Elon Proves Them Wrong Again. $1 Trillion Terafab Is NEXT.

以下是播客新闻内容的中文翻译： * **特斯拉 Cybercab 在奥斯汀：** 据报道，“特斯拉 Cybercab”（完全自动驾驶、安全、纯视觉的机器人出租车）正在“涌入奥斯汀市中心”。 * **Cybercab 正式发布：** Cybercab 的正式发布定于2026年9月3日，这距离播客发布日期还不到一天。 * **驳斥专家对自动驾驶的说法：** 播客强调，“所谓的专家”曾声称使用纯视觉系统实现安全、经济的自动驾驶汽车是不可能的，并且特斯拉将需要一个包含激光雷达、雷达、GADAR、盲文、声纳、烟雾信号、摩尔斯电码、X射线、红外线、紫外线和伽马射线探测器在内的综合传感器套件，而这些说法已被证明是错误的。这被列为埃隆·马斯克证明怀疑者是错误的第328个例子。 * **万亿美元的 TerraFab 项目：** TerraFab 项目旨在每年生产一太瓦的芯片，预计需要万亿美元的资本。该项目被介绍为专家们错误的第329个例子，因为他们目前认为其不可能实现或不会成功。 * **SpaceX 作为 TerraFab 的主要资金来源：** SpaceX 被认为是庞大 TerraFab 项目的潜在“金主”和主要出资方。 * **人工智能投资对其他融资的影响：** 人工智能领域的高“资本回报率”（IRR）预计将“耗尽”其他大型政府或基础设施项目的融资。 * **路易斯安那州1000亿美元的星舰发射设施：** 一项1000亿美元的投资正用于路易斯安那州的一个新星舰发射设施，旨在实现“每年数千次甚至数万次星舰发射”。 * **Figure 发布 Index：** Figure 正在发布“Index，最大的机器人训练数据数据库”。 * **路易斯安那州星际基地发射频率目标：** 仅路易斯安那州基地，目标是每年约10,000次星舰发射。 * **路易斯安那州项目的规模：** 该项目被描述为有史以来最大的基础设施项目之一，可与美国高速公路系统（2026年美元价值超过7000亿美元）相媲美，其雄心壮志超越了巴西和中国的巨型水电大坝项目。 * **路易斯安那州设施的战略位置：** 选择路易斯安那州是因为它能够为人工智能卫星“正南”方向飞越美利坚湾（前身为墨西哥湾）进行发射，以及其丰富的天然气供应，可用于转化为甲烷火箭燃料。 * **SpaceX 的高资本回报率：** SpaceX 预计其业务将获得“极高的资本回报率”，其中星链（Starlink）预计在第10次火箭发射后实现100%的年化回报。 * **人工智能卫星回报预测：** 第100次人工智能卫星发射的回报预计将达到“高二十位数”（指20%以上），并在第1,000次人工智能卫星发射时超过星链的回报。 * **星链和人工智能计算的快速回本周期：** 星链卫星部署可在一年内回本，地基人工智能计算的回报周期不到一年，而规模化轨道人工智能计算的回报周期也将少于一年。 * **SpaceX 的人工智能计算收入潜力：** SpaceX 第二季度财报电话会议透露，计划上线大量人工智能计算。按照当前现货价格，这每年可能产生“数千亿美元”的收入（例如，10吉瓦以每瓦40美元计算，每年总计4000亿美元），有望在几年内仅凭人工智能计算收入就为 TerraFab 的全部资本支出提供资金。 * **芯片成为卫星主要成本：** 计算机芯片的成本正成为“卫星实际成本的主要部分”。 * **黄仁勋对 XAI 的赞扬：** 英伟达（NVIDIA）首席执行官黄仁勋公开赞扬埃隆·马斯克和 XAI 团队能够“比其他任何人都更快地”将人工智能计算投入使用。 * **英伟达和 SpaceX AI 共同设计的芯片：** 英伟达和 SpaceX AI 已经“共同为 StarMind 人工智能卫星设计了一款新芯片”。 * **英伟达为 SpaceX AI 提供融资（阴谋论）：** 播客推测，鉴于 SpaceX 芯片的快速变现能力，英伟达可能很快会为 SpaceX AI 的芯片采购提供融资，甚至为 StarMind 星座的资本支出做出贡献。 * **TerraFab 在美国为英伟达生产芯片（阴谋论）：** 有人建议，TerraFab 最终可能不仅为 SpaceX AI 的需求，还将在美国本土为英伟达生产芯片。 * **台湾集中风险：** 全球90%以上最新、最好的AI芯片，包括英伟达的芯片，都在台湾由台积电（TSMC）生产，由于“略带敌意的邻居”（中国），这造成了“岌岌可危的局面”。 * **TerraFab 的国家安全和经济利益：** TerraFab 项目被认为对国家安全和经济繁荣至关重要，因为它能大幅“降低”全球经济对台湾芯片生产的依赖风险，可能导致美国境内技术公司和政府“近乎一致的支持”。

Here are the news items from the podcast: * **Tesla Cybercabs in Austin:** Fully autonomous, safe, vision-only robotaxis, referred to as "Tesla cybercabs," are reportedly "flooding downtown Austin." * **Official Cybercab Launch:** The official launch of the cybercab is scheduled for September 3rd, 2026, which is less than a day away from the podcast's publication date. * **Refutation of Expert Claims on Autonomy:** The podcast highlights that "so-called experts" who previously stated it was impossible to have safe, affordable, autonomous vehicles using a vision-only system, and that Tesla would require a comprehensive sensor suite including LiDAR, radar, GADAR, Braille, sonar, smoke signals, Morse code, X-ray, infrared, ultraviolet, and gamma ray detectors, have been proven wrong. This is cited as example #328 of Elon Musk proving doubters wrong. * **The Trillion-Dollar TerraFab Project:** The TerraFab project, which aims to produce a terawatt of chips annually and is expected to require a trillion dollars in capital, is introduced as example #329 of experts being wrong, as they currently deem it impossible or unsuccessful. * **SpaceX as Primary Funder of TerraFab:** SpaceX is identified as the likely "sugar daddy" and primary funder for the massive TerraFab project. * **AI Investment Impact on Other Financing:** The high "return on capital" (IRR) in AI is projected to "dry up financing" for other large-scale government or infrastructure