自动驾驶 · 272

Alibaba: The latest generation of AI processor Zhenwu M890 has been widely commercialized in more than 650 external customers

Comparatively, Alibaba announced that its latest generation AI processor, the Zhenwu M890, has been widely commercialized in more than 650 external customers. The chip has been used in more than 20 industries such as autonomous driving, the Internet, and financial services through Alibaba Cloud services, and has verified its ability to support full-process AI workloads.

2d ago

US media: Tesla Cybercab will be released in Austin this month, but the safety of autonomous driving is still questionable

Comparing news, Tesla plans to publicly launch the driverless taxi Cybercab as early as this month in Austin, Texas. The model is designed without a steering wheel and no pedals. Currently, Tesla employees have tested a fully driverless version on private roads in the company park. The report said that Cybercab is based on Tesla's FSD software, but its autonomous driving capability and safety are still being questioned. The US National Highway Traffic Safety Administration (NHTSA) is still investigating traffic rule compliance issues involving Tesla's FSD, and existing Robotaxi services are still equipped with human safety personnel in some markets. Cybercab, on the other hand, is designed to require no intervention from people in the car. Tesla plans to respond to emergencies through a remote operator, and has begun adding Starlink connectivity to the vehicle. The size of the data is also an issue. Tesla said in July that it needed to accumulate dedicated driving data for Cybercab. Up to now, Tesla's unsupervised Robotaxi has traveled a total of about 380,000 miles in 6 cities, while Waymo has completed more than 220 million miles of fully driverless miles on public roads since 2020. Additionally, Cybercab lacks traditional driving controls and may be restricted by US federal vehicle safety regulations. It is unclear whether Tesla is seeking regulatory exemptions, the report said. The Verge notes that as Cybercab nears launch, there is still a great deal of uncertainty about the autonomous driving safety and regulatory approvals required for its commercial operation.

4d ago

Einride to deploy 500 Tesla Semi trucks

According to the news, Swedish electric freight and autonomous driving technology company Einride announced that it will deploy 500 Tesla Semi trucks on SAGA's artificial intelligence platform to serve Amazon and other customers, covering the North American market. The deployment will be carried out in multiple phases, and is scheduled to begin in September, with financial support provided through a third-party financing scheme.

4d ago

SoftBank invests $200 million in construction robotics startup Gravis Robotics

Comparing news, according to Bloomberg, SoftBank Group invested $200 million in Swiss construction robotics startup Gravis Robotics to complete its Series A financing. Gravis was born out of the Zurich Federal Institute of Technology and mainly develops autonomous driving software and hardware for earthmoving construction equipment, which enables autonomous operation of heavy machinery and remote supervision of robot fleets.

5d ago
Jeff Dean's Last Conversation Before Leaving His Job: I Underestimated AI and Seen the Entrepreneur's Only Way to Live

Jeff Dean's Last Conversation Before Leaving His Job: I Underestimated AI and Seen the Entrepreneur's Only Way to Live

