首页  >>  来自播客: Electrified 更新   反馈  

Electrified - Tesla Finally Adding More Robotaxis & the SpaceX AI Satellite V1: Running the Numbers ⚡️

发布时间:   原节目
以下是视频内容的中文翻译,涵盖了所有讨论的新闻项目: **主持人免责声明:** 尽管迪伦·卢米斯(Dylan Loomis)患有严重过敏,他仍在主持《电动化》(Electrified)节目。 **特斯拉机器人出租车与自动驾驶更新:** * **德克萨斯州机器人出租车VINs:** 德克萨斯州汽车运输公司网站目前列出了59个特斯拉用于其机器人出租车业务的车辆识别码(VINs),两周前这一数字刚超过40个。所有车辆均为Model Y。这表明特斯拉可能正在部署,尽管并非所有车辆都已激活或处于无人监督状态。 * **德克萨斯州投诉:** 一项针对特斯拉的投诉被列出,但已因“无管辖权”而结案。迪伦已根据《公共信息法案》提交了请求以获取更多详细信息。 * **休斯顿Cybercab(赛博出租车)目击事件:** S.E.罗宾逊对休斯顿车场的航拍显示有21辆Cybercab和许多Model Y。迪伦重申,在看到没有方向盘或踏板的Cybercab之前,对它们的部署期望仍然很低。 * **达拉斯Cybercab目击事件:** 一段达拉斯车场的视频显示有数十辆Cybercab排成一列。 * **美国测试范围:** Cybercab目前已在美国大约20个不同的州进行测试。 * **亚利桑那州急救响应计划:** 一份日期为4月24日的文件披露了特斯拉在亚利桑那州的急救响应计划,详细说明了在钱德勒、吉尔伯特、格伦代尔、马里科帕县、梅萨、天堂谷、凤凰城、斯科茨代尔和坦佩(均位于马里科帕县,人口约480万)进行SAE L4级自动驾驶操作的细节。特斯拉的网站上也包含这些计划。 * **亚利桑那州上线预期:** 根据机器人出租车追踪器,凤凰城是亚利桑那州唯一被官方明确提及的城市,这表明它将是该州的首个上线地点,并有望迅速扩展。 * **第一季度时间表未达标:** 距离第一季度幻灯片中设定的在五个不同城市上线的目标只剩下20天,迪伦认为这一目标不太可能实现。 **监督FSD进展:** * **荷兰安全数据:** 特斯拉报告称,在荷兰道路上,监督FSD比人工驾驶安全三倍多,碰撞次数减少3.5倍。其他指标虽然难以量化,但包括:自动紧急制动事件减少近15倍,急加速减少8.8倍,急刹车减少7.3倍,急转弯减少8倍。 * **公路与非公路安全(荷兰):** 在高速公路上,FSD安全3.4倍(行驶1660万公里,零碰撞);在非高速公路上,FSD安全1.6倍(行驶700万公里,发生三次碰撞)。从4月10日到6月5日,FSD在荷兰的总行驶里程达到2360万公里。非高速公路上的碰撞很可能是低速停车场事故。 * **丹麦批准:** 埃隆·马斯克确认,监督FSD已在丹麦获得官方批准,预计将很快推出。 * **欧洲累计批准:** 监督FSD现已在荷兰、立陶宛、爱沙尼亚和丹麦获得批准,覆盖约21.5万辆特斯拉(累计数量,需经AI4/硬件3过滤)。 * **欧盟范围投票展望(TCMV 6月30日):** 汽车技术委员会(TCMV)6月30日会议议程显示,将“继续讨论”荷兰第39号法案的授权,但未提及投票。迪伦预计不会有投票,因为正式法案通常会在两周前起草并分发。 * **未来欧盟批准:** 下一个可能进行全欧盟范围投票的窗口期将是10月。这意味着监督FSD的批准路径目前可能仍将是逐国进行。欧盟范围的批准可能会推迟到2027年第四季度进行投票,实际软件推出将推迟到2027年初。 * **临时批准风险:** 丹麦一家媒体澄清,目前的批准都是临时性的。如果欧盟委员会拒绝该系统,荷兰的临时批准(以及随后的丹麦批准)将在六个月后失效。如果欧盟正式拒绝FSD,这可能导致所有当前的临时批准失效。欧盟委员会的长期拖延也可能意味着这种逐国批准的零散状态将无限期持续下去。 **特斯拉业务与产品新闻:** * **美国贸易代表访问:** 美国贸易代表詹姆斯·格里尔(James Greer)本周将访问特斯拉弗里蒙特工厂,作为促进美国再工业化制造之旅的一部分。 * **关税担忧:** 特斯拉此前在2023年3月致信格里尔,警告美国出口商容易受到报复性关税的影响,并强调在国内采购锂电池组件的困难。特斯拉预计将在访问期间重申这些担忧。 * **特斯拉Semi卡车欧洲扩张:** 特斯拉已在德国帕斯多夫(慕尼黑附近)发布了特斯拉Semi卡车业务发展经理的招聘信息,这标志着Semi卡车在欧洲的商业基础工作正在进行。然而,预计最早也要到2027年才能部署/销售。 * **Model 3 后驱版效率:** 埃德蒙兹(Edmunds)在其实际续航测试中,赞扬特斯拉Model 3后轮驱动版是其“效率冠军”。它以393英里的续航里程(+8.3%)超出了EPA的估算(363英里),并且比其EPA评级效率高得多,达到了21.7千瓦时/百英里(4.61英里/千瓦时),而EPA评级为25千瓦时/百英里(4英里/千瓦时),效率高出13.2%。 * **特斯拉车辆寿命:** IC Cars对1.74亿辆汽车进行的一项研究发现,特斯拉达到25万英里的可能性是斯巴鲁的两倍。特斯拉总体排名第六,有4.6%的可能性(与GMC并列),领先于凯迪拉克、梅赛德斯、宝马和保时捷等豪华品牌。在豪华品牌中,特斯拉仅次于雷克萨斯和讴歌,排名第三。 **SpaceX AI-1卫星与星舰讨论:** * **SpaceX采访:** 埃隆·马斯克和SpaceX首席财务官的两份新一般性采访发布。 * **AI-1卫星描述:** 视频展示了AI-1卫星的图像,突出了其散热器、计算模块和巨大的太阳能电池板。它宽70米(约75码),高20米(约20码)。 * **马斯克的雄心勃勃目标:** SpaceX的目标是到2027年底实现1吉瓦(年化)的轨道人工智能计算能力,迪伦认为这过于乐观。 * **AI-1功率与重量:** 每颗AI-1轨道卫星在峰值时可提供150千瓦(0.15兆瓦)功率,持续/平均功率为120千瓦(0.12兆瓦)。它们重约2吨(4,400磅),与一辆Model Y相当。 * **与Colossus 1对比:** 要匹配一个500兆瓦的数据中心(如Colossus 1),需要3,333颗AI-1卫星(使用峰值功率)或4,000颗卫星(使用持续功率)。 * **星舰V3发射要求:** 星舰V3预计每次飞行可运载100吨(约50颗AI-1卫星)。