projects. * **$100 Billion Starship Launch Facility in Louisiana:** A $100 billion investment is being made into a new Starship launch facility in Louisiana, intended for "thousands and thousands of Starship launches annually." * **Figure Index Release:** Figure is releasing "Index, the largest database of robot training data." * **Starbase, Louisiana Launch Cadence Goal:** The goal for the Louisiana site alone is approximately 10,000 Starship launches per year. * **Scale of Louisiana Project:** This project is described as one of the largest infrastructure projects of all time, comparable to the U.S. highway system (over $700 billion in 2026 dollars), and ahead of giant hydropower dam projects in Brazil and China in terms of ambition. * **Strategic Location of Louisiana Facility:** The Louisiana location is chosen for its ability to launch "due south" over the Gulf of America (formerly Gulf of Mexico) for AI satellites, and its abundant natural gas supply needed to convert into methane rocket fuel. * **SpaceX's High Return on Capital:** SpaceX anticipates "extraordinarily high return on capital" for its businesses, with Starlink projected to achieve 100% annualized returns by its 10th rocket launch. * **AI Satellite Return Projections:** Returns for the 100th AI satellite launch are expected to be in the "high 20s," exceeding Starlink's returns by the 1,000th AI satellite launch. * **Rapid Payback Period for Starlink and AI Compute:** Starlink satellite deployments pay for themselves within a year, earth-based AI compute returns are under one year, and orbital AI compute at scale will also see returns in less than a year. * **SpaceX's AI Compute Revenue Potential:** SpaceX's Q2 earnings call revealed plans to bring significant AI compute online. At current spot prices, this could generate "multiple hundreds of billions of dollars per year" (e.g., 10 gigawatts at $40/watt totals $400 billion/year), potentially funding the entire TerraFab CapEx from AI compute revenue alone within a few years. * **Chips as Major Satellite Cost:** The cost of computer chips is becoming the "majority actual part of the cost of the satellite." * **Jensen Huang's Praise for XAI:** NVIDIA CEO Jensen Huang has publicly praised Elon Musk and the XAI team for their ability to get AI compute online "much faster than anyone else." * **NVIDIA and SpaceX AI Co-designed Chip:** NVIDIA and SpaceX AI have "co-designed a new chip for the StarMind AI satellites." * **NVIDIA Financing for SpaceX AI (Conspiracy Theory):** The podcast speculates that NVIDIA may soon offer financing for SpaceX AI's chip purchases and even contribute to the capital expenditure of the StarMind constellation, given SpaceX's rapid monetization of chips. * **TerraFab Producing Chips for NVIDIA in US (Conspiracy Theory):** It's suggested that TerraFab could eventually produce chips domestically in the US for NVIDIA, beyond just SpaceX AI's needs. * **Taiwan Concentration Risk:** Over 90% of the world's newest, best AI chips, including NVIDIA's, are produced in Taiwan by TSMC, creating a "precarious situation" due to a "slightly hostile neighbor" (China). * **National Security and Economic Stakes of TerraFab:** The TerraFab project is deemed critical for national security and economic prosperity, as it could massively "de-risk" the global economy's reliance on Taiwan for chip production, potentially leading to "near-unanimous support within the US" from technology companies and governments.