Source | InfoQ Compilation | Curated by Yu Qi | Tina A year ago, Google Chief Scientist Jeff Dean predicted at the AI Ascent 2025 Summit: By 2026, there may be AI systems that can work around the clock and are as capable as junior software engineers. A year later, six days ago, he admitted in an interview with YC that he had underestimated how fast AI is progressing. The model's ability to handle complex tasks grew much faster than he had anticipated at the time. So, according to Jeff Dean, how fast will AI move forward in the future? How can startups survive in an era where generic models continue to expand the boundaries of capabilities? Early this morning, this interview had a different weight. Jeff Dean announced that tomorrow will be his last day at Google. After working at Google for 27 years, this legendary engineer, known as the “programmer among programmers” in Silicon Valley and deeply involved in the construction of Google's system architecture and AI technology, co-founded Discovery Loop with long-term partners Sanjay Ghemawat, Oriol Vinyals, and Quoc Le, a public welfare company focusing on cutting-edge research in machine learning, science, and engineering. Google will continue to work with them as a founding investor and cloud computing partner. According to WIRED, this startup idea actually only surfaced a few weeks ago. In order to retain this core team, Alphabet CEO Sundar Pichai also tried to persuade them to “not lose the job card” during many meetings. But in the end, a few people decided to leave the big company system in exchange for the fun, speed, and freedom that only a startup can have. Screenshot from: https://x.com/JeffDean/status/2085035498222002595/photo/1Jeff Dean wrote in his farewell letter that he saw Google grow from a company of just 25 people to a tech giant with more than 190,000 employees. Today, Google has 13 products with over 1 billion users. From search, email, translation, and video to large-scale computing, autonomous driving, and AI systems, the technology he participated in building has spanned almost the entire evolution of Google. And one of the main reasons that prompted them to leave was precisely inertia, which is difficult for large companies to get rid of. As Oriol Vinyals said, within large organizations, driving any radical change requires overcoming layers of resistance; they want to do something different. What's interesting is that until now, the new company hasn't even had time to recruit people or rent an office. As for who will be the CEO, after a short pause within the team, everyone has their eyes on Jeff Dean — “I think it's me.” he said. As a result, this interview, published on the eve of Jeff Dean's departure, is like a focused judgment on the next stage of AI as he stands at a turning point in his career. On the program, he and YC partner Diana Hu discussed the paradigm shift in AI from “model centered” to “context engineering,” the huge opportunities that inference hardware is emerging, and how entrepreneurs can find a real living space worth sticking to in an age where generic models are becoming stronger and more applications may be directly incorporated by models. This article is based on a video compilation of this interview, edited by InfoQ. Too long without reading the Q edition: Last year you said 2026 would have AI with capabilities close to those of junior engineers. A year has passed, does this prediction punch you in the face? A: That's pretty accurate, but I've underestimated one thing: the model's ability to handle increasingly complex tasks is growing much faster than I expected. Moreover, this ability is spilling over into fields other than coding, and Agent-based systems are starting to really take off. Q: What are the bold predictions for 2027? A: The deep learning system will implement a fully automated problem decomposition and automated experiment cycle: split the problem into sub-problems, run experiments automatically, integrate the results, and obtain an improved system. And this doesn't just apply to machine learning; it can be used in any field of science and engineering with measurable goals. Q: In 2001, Google loaded the search index into memory,...

9d agoburnking#AI #Jeff Dean #Google
[Comparative Daily News Picks] Blackstone plans to build a financing plan of more than 36 billion US dollars for Anthropic; the Dow Jones Index closed up 907.53 points, and both the S&P 500 and NASDAQ reached new highs; SpaceX's revenue exceeded expectations, and the 100 billion unban wave is approaching this week; Polymarket is seeking financing at a valuation of more than 20 billion US dollars; the White House does not plan to disclose the details of the advanced AI model evaluation framework

[Comparative Daily News Picks] Blackstone plans to build a financing plan of more than 36 billion US dollars for Anthropic; the Dow Jones Index closed up 907.53 points, and both the S&P 500 and NASDAQ reached new highs; SpaceX's revenue exceeded expectations, and the 100 billion unban wave is approaching this week; Polymarket is seeking financing at a valuation of more than 20 billion US dollars; the White House does not plan to disclose the details of the advanced AI model evaluation framework

