Tesla's New Questionable Cybercab Plans / A SpaceX Surprise ⚡️

以下是播客内容的总结，包括新闻、事实和观点： **介绍与特斯拉CyberCab战略** Dylan Loomis感谢新赞助人Bruce D.，并提出了一个关键问题：如果特斯拉的CyberCab车队有利可图，为什么特斯拉要将其出售给客户，特别是车队所有者，从而放弃部分利润？其理由是为了更快地扩大规模，使资产负债表多元化，并实现更快的投资回收期，尤其是在具体细节信息有限的情况下。 **特斯拉的车辆销售方式** 特斯拉的机器人出租车服务战略将与其过去的车辆销售（直销或租赁）类似，提供管理资产负债表和卸载运营基础设施的灵活性。这与Farzad的论点相符，即特斯拉生产CyberCab的速度将远快于其在没有方向盘和踏板的情况下部署这些车辆的速度。 **特斯拉的核心资产与外包组件** * **稀缺的专有资产**：FSD（自动驾驶软件和训练数据）、乘客应用程序、移动体验、新的动态定价算法、客户档案和数据、远程操作员基础设施、品牌以及客户体验。 * **非稀缺的外包组件**：实体车辆（“金属”）、停车场、充电、清洁、保险（包括索赔和文书工作）、许可证以及与当地市政府官员/法规的沟通。这些被认为是“繁琐的工作”。 **第三方车队所有权的好处** * **利润与风险**：虽然特斯拉100%所有权可以获得100%的利润，但它也承担了100%的折旧、保险费飙升、车库资本支出、利用率低的空闲时间和维护的风险。 * **财务优势**：出售/租赁CyberCab并从收入中分成，将一个多年、运营支出繁重的过程转化为更快的收入、更快的投资回收期，以及利用“他人的资金”进行扩展的轻资产收入流。特斯拉旨在将“最差的、价值最低的部分”外包出去。 * **规模化挑战**：机器人出租车业务需要高利用率、合理的保险、便宜的仓库/车库和低运营成本。目前，特斯拉、Waymo和Zoox的车队规模都很小（均低于10,000辆），缺乏规模经济。 * **基础设施密集型**：Ashok对所需的基础设施数量“相当惊讶”。特斯拉今年的资本支出约为250亿美元，自由现金流已转为负值。特斯拉没有无限的资金来扩大数十万辆机器人出租车的规模，同时将它们全部保留在其资产负债表上。 **历史背景与类比** * **宏图计划第二部分**：特斯拉最初计划让客户将他们的车辆部署到共享自动驾驶网络上，并用特斯拉自有车辆填补空缺，以避免自行资助车队。由于软件和法规问题，这尚未实现。 * **租赁策略**：特斯拉此前曾推出无买断选项的租赁服务，旨在收回车辆用于机器人出租车运营。大约两年前，这一计划被放弃，车辆作为二手车出售，以利用高转售价值和快速现金流，因为机器人出租车软件尚未准备好。 * **超级充电桩网络**：特斯拉现在向第三方出售超级充电桩硬件，以实现更快的规模化和运营卸载，尽管其内部有利可图，这表明存在多种可行途径。 **财务机制与风险规避** * 一个100%由特斯拉拥有的机器人出租车车队无法以工厂生产CyberCab的速度进行融资、许可和设置。 * **SpaceX类比**：SpaceX的计算资本支出通过租赁可在一年内收回，而像飞机库这样的实体资产则需要5-10年。“并非所有资本支出都是相同的。”第三方所有权可能将CyberCab的投资回收期从1-2年缩短到几天或几周。 * **车队成本**：CyberCab车队所有权涉及库存、保险资金、每个城市的车库和充电器资本支出、用于清洁、轮胎和损坏的营运资金，以及利用率不足带来的剩余风险。 * **未来灵活性**：当其资金状况发生变化时，特斯拉日后可以从运营商手中回购车辆。 * **行业趋势**：Waymo已经将其车库和车队工作推给Avis等合作伙伴，以将低价值任务外包，专注于自动驾驶软件和硬件。 * **物理网络**：特斯拉面向车队合作伙伴的在线表格列出了“移动出行枢纽和基础设施”，表明他们希望其他人来建设物理网络。 * **需求生成**：车队买家（机场、酒店、医院、大学、运动队、市政当局）将负责需求生成，从而将此任务从特斯拉身上卸下。 * **运营开销**：分区、许可、停车、充电、清洁和机场合同可以部分外包给车队运营商。 **特斯拉多元化战略总结** 特斯拉旨在为其机器人出租车网络的资本密集度、投资回收期和风险结构进行多元化。由于缺乏市场替代方案，特斯拉可能会战略性地在利用率高的地区拥有车队，并提出对自己有利的报价。该战略可能会随着未来的资金状况而改变。 **当前状态与先决条件** * 这些交易要实现规模化还需要大量时间。 * 需要V15软件的广泛可用。 * NHTSA对CyberCab的审计查询必须得到解决。 **机器人出租车测试与安全担忧** * “Dirty Tesla”因“不安全行为”（走到CyberCab前面）被禁止使用机器人出租车服务7天。 * David Moss拍摄的跑到街上测试CyberCab刹车能力的视频走红。 * 主持人Dylan同意“Larry”的观点，即这种测试毫无益处，应该停止，因为特斯拉已经进行了广泛测试，其自动驾驶车辆从未撞到人类。 **V15更新** * Ashok表示，V15将具备更早的危险预测、更快的反应时间，并显著提高安全性和防撞能力。 * Dylan认为，V15（或V15.1、.2、.3、.4）必须为特斯拉“解锁巨大的规模”，尤其是在没有方向盘/踏板且CyberCab产量正在提高的情况下。 **全球CyberCab推出与法规** * **日本**：特斯拉日本展示了CyberCab。《日经亚洲》指出，日本要求L4级自动驾驶车辆认证（无方向盘/踏板）和特定运营许可证，目前限于受限区域。特斯拉日本总裁表示，CyberCab使用与Model Y相同的FSD技术，今年在日本将FSD合法化是首要任务，希望CyberCab的推出不远了。Dylan认为FSD批准和CyberCab批准是“两回事”。 * **德国**：KBA（德国监管机构）正在考虑一条豁免途径，并在今年晚些时候的TCMV会议之前承认FSD的批准。德国当局此前曾表示有兴趣在缺乏司机的农村地区引入特斯拉机器人出租车，这可能加速FSD的进程。 **特斯拉杂项新闻** * 一位特斯拉车主成功推翻了一张超速罚单，他通过行车记录仪录像证明自己在70英里/小时的区域内行驶77英里/小时，而非警察雷达测得的82英里/小时。 **Cybertruck与Powerwall集成** * Cybertruck和Powerwall 3现在可以协同为房屋供电，将家庭备用电源延长三天以上（相当于九个额外的Powerwall）。这利用了Cybertruck的123千瓦时电池组除以Powerwall 3的13.5千瓦时容量。 * Powerwall 2的兼容性将于今年晚些时候推出。 * 这种与Powerwall协同工作的“PowerShare”是“缺失的一环”，因为一个Powerwall 3通常只能为房屋供电约一天。 * Cybertruck工程师Wes Morrill解释说，认证跨多个Powerwall世代的两个电网形成设备比预期花费了更长时间。 * 车辆到电网（V2G）功能已在德克萨斯州部分地区推出，允许拥有PowerShare家庭备用电源硬件的Cybertruck车主选择加入并在电网紧张时将多余能量反馈给电网以获得补偿。 * Dylan认为，由于Elon的政治立场，Cybertruck变得“政治色彩过于浓厚”，迄今为止，其销售数字一直是个“失败者”，他希望人们最终能理解其卓越之处。 **特斯拉Semi（欧洲）** * 特斯拉Semi将进入欧洲，下周将举行一场活动。 * 欧洲规格：在超过88,000磅的总组合重量（GCVW）下，续航里程为342英里。能耗为1.61千瓦时/英里。 * Megacharger速度：高达800千瓦（美国为1.2兆瓦）。 * 欧洲的满载额定值比美国更重（82,000磅GCVW），但效率与其保持一致。这是标准续航里程的等效值。 * 欧盟的计算：550公里 * 1千瓦时/公里 = 约550千瓦时可用电量，与美国的548千瓦时标准续航里程电池组相符。美国的标准续航里程Semi为325英里，略低于欧洲版本，这归因于不同的测试协议，而非电池容量更大。 * 德国特斯拉Semi网页列出了从2027年起开始交付，并宣传提供基本充电器以进行延长/过夜充电。 * 由于欧洲法规（美国仍要求传统后视镜），欧洲版本将传统侧视镜替换为摄像头监控系统（CMS）。 * 首批欧洲Semi预计将从内华达州运来，希望未来能在柏林超级工厂生产。 * 欧洲版本取消了连续的前灯条，采用了类似于Model Y的独立前灯。 **TerraFab建设** * Joe Tettmeyer的无人机视频显示，格莱姆斯县的TerraFab（而非奥斯汀研究工厂）已正式开始建设，主要地基工程、基础方形开挖和钢筋已到位。 **SpaceX新闻** * Brett Johnson宣布与一家未披露名称的公司达成了另一项计算租赁协议，每月11亿美元，附带90天退出条款。 * SpaceX有望在今年年底实现1000亿美元的经常性收入运行率，高于2025年全公司不到190亿美元的水平。 * 计算租赁协议的价格范围为每瓦计算能力30-50美元，SpaceX处于高端。Brett认为，他们通过扩展内部模型（Grok系列）将获得更多收入，以避免租赁协议带来的货币化贬值。 * 使用了多年的计算设备价格保持好于预期，表明芯片和租赁协议具有强大的残值。 * Elon Musk表示，Grok 4.7需要“再烹饪几天”，并指出在惩罚性响应长度、放弃困难任务和检查工作不够严谨方面存在问题。该版本预计将添加“大量”SpaceX公司训练数据，可能使其在工程方面表现出色。 **The Boring Company** * 在D轮融资中筹集了30亿美元，由阿联酋牵头，公司估值为230亿美元。 * Dylan认为，CyberCab在这些隧道中运营“只是时间问题”。 **个人推荐与结束语** * Dylan推荐了书籍《好消息孩子们，给幼儿的福音》，称其高质量，配有可爱的插图和精彩的故事。 * 他还链接了他新买的椅子。 * 特斯拉股价收于365.44美元（上涨0.52%），SpaceX收于151.21美元（上涨2.04%），NDX上涨0.91%。 * 他鼓励大家查看Whisperflow及其笔记功能，强调他近10万字的记录和连续58天不间断使用。

Here's a summary of the podcast, including news, facts, and opinions: **Introduction & Tesla CyberCab Strategy** Dylan Loomis thanks new patron Bruce D. and addresses a critical question: If Tesla's CyberCab fleet is profitable, why would Tesla sell it to customers, specifically fleet owners, giving up some profitability? The rationale is to scale faster, diversify the balance sheet, and achieve quicker payback periods, especially given limited information on specifics. **Tesla's Approach to Vehicle Sales** Tesla's strategy for robo-taxi service will be similar to its past vehicle sales (direct or lease), offering flexibility in managing its balance sheet and offloading operational infrastructure. This connects to Farzad's argument that Tesla will manufacture CyberCabs much faster than it can deploy them without a wheel and pedals. **Tesla's Core Assets vs. Outsourced Components** * **Scarce, Proprietary Assets:** FSD (autonomous software and training data), rider app, mobile experience, new dynamic pricing algorithm, customer profiles and data, remote operator infrastructure, brand, and customer experience. * **Non-Scarce, Outsourced Components:** The physical cars ("metal"), parking lots, charging, cleaning, insurance (including claims and paperwork), permits, and conversations with local city officials/regulations. These are considered "grunt work." **Benefits of Third-Party Fleet Ownership** * **Profit vs. Risk:** While 100% Tesla ownership captures 100% profit, it also absorbs 100% depreciation, insurance spikes, depot capital expenditures, risk of underutilized windows, and maintenance. * **Financial Advantages:** Selling/leasing CyberCabs and taking a cut of revenue converts a multi-year, OpEx-heavy process into quicker revenue, faster