要将3,333颗AI-1卫星(6,666吨)送入轨道以实现500兆瓦的等效功率,需要67次星舰发射。要达到1吉瓦的年化目标,这意味着大约150次星舰发射(每月超过12次发射)。 * **星舰发射频率挑战:** 星舰在三年内发射了12次(2025年5次,今年因V3升级仅1次),但尚未成功入轨。大多数发射只携带了少量有效载荷;最大载荷为45吨(22个模拟星链单元)。 * **星链V3背景:** 仅有两颗经过改装的星链V2(模拟V3硬件)发射用于隔热罩测试;目前尚无实际的星链V3卫星投入服务。 * **乐观与现实:** 迪伦总结道,考虑到目前的进展,到2027年底实现1吉瓦的目标,需要如此高的星舰发射频率和新工厂的建设,是高度乐观的。 **市场表现与赞助商:** * **特斯拉股票:** 特斯拉股票收盘价为396.68美元,下跌3%,交易量高于平均水平14%。纳斯达克100指数(NDX)下跌1.5%。 * **赞助商:** Surfshark,重点介绍了他们的新Dowsos VPN协议,该协议具备后量子安全,提供专用加密隧道,速度提高30%,经过独立审计,并包含分流隧道功能。 迪伦希望明天能感觉好些,并感谢他的Patreon支持者。

Here's a comprehensive summary of the video transcription, including every news item discussed: **Host's Disclaimer:** Dylan Loomis is hosting Electrified despite suffering from severe allergies. **Tesla Robotaxi & Autonomy Updates:** * **Texas Robo-taxi VINs:** The Texas Motor Carrier website now lists 59 Tesla VINs for its robo-taxi business, up from just over 40 two weeks ago. All are Model Ys. This indicates potential deployment, though not all are active or unsupervised yet. * **Texas Complaint:** One complaint against Tesla is listed but closed with "no jurisdiction." Dylan has submitted a Public Information Act request for more details. * **Houston Cybercab Sightings:** S.E. Robinson's flyover of the Houston lot showed 21 cybercabs and many Model Ys. Dylan reiterates low expectations for cybercab deployment until they are seen without steering wheels or pedals. * **Dallas Cybercab Sightings:** A video of the Dallas lot revealed dozens of cybercabs lined up. * **US Testing Reach:** Cybercabs have now been spotted testing in approximately 20 different U.S. states. * **Arizona First Responder Plan:** A document dated April 24th revealed Tesla's first responder interaction plan for Arizona, detailing SAE Level 4 autonomy operations in Chandler, Gilbert, Glendale, Maricopa County, Mesa, Paradise Valley, Phoenix, Scottsdale, and Tempe (all within Maricopa County, population ~4.8 million). Tesla's website also contains these plans. * **Arizona Launch Expectations:** Based on the Robotaxi tracker, Phoenix is the only city officially called out for Arizona, suggesting it will be the first launch location there, with quick expansion hoped for. * **Q1 Timeline Miss:** With 20 days left to launch in five different cities to meet its Q1 slide deck timeline, Dylan believes this goal is unlikely. **Supervised FSD Developments:** * **Netherlands Safety Data:** Tesla reported that supervised FSD is over three times safer than manual driving on Dutch roads, with 3.5 times fewer collisions. Other metrics, though tough to quantify, include: nearly 15 times fewer automatic emergency braking events, 8.8 times less harsh acceleration, 7.3 times less harsh braking, and 8 times fewer hard swerves. * **Highway vs. Non-Highway Safety (Netherlands):** FSD was 3.4 times safer on highways (16.6 million km, zero collisions) and 1.6 times safer on non-highway roads (7 million km, three collisions). Total FSD kilometers in the Netherlands from April 10th to June 5th reached 23.6 million. Non-highway collisions are likely low-speed parking lot incidents. * **Denmark Approval:** Elon Musk confirmed that supervised FSD has been officially approved in Denmark, with rollout expected soon. * **Cumulative European Approvals:** Supervised FSD is now approved in the Netherlands, Lithuania, Estonia, and Denmark, covering roughly 215,000 Teslas (cumulative, needing AI4/Hardware 3 filtering). * **EU-Wide Vote Outlook (TCMV June 30th):** The agenda for the Technical Committee of Motor Vehicles (TCMV) meeting on June 30th shows "continuation of discussions" for the Netherlands' Act 39 authorization but no mention of a vote. Dylan expects no vote, as a formal act would typically be drafted and circulated two weeks prior. * **Future EU Approval:** The next possible window for a block-wide vote would be October. This means the path to approval for supervised FSD will likely remain country-by-country for now. EU-wide approval might be delayed until Q4 2027 for the vote, pushing actual software rollout into early 2027. * **Provisional Approval Risks:** An outlet in Denmark clarified that current approvals are provisional. If the EU Commission rejects the system, the Dutch provisional approval (and subsequently, Denmark's) would become invalid after six months. This could lead to all current provisional approvals lapsing if the EU formally rejects FSD. A prolonged delay by the EU Commission could also mean the country-by-country patchwork continues indefinitely. **Tesla Business & Product News:** * **US Trade Representative Visit:** James Greer, the US Trade Representative, is scheduled to visit Tesla's Fremont factory this week as part of a manufacturing tour promoting US re-industrialization. * **Tariff Concerns:** Tesla previously submitted a letter to Greer in March 2023, warning that US exporters are vulnerable to retaliatory tariffs and highlighting difficulties in sourcing lithium battery components domestically. Tesla is expected to reiterate these concerns during the visit. * **Tesla Semi European Expansion:** Tesla has posted a job listing for a Business Development Manager for the Tesla Semi in Parsdorf (near Munich), Germany, signaling commercial groundwork for the Semi in Europe. However, deployment/sales are not expected until 2027 at the earliest. * **Model 3 RWD Efficiency:** Edmunds lauded the Tesla Model 3 Rear-Wheel Drive as the "efficiency champ" in their real-world range test. It exceeded its EPA range estimate (363 miles) by going 393 miles (+8.3%) and was significantly more efficient than its EPA rating, achieving 21.7 kWh/100 miles (4.61 