Daily AI · Crypto · Macro · Market News, Bitpush helps you set priorities ↓ AI · News [Blackstone plans to build a financing plan of over 36 billion US dollars for Anthropic] Comparing news, Blackstone Inc. is in preliminary communication with investors to assess the viability of raising a second hyperscale debt financing for Anthropic to support its plans to purchase Google AI chips. The financing plan reflects that AI companies are obtaining computing power resources through large-scale capital operations. As demand for big model training and inference continues to grow, AI companies are seeking new financing methods to support the construction of expensive computing infrastructure. People familiar with the matter revealed that a preliminary plan proposed by Blackstone reached at least $36 billion. The size of the funding, structure, and whether Blackstone will eventually dominate the deal are still under discussion, and details are subject to change. (Bloomberg) [Nvidia Launches Alpamayo 2 Super to Accelerate AI Development for Autonomous Driving] In comparison, Nvidia CEO Wong In-hoon posted an article on the X platform announcing the launch of Alpamayo 2 Super, a cutting-edge open reasoning model (open reasoning model) for the field of autonomous driving, adding that the development focus of the next phase of AI will expand from software intelligence to the field of robotics, and autonomous vehicles will become an important entry point into the robotics era. Hwang In-hoon said that Alpamayo 2 Super can not only “see” the road environment, but also understand and reason complex scenes and think before performing actions. It can be used as a core AI model for autonomous taxis, unmanned trucks, shuttle vehicles, delivery vehicles, agricultural robots, and future large-scale mobile robot systems. According to reports, Nvidia opened the model for commercial applications under the OpenMDW-1.1 license, and developers and enterprises can research, fine-tune, and deploy the model. Hwang In-hoon said that the open model will help improve the safety and reliability of autonomous driving systems. [AMD's 2026 Q2 revenue is US$11.5 billion, market forecast is US$11.284 billion] In comparison, AMD (AMD.O) has Q2 revenue of US$11.5 billion in 2026, with market expectations of US$11.284 billion, compared to US$7.685 billion for the same period last year. The estimated Q3 revenue for 2026 is US$12.7 billion to US$13.3 billion, and the market is expected to be US$12.51 billion. [SpaceX's revenue exceeds expectations, the 100 billion ban will be lifted this week] Comparing news, SpaceX announced Q2 revenue (7.8 billion US dollars), which exceeded Wall Street expectations (6.8 billion US dollars). The company previously raised 86 billion US dollars through the largest IPO in history. Since then, the stock price has dropped sharply due to post-IPO fluctuations and the overall sell-off of the AI sector. More than 1 trillion US dollars of market value has evaporated from its peak, and Musk's status as the world's first trillionaire has also been lost. Another factor adding uncertainty to the company is that the ban on stocks with a market value of more than 100 billion US dollars will be lifted later this week, which may put further downward pressure on stock prices. During the conference call, analysts are likely to ask Musk for details on the company's huge AI spending, Starlink's Internet business expansion plans, and how fast Starship can launch satellites and people. Crypto · Market [The Dow Jones Index closed up 907.53 points, and both the S&P 500 and NASDAQ reached new highs]. Comparing news, the Dow Jones Index closed up 907.53 points, or 1.71%, to 54,085.94 points; the S&P 500 rose 136.02 points, or 1.79%, to 7,736.52 points; the Nasdaq Composite Index rose 671.1 points, or 2.59%, to 26,584.99 points . [Polymarket seeks financing at a valuation of more than 20 billion US dollars] Comparing news, it is predicted that the market Polymarket will seek financing at a valuation of more than 20 billion US dollars. [New York Federal Court of Appeals Rejects SBF's Appeal and Conspiracy Convictions] In comparison, the New York Federal Court of Appeals on Tuesday upheld SBF's conviction for fraud and conspiracy after its crypto trading platform FTX went out of business. In 2024, S...

17d agoWendy#Compare Daily Picks

Hwang In-hoon announced the launch of Alpamayo 2 Super, Nvidia's open source inference model for autonomous driving to accelerate the robotics era

Comparing news, Hwang In-hoon posted an article on the X platform announcing the launch of Alpamayo 2 Super, a cutting-edge open source reasoning model created by Nvidia for the field of autonomous driving. Hwang In-hoon said that Alpamayo 2 Super can not only “see” the environment, but also understand and reason complex worlds, think before acting, and can be used as a core model for robot taxis, autonomous trucks, shuttle cars, delivery vehicles, agricultural machinery, and future large-scale mobile robot systems. The model will be open for commercial use based on OpenMDW-1.1, allowing developers and enterprises to inspect, fine-tune, and deploy the model. Hwang In-hoon said that the open model will help improve the safety and reliability of autonomous driving systems. He said that the next wave of artificial intelligence will belong to robotics, and autonomous vehicles will be an important starting point for this trend.