payback, and an asset-light income stream that scales with "other people's money." Tesla aims to outsource the "worst, lowest value parts." * **Scaling Challenges:** Robo-taxi businesses require high utilization, reasonable insurance, cheap depots, and low operating costs. Currently, Tesla, Waymo, and Zoox have small fleets (under 10,000 units), lacking economies of scale. * **Infrastructure Intensive:** Ashok was "quite surprised" by the amount of infrastructure needed. Tesla's CapEx is around $25 billion this year, and free cash flow has flipped negative. Tesla doesn't have unlimited funds to scale hundreds of thousands of robo-taxis while keeping them all on its balance sheet. **Historical Context and Parallels** * **Master Plan Part 2:** Tesla initially planned for customers to deploy their vehicles onto a shared autonomous network, filling gaps with Tesla-owned vehicles, to avoid financing the fleet itself. This hasn't come to fruition due to software and regulations. * **Leasing Strategy:** Tesla previously had leases with no buyout option, intending to take cars back for robo-taxi operations. Around two years ago, this was abandoned, and cars were sold as used vehicles to leverage high resale values and quick cash, as the robo-taxi software wasn't ready. * **Supercharger Network:** Tesla now sells Supercharger hardware to third parties for faster scaling and offloading operations, despite internal profitability, showing multiple viable paths. **Financial Mechanics and Risk Mitigation** * A 100% owned Tesla RoboTaxi fleet cannot be financed, permitted, and set up at the speed the factory can produce CyberCabs. * **SpaceX Analogy:** SpaceX's compute CapEx has under one-year payback periods with leasing, compared to 5-10 years for physical assets like hangars. "Not all CapEx is the same." Third-party ownership could shift CyberCab payback from 1-2 years to days or weeks. * **Fleet Costs:** CyberCab fleet ownership involves inventory, insurance capital, depot and charger CapEx in every city, working capital for cleaning, tires, and damage, and residual risks from underutilization. * **Future Flexibility:** Tesla could buy back vehicles from operators later when its capital situation changes. * **Industry Trend:** Waymo already pushes depot and fleet work to partners like Avis to outsource low-value tasks and focus on self-driving software and hardware. * **Physical Network:** Tesla's online form for fleet partners lists "mobility hubs and infrastructure," indicating they want others to build the physical network. * **Demand Generation:** Fleet buyers (airports, hotels, hospitals, colleges, sports teams, municipalities) will handle demand generation, offloading this from Tesla. * **Operational Overhead:** Zoning, permitting, parking, charging, cleaning, and airport contracts can be partially offloaded to fleet operators. **Summary of Tesla's Diversified Strategy** Tesla aims to diversify its capital intensity, payback periods, and risk structure for the robo-taxi network. Tesla will likely strategically own fleets in high-utilization