 构建一个世界模型 Waymo表示，他们开发了一个名为 Waymo 基础模型的大 ...

Core Viewpoint - Dustin Tran, a former senior researcher at Google DeepMind, has joined xAI and is recognized for his significant contributions to the development of the Gemini AI model, which has achieved state-of-the-art reasoning capabilities and won multiple prestigious competitions [1][2][12]. Group 1: Dustin Tran's Contributions - Tran played a pivotal role in the development of the Gemini product line, which helped Google regain its position in the AI landscape after the decline of GPT [2][12]. - Under Tran's leadership, the Gemini series, particularly Gemini 1.5 Pro, excelled in various AI benchmarks, marking a significant turnaround for Google [15][16]. - Tran's team was instrumental in the rapid development of Gemini's predecessor, Bard, despite its initial poor reception [13][14]. Group 2: Transition to xAI - Tran's decision to join xAI was influenced by three main factors: superior computing power, innovative data strategies, and alignment with Elon Musk's corporate philosophy [27][28][29]. - He expressed admiration for the extensive resources available at xAI, which he found unparalleled even during his tenure at Google [30][31]. - Tran believes that xAI has the potential to achieve rapid advancements in AI capabilities, surpassing other companies in a short timeframe [35][36]. Group 3: Background and Achievements - Tran has an impressive academic background, having graduated from UC Berkeley, earned a master's degree from Harvard, and pursued a PhD at Columbia University [22]. - He has contributed to several influential projects and publications in the AI field, with over 24,000 citations on Google Scholar [25][23]. - His early career included a brief internship at OpenAI, where he was involved in notable projects like the Dota 2 AI [21][19].

本文来自微信公众号：小马宋，作者：小马宋，题图来自：AI生成 一家餐厅使用还是没有使用预制菜。 这个叫事实。 我不知道它家是不是用了预制菜，但我觉得它用的是预制菜。 这个叫猜测。 我觉得这家餐厅不好吃，很贵。 这个叫观点。 一个人觉得这家餐厅好吃还是不好吃，没有对错。 因为好不好吃是主观的。 我通过1万人的测试，发现51%的人说好吃，49%的人觉得不好吃。 这个叫数据，也是一种事实。 我觉得它们家很贵，又不好吃，非常气愤。 这是基于观点的情绪。 如果一家餐厅没有使用预制菜，但我吃完说它几乎都是预制菜。 私下说，这叫发泄情绪；公开说，这会变成诬陷。 究竟什么算是预制菜，国家制定了一个标准。 这个叫定义。 本来没什么预制菜的标准，但大家普遍认同某个标准。 这个叫共识。 我觉得这家餐厅好吃，他觉得这家餐厅不好吃，我们吵起来了。 这个叫争论。 吵架之前，先确定标准或者定义；然后基于事实，确定事实；根据事实来确定对错，公开结论。 这个叫程序。 如果只是个人观点，其实不必吵架，吵也没用，永远不可能有结论。 本文来自微信公众号：小马宋，作者：小马宋 ...

Core Insights - The INBOUND conference, attended by approximately 13,000 participants, was held in San Francisco for the first time after 15 years in Boston, showcasing the company's commitment to innovation [1] - HubSpot announced over 200 updates and features across its platform, indicating a strong pace of innovation [1] - A key theme from the conference was the importance of data as a foundational element for AI initiatives [2]

Separation Journey - The spin-off of the Solstice business is on schedule for Q4 this year, with an Investor Day planned for October 8 [1] - The Aero spin-off is expected to occur in the second half of next year, approximately one year from now [1] Complexity and Opportunities - The previous spin-off experience with AM provided a preparatory phase for the more complex Aero spin-off, which is being managed by a dedicated project team [2] - The CEO of RemainCo Automation has identified a significantly larger opportunity set in automation, driven by the convergence of cloud, data, and AI, than initially anticipated during the separation decision [2]