18d ago
Musk's empire is being besieged by the market on all sides

Musk's empire is being besieged by the market on all sides

来源｜字母榜作者｜袁心玥、王靖原标题｜马斯克两个核心资产，正在同时遭遇市场拷问7月29日，SpaceX股价盘中一度跌至107.01美元，较IPO发行价135美元下跌超过20%，较上市后的高点225.64美元更是回落超过50%。按照上市初期约2.6万亿美元的峰值估值计算，SpaceX市值已经蒸发约1.2万亿美元，相当于亏了一整个特斯拉。另一边，在特斯拉公布第二季度财报的第二天，特斯拉股价收跌14.52%，报319.69美元，创下一年多来最大单日跌幅；公司市值单日蒸发约2145亿美元（约合人民币1.5万亿元），成为当天标普500指数表现最差的成分股。截至目前，Tesla股价并未修复，仍较财报公布前低约18%。如果只看营收和交付数据，这份财报并不算差。第二季度，特斯拉全球交付汽车480,126辆，创下历史同期新高；营收达到282.4亿美元，同比增长26%，同样高于市场预期。但市场关注的重点已经发生了变化：由于AI、Robotaxi、Optimus以及相关基础设施投入持续增加，特斯拉第二季度资本开支达到58亿美元，自由现金流转为-11亿美元——这是公司两年多以来首次出现季度自由现金流为负。与此同时，公司维持全年超过250亿美元资本开支的指引。除了现金流之外，还有两项信息值得关注。根据特斯拉财报披露的累计付费载客里程拆分计算，Robotaxi车队今年第二季度新增付费运营里程约70万英里，低于第一季度约110万英里，环比下降约36%。本季度的财报电话会上，马斯克在谈到Cybercab部署计划时表示，特斯拉将首先利用配备方向盘和踏板的Cybercab测试车辆积累驾驶数据，再逐步扩大无方向盘版本的部署。这说明，尽管特斯拉目前已有约1000万辆在路车辆持续收集真实道路数据，但Cybercab作为新的车辆平台，并不能直接依赖现有车型的数据，仍需要完成针对自身平台的数据积累与验证。资本市场愿意给予特斯拉远高于传统汽车公司的估值，很大程度上并不在于它卖出了多少辆车，而是相信Robotaxi和Optimus能够带来下一阶段的增长。其中，Robotaxi最重要的逻辑之一，就是特斯拉能够依靠近1000万辆汽车持续采集真实道路数据，不断强化自动驾驶能力，最终形成一个持续自我强化的数据飞轮。而这次被暴露出的问题，让市场开始怀疑：特斯拉长期强调的数据飞轮，究竟能够在多大程度上支撑Robotaxi的未来？Robotaxi一直是特斯拉估值体系中最重要的故事之一。这套故事建立在一个非常简单的逻辑上：特斯拉拥有全球规模最大的智能汽车车队。截至今年第二季度，公司已有约1000万辆汽车在全球道路上持续运行，全球付费FSD（Full Self-Driving）用户接近150万。每天，大量真实道路数据不断回传，用于训练自动驾驶模型。马斯克也多次强调，特斯拉真正的竞争优势不是激光雷达，也不是高精地图，而是这支不断产生真实驾驶数据的车队。一旦数据积累足够，FSD能力会不断提升，最终形成Robotaxi网络，再吸引更多车辆加入、产生更多数据，形成持续强化的数据飞轮。按照这一叙事，随着真实道路数据不断增加，Robotaxi理应进入持续扩张阶段。然而，根据特斯拉披露的累计Robotaxi里程数据拆分，分析人士发现，二季度新增付费里程约70万英里，较第一季度约110万英里下降约36%。需要说明的是，特斯拉此次财报展示的是Robotaxi累计付费运营里程，而非季度运营数据——从图表上看，累计里程仍在持续上升，Robotaxi业务似乎保持增长。也就是说，累计数据掩盖了季度运营节奏的变化，而拆分后的季度数据，则呈现...

23d agoWendy#SpaceX #Tesla #Musk

Due to the surge in orders for Samsung's 2 nanometer process, manpower is tight, and Google plans to outsource the TPU I/O chip back-end design

Comparatively, Samsung Electronics is considering outsourcing the I/O chip back-end design work for Google's 10th generation TPU (codename Icefish). Google's 2 nanometer-based TPU consists of a Compute processor and an I/O chip — the former is expected to be manufactured by TSMC using the 1.4 nm process, while the latter is produced by Samsung on the 2 nm process and is responsible for data transmission between the Compute processor and HBM. Google is co-designing the chip with MediaTek and will be mass-produced as soon as 2028. However, Samsung's recent rapid influx of 2nm orders — in addition to Google and Tesla, it has taken over two customers, Anthropic and DeepX — has strained internal manpower. Previously, the back-end design of Tesla's 2nm autonomous driving chip was still done by Samsung itself. Potential outsourcing partners currently mentioned include ADTechnology, Gaonchips, and Alphachips. The first two companies have each invested in large-scale projects — ADTechnology is focusing on the 2-nm CPU project ADP620, which aims to exceed 1 trillion won in annual revenue between 2028 and 2029; Gaonchips is preparing to participate in the K-On-Device AI project of the Ministry of Industry, Trade and Energy of about 800 billion won to develop 5-nm ADAS chips in cooperation with Hyundai Motor Company and others. Since back-end design is essentially a service contract with low added value (usually at the level of tens of billions of won), far less than ASIC projects that cover everything from design to production (up to hundreds of billions to trillion won), the two companies are not positive, and only tend to undertake limited work to establish advanced process project performance. Alphachips, on the other hand, sees the Google TPU project as a growth driver and is more willing to participate. Industry sources pointed out that TSMC's orders in the 2 nm market, which cannot be digested due to capacity bottlenecks, are spilling over to Samsung, which is also the root cause of the latter's tight manpower.

38d ago