areas and make offers favorable to itself due to a lack of market alternatives. The strategy can change with future capital conditions. **Current Status and Prerequisites** * Significant time is needed before these deals scale. * Requires V15 software to be broadly available. * NHTSA's audit query into the CyberCab must be resolved. **Robo-Taxi Testing and Safety Concerns** * "Dirty Tesla" was banned from robo-taxi service for 7 days for "unsafe conduct" (stepping in front of a CyberCab). * David Moss's video of running into the street to test a CyberCab's stopping ability went viral. * Dylan, the host, agrees with "Larry" that this type of testing is unhelpful and should stop, as Tesla has extensively tested it, and its autonomous vehicles have never run over a human. **V15 Update** * Ashok stated V15 will have earlier hazard prediction, faster reaction time, and significantly better safety and collision avoidance. * Dylan believes V15 (or V15.1, .2, .3, .4) must "unlock some serious scale" for Tesla, especially with no wheel/pedals and CyberCab production ramping. **Global CyberCab Rollout & Regulations** * **Japan:** Tesla Japan showcased the CyberCab. Nikkei Asia notes Japan requires Level 4 autonomous vehicle certification (no wheel/pedals) and a specific operation permit, currently limited to restricted areas. Tesla Japan's president stated the CyberCab uses the same FSD tech as Model Y, and legalizing FSD in Japan this year is a top priority, hoping the CyberCab launch isn't distant. Dylan believes FSD approval and CyberCab approval are "two very different things." * **Germany:** The KBA (German regulatory authority) is considering an exemption route and recognizing FSD's approval before a TCMV meeting later this year. German authorities previously expressed interest in Tesla robo-taxis for rural areas lacking drivers, potentially fast-tracking FSD. **Miscellaneous Tesla News** * A Tesla owner successfully overturned a speeding ticket by using dashcam footage to prove they were going 77 mph in a 70 mph zone, not 82 mph as clocked by police radar. **Cybertruck & Powerwall Integration** * The Cybertruck and Powerwall 3 can now power a house together, extending home backup by over three days (equivalent to nine additional Powerwalls). This uses the Cybertruck's 123 kWh battery pack divided by the Powerwall 3's 13.5 kWh capacity. * Powerwall 2 compatibility is coming later this year. * This "PowerShare" working in tandem with Powerwall is the "missing piece" as one Powerwall 3 typically only powers a home for about a day. * Cybertruck engineer Wes Morrill explained that certifying two grid-forming devices across multiple Powerwall generations took longer than expected. * Vehicle-to-grid (V2G) functionality was launched in parts of Texas, allowing Cybertruck owners with PowerShare home backup hardware to opt-in and discharge excess energy back to the grid during stress events for compensation. * Dylan feels the Cybertruck has become "too politically charged" due to Elon's politics, and to date, it's been a "dud" in sales