Honeywell FY Conference Summary Company Overview - **Company**: Honeywell (NasdaqGS: HON) - **Date of Conference**: September 10, 2025 Key Points on Separation Journey - The spin-off of the Solstice business is on schedule for Q4 2025, with an investor day planned for October 8, 2025 [7] - The ERO spin-off is expected in the second half of 2026, with no major surprises reported during the execution process [7] - The CEO noted that the opportunity set in automation, driven by cloud, data, and AI, is larger than initially anticipated [8] - Honeywell has connected 20,000 customers, enhancing service capabilities and reducing costs [9] Strategic Focus and Portfolio Management - The separation allows each entity to focus on its own strategy, with aerospace and automation projected to generate approximately $20 billion in revenue each [13] - Honeywell is actively managing its portfolio, having made six acquisitions, four in automation and two in aerospace, while also conducting strategic reviews of existing businesses [15][16] - The company aims to drive common outcomes such as energy efficiency and operational excellence through its portfolio [14] Quantum Fundraising - Honeywell completed a quantum fundraise of $600 million, aiming to increase it to $700 million, indicating growing investor interest in quantum technology [17] - The focus areas for quantum applications include research and life sciences, banking, and cybersecurity [18] Aerospace Business Insights - The aerospace business is expected to grow to $30 billion in less than a decade, with mid-single to high-single-digit growth anticipated [23] - Margins have stagnated around 25-26%, influenced by acquisition costs and unfavorable OE mix, but are expected to stabilize [25][29] - The backlog for aerospace has reached $70 billion, indicating strong future growth potential [27] Automation Business Developments - The automation business is focusing on high-growth verticals and leveraging data and AI for operational improvements [39] - Recent acquisitions in LNG and cybersecurity are aimed at strengthening Honeywell's position in critical future markets [37] - The Access Solutions business has shown strong sales synergies, particularly in data centers [42] Market Conditions and Pricing Strategy - Honeywell is cautious about the impact of tariffs and economic conditions on its business, particularly in international markets [46][50] - The company has adopted a strategy to protect volume while managing pricing, with a focus on maintaining margins despite inflationary pressures [52][53] R&D and Growth Outlook - Increased R&D spending is expected to drive organic growth, with a typical product development cycle of 18 months [54][55] - Honeywell aims to achieve mid to high single-digit growth across its businesses, contingent on favorable market conditions [57] Conclusion - Honeywell is strategically positioning itself for future growth through separation, focused acquisitions, and leveraging technology advancements in automation and aerospace sectors. The company remains vigilant about market conditions and is committed to maintaining a strong growth trajectory.

"国内自媒体总说，自主品牌车企的智能辅助驾驶（实力）远超特斯拉，但事实是特斯拉先提概念，我们 随后跟进。从算力上论，国内所有车企AI（芯片）的显卡加起来都没有特斯拉Dojo多。"前不久，中国汽车工 业协会副总工程师王耀在行业论坛上的一番话，再次引发行业对智能辅助驾驶算力、算法的关注。 "特斯拉2016年启动自动驾驶自主研发，我们2018年跟进；特斯拉2021年发布FSD Beta，我们2023年推出类似 功能，这种时间差背后是算力基建的代差。"王耀提到的算力差距，在多大程度上影响智能辅助驾驶的发展？ 对于这个问题，行业专家和企业人士给出了自己的解答。 特斯拉追求算力的自研账 走纯视觉路线的特斯拉，对算力有着近乎疯狂的追求。这与其技术路线选择和数据驱动路径密不可分。 与ChatGPT、Deepseek等聊天机器人模型主要以图片、文本不同，自动驾驶系统的数据训练需要真实的驾驶视 频，相比之下占用空间大得多。海量的实时视觉信息处理、复杂场景的模拟预测、神经网络模型的快速迭代训 练，每一项都对自动驾驶的底层算力提出指数级增长需求。 尤其特斯拉的目标不只是FSD，还有Robotaxi及无人驾驶。在2023年二季度公司财 ...

Report Industry Investment Rating - The global economy in the macro and financial sector is rated as (bullish) [1] Core View - The global economy maintains an upward trend. China strengthens its domestic cycle, Asian exports are strong, and the terminal demand is considered strong. Market expects the Fed to cut interest rates in September 2025 and accelerate rate - cuts in 2026. China's comprehensive rectification of involution - style competition is expected to boost listed company performance. The European Central Bank has cut interest rates 8 times, Germany is expanding its military by 30%, and Meta plans to invest hundreds of billions of dollars in building large - scale data centers [1] Summary by Related Information AI Investment - AI investment covers three major directions: computing power, data, and downstream applications. The global data center construction in computing power is booming, and in application fields such as AI agents, intelligent driving, and intelligent terminals, China has many globally competitive enterprises [1] Eurozone Government Bonds - As the US dollar's safe - haven status is questioned, central banks around the world are increasing their allocation of eurozone government bonds, and the subscription ratio in eurozone government bond issuance has risen from 16% last year to 20% so far this year [1] Central Bank Independence - After JPMorgan Chase's Dimon took the lead, executives from Goldman Sachs, Bank of America, and Citigroup also voiced their support for the central bank's independent operation without White House intervention [1] Japanese Stock Market - With the upcoming Japanese election, Japan's $6.8 trillion stock market faces a "reckoning" moment. Polls suggest that the ruling coalition led by Prime Minister Ishiba Shigeru may lose its majority in the Senate election this weekend [1] Stablecoin Business - JPMorgan Chase's CEO Dimon and Citigroup's CEO Fraser said they will participate in the stablecoin business, and JPMorgan needs to accept stablecoins to keep up with competitors [1] Tariffs and Inflation - New York Fed President Williams expects tariffs to add about one percentage point to inflation from the second half of 2025 to 2026, although the current comprehensive data shows only a relatively limited impact of tariffs [1] xAI's Expansion - Musk's xAI is looking to expand its infrastructure in the Middle East to power its computing - intensive AI models with the region's cheap energy, abundant capital, and political goodwill [1] Trump's Tariff Plan - Trump plans to impose tariffs slightly higher than 10% on at least 100 countries, including those in Africa and the Caribbean [1] Economic Data - The final value of the US Markit Manufacturing PMI in June 2025 was 52.0, continuing to expand. China's PMI production index and new order index in June 2025 resumed expansion. Germany's industrial output in May 2025 increased by 1.2% month - on - month [1]