miles/kWh) compared to 25 kWh/100 miles (4 miles/kWh), making it 13.2% more efficient. * **Tesla Vehicle Longevity:** A study by IC Cars covering 174 million vehicles found that Teslas are twice as likely to reach 250,000 miles as Subarus. Tesla ranked 6th overall with a 4.6% chance (tied with GMC), ahead of luxury brands like Cadillac, Mercedes, BMW, and Porsche. Among luxury brands, Tesla was third behind Lexus and Acura. **SpaceX AI-1 Satellite & Starship Discussion:** * **SpaceX Interviews:** Two new general interviews with Elon Musk and the CFO of SpaceX were released. * **AI-1 Satellite Description:** An image of the AI-1 satellite was shown, highlighting its radiators, compute module, and massive solar arrays. It measures 70 meters across (roughly 75 yards) and 20 meters tall (roughly 20 yards). * **Elon's Ambitious Target:** SpaceX is targeting one gigawatt of orbital AI compute (annualized) by the end of 2027, which Dylan views as overly optimistic. * **AI-1 Power & Weight:** Each AI-1 orbital satellite provides 150 kilowatts (0.15 megawatts) at peak power, or 120 kilowatts (0.12 megawatts) sustained/average. They weigh about 2 tons (4,400 lbs), comparable to a Model Y. * **Comparison to Colossus 1:** To match a 500-megawatt data center (like Colossus 1), it would require 3,333 AI-1 satellites (using peak power) or 4,000 satellites (using sustained power). * **Starship V3 Launch Requirements:** Starship V3 is projected to carry 100 tons (about 50 AI-1 satellites) per trip. To put 3,333 AI-1 satellites (6,666 tons) into orbit for a 500 MW equivalent, it would require 67 Starship launches. To reach 1 gigawatt annualized, this would mean roughly 150 Starship launches (>12 launches per month). * **Starship Launch Cadence Challenges:** Starship has launched 12 times in three years (5 times in 2025, 1 time this year due to V3 upgrades) and has not yet reached orbit. Most launches carried minimal payloads; the max was 45 tons (22 simulated Starlink units). * **Starlink V3 Context:** Only two modified Starlink V2s (simulating V3 hardware) have launched for heat shield testing; zero actual Starlink V3 satellites are operating for service. * **Optimism vs. Reality:** Dylan concludes that the target of 1 gigawatt by end of 2027, requiring such a high Starship launch cadence and new factory build-out, is highly optimistic given current progress. **Market Performance & Sponsor:** * **Tesla Stock:** Tesla stock closed the day at $396.68, down 3%, with 14% above-average volume. The NDX was down 1.5%. * **Sponsor:** Surfshark, highlighting their new Dowsos VPN protocol, which is post-quantum secure, offers dedicated encrypted tunnels, is 30% faster, independently audited, and includes split-tunneling. Dylan hopes to feel better by tomorrow and thanks his Patreon supporters.