numbers, hoping people will eventually understand its exceptionalism. **Tesla Semi (Europe)** * The Tesla Semi is coming to Europe, with an event next week. * European specs: 342 miles range at over 88,000 pounds Gross Combination Weight (GCVW). Energy consumption at 1.61 kWh/mile. * Megacharger speeds: up to 800 kW (compared to 1.2 MW in the US). * Europe's fully loaded rating is heavier than the US (82,000 lbs GCVW), but efficiency is in line. This is a standard range equivalent. * EU math: 550 kilometers * 1 kWh/km = ~550 kWh usable, aligning with the US 548 kWh standard range pack. The US standard range semi has 325 miles, slightly lower than the European version, due to different test protocols, not a bigger battery. * Germany's Tesla Semi webpage lists deliveries from 2027 onwards and touts availability of the base charger for extended/overnight charging. * The European variant replaces traditional side view mirrors with a Camera Monitoring System (CMS) due to European regulations (US still requires traditional mirrors). * Initial European Semis are expected from Nevada, with hopes for Giga Berlin production later. * The European version removes the continuous front light bar, featuring standalone headlights similar to the Model Y. **TerraFab Construction** * Joe Tettmeyer's drone video shows construction has officially started on the TerraFab in Grimes County (not the Austin research fab), with main foundation work, footing square excavations, and rebar. **SpaceX News** * Brett Johnson announced another compute leasing deal with an unnamed company for $1.1 billion per month, with a 90-day out clause. * SpaceX is on track for a $100 billion recurring revenue run rate by the end of this year, up from less than $19 billion company-wide in 2025. * Compute leasing deals range from $30-$50 per watt of compute capacity, with SpaceX at the higher end. Brett believes they will earn more scaling internal models (Grok family) to avoid monetization degradation from lease deals. * Multi-year-old compute is holding its rates better than expected, indicating strong residual values for chips and lease deals. * Elon Musk said Grok 4.7 needs "a few more days to cook," noting issues with penalized response length, giving up on hard tasks, and insufficient rigor in checking work. This version is expected to add a "massive amount" of SpaceX company training data, potentially making it strong in engineering. **The Boring Company** * Raised $3 billion in Series D funding led by the UAE, valuing the company at $23 billion. * Dylan believes it's "only a matter of time" until the CyberCab operates in these tunnels. **Personal Recommendation & Closing** * Dylan recommends the book "Good News Children, the Gospel for Toddlers," describing it as high-quality with cute drawings and a well-told story. * He also links his new chair. * Tesla stock closed at $365.44 (up 0.52%), SpaceX at $151.21 (up 2.04%), and the NDX was up 0.91%. * He encourages checking out Whisperflow and its note-taker feature, highlighting his nearly 100,000 words and 58-day streak.