Elon's Unexpected Tesla Announcement / New Tesla Semi Design / Unsupervised Finally Launches ⚡️
发布时间 2026-01-23 01:00:18 来源
以下是新闻内容的中文摘要：
**特斯拉自动驾驶出租车和自动驾驶技术：**
* **无人驾驶付费乘车服务已在奥斯汀上线：** 特斯拉已正式在奥斯汀推出无人驾驶、顾客付费的自动驾驶出租车服务，这一消息得到了“Tesla 99T”的视频以及多位特斯拉工程师的证实。
* **混合车队与规模化：** 初期，无人驾驶车辆将与更广泛的、配备安全员的车辆混合运营，预计随着时间推移，无人驾驶车辆的比例将逐渐增加。
* **伴随车辆：** 目前，无人驾驶车辆会有特斯拉伴随车辆进行观察验证，预计还将持续2-3周。
* **有限的运行区域：** 一位乘客（Joe Tettmire）无法更改目的地，这表明这些无人驾驶车辆可能在更有限的地理围栏或特定区域内运行，很可能是特斯拉数据最丰富的奥斯汀南部地区。
* **首位报道者：** 首位报道此消息的人士是一名特斯拉前人工智能工程师。
* **车辆交互：** 方向盘被锁定，无法进行手动输入，但乘客座位可用。如果乘客随意触碰方向盘，将触发警报并最终导致车辆靠边停车。
* **定价一致性：** 自动驾驶出租车（3.31英里行程收费4.31美元）的定价与有安全员时保持一致，这证实了特斯拉的长期策略，即不将安全员的工资纳入成本模型。
* **小费选项：** 应用程序中存在小费选项，但过去有时会出现故障，显示“刺猬通知”。
* **埃隆确认：** 埃隆·马斯克在X上证实了此次发布，祝贺人工智能团队，并邀请更多人加入特斯拉AI，共同解决现实世界中的AI挑战。
* **历史时间线：** 从奥斯汀自动驾驶出租车服务推出到首次无人驾驶付费客户乘车，恰好用了7个月时间，这为未来在其他城市推出服务树立了标杆。
* **纯视觉成就：** 对于纯视觉自动驾驶系统来说，这是一个历史性时刻，其能力与消费级FSD车辆相同。
* **摩根士丹利分析师更新：** 摩根士丹利新任特斯拉股票分析师安德鲁承认了纯视觉方案的关键性时刻，但对自动驾驶出租车的扩张提出了保守预测（年底前1,000辆，2030年30,000辆，2040年500万辆），主持人认为这些预测过于保守。
* **FSD演示扩展：** 特斯拉已将FSD演示驾驶扩展到欧盟各地，现在包括西班牙和芬兰，试驾预约已排到第一季度。
* **FSD横贯大陆挑战（洛杉矶至纽约）：** 一辆搭载硬件4.0和FSD 14.2.2.3的2024款Model S完成了从洛杉矶到纽约市3,081英里、零干预的行程，耗时58小时22分钟（平均时速64英里/小时），其中17%的时间用于充电。FSD系统在极端冬季条件（雪、冰、泥泞）下表现出色，获得了驾驶员的高度赞扬。
* **埃隆的FSD状态：** 埃隆·马斯克表示，自动驾驶是一个“已解决的问题”。
* **全球FSD/自动驾驶出租车推广：** 特斯拉预计其自动驾驶出租车服务将在今年年底前在美国“非常普遍”，并希望下个月在欧洲和中国在类似的时间线获得有监督的批准。
**特斯拉与SpaceX能源和AI计划：**
* **埃隆在达沃斯世界经济论坛的采访：** 埃隆重申，人工智能成本正在迅速下降，能源很快将成为限制因素，导致到年底时生产的芯片数量将超过可供电的芯片数量。
* **大规模太阳能生产：** 特斯拉和SpaceX现在都专注于建设大规模太阳能，目标是在大约3年内在美国每年合计生产100千兆瓦（GW）的太阳能。这比特斯拉历史上的部署量有了显著增加，可能主要用于太空中的AI数据中心。
* **国内太阳能生产：** 埃隆强调，由于高额进口关税，国内太阳能电池生产（不仅仅是组装）可以而且将会实现，这标志着垂直整合的转变，将使纽约布法罗等地受益。
**Optimus人形机器人：**
* **进展：** Optimus目前正在工厂中执行简单任务，预计今年晚些时候将处理更复杂的工业任务。
* **公开发售时间线：** 埃隆预测特斯拉将在明年年底前向公众销售Optimus，他对其可靠性、安全性和功能性充满信心。主持人对此在家用场景下的时间线表示怀疑，因为家用对安全性和可靠性有更严格的要求。
**ARK Invest “2026年大创意”报告：**
* **Optimus难度：** ARK估计Optimus的规模化难度是自动驾驶出租车的20万倍（相比埃隆所说的100倍）。
* **自动驾驶出租车市场份额：** Waymo在旧金山的市场份额停滞在15-20%，Uber仍占据主导地位。
* **自动驾驶里程：** 特斯拉的累计自动驾驶里程接近80亿英里，远超所有其他公司（<10亿英里）。
* **每英里成本优势：** ARK预测，特斯拉的改款Model Y在每英里增量成本上比Waymo的第五代车型（捷豹I-Pace）便宜35%。到2030年，赛博出租车（CyberCab）预计将比Waymo的第六代车型（Olli）便宜50%，目标是每英里0.20美元。埃隆称这一说法“可能属实”，但主持人对这一水平的盈利能力表示怀疑。
**其他特斯拉新闻：**
* **Megapack部署：** Copia Power宣布，亚利桑那州的Maricopa Energy已安装了594个特斯拉Megapack，总计2,200兆瓦时（MWh）的电池储能，并与一个550兆瓦（MW）的太阳能项目配对，预计将于年底投入运营。
* **特斯拉Semi更新版（“Atlas”）：** 特斯拉Semi卡车的新版更新车型，代号“Atlas”，在展示时没有配备整流罩。它似乎略短，车顶高度不同，主要是为了适应各种拖车类型（平板车、油罐车）并便于运输，而非表示长途或短途续航。预计将在第四季度财报电话会议上公布更多更新信息。
* **Cybertruck交付：** Cybertruck已开始在阿联酋交付首批车辆。
**竞争对手新闻：**
* **Waymo进驻迈阿密：** Waymo已开始在迈阿密向公众开放其完全自动驾驶的叫车服务，服务范围为60平方英里，已吸引近1万名注册用户。
**SpaceX IPO与AI卫星：**
* **IPO计划：** SpaceX已与四家银行（美国银行、高盛、摩根大通、摩根士丹利）合作，以主导其潜在的首次公开募股（IPO）。
* **AI卫星规模化：** 埃隆澄清，“每年100千兆瓦”的SpaceX目标是指在3-4年内从地球发射的*太空AI卫星*所需的能量，纠正了之前“太瓦”的说法。他指出，每年数百太瓦的能源将需要月球制造，并且至少还需要10年以上才能实现，并补充说美国平均每年电力消耗仅为0.5太瓦。
**市场表现：**
* **特斯拉股价：** 特斯拉股价当日收盘上涨4.15%，报449.36美元，表现优于纳斯达克100指数（上涨0.76%）。
Here's a summary of the news items from the transcription:
**Tesla Robotaxi & Autonomy:**
* **Unsupervised Paid Rides Live in Austin:** Tesla has officially launched unsupervised, customer-paid robotaxi rides in Austin, confirmed by a video from "Tesla 99T" and various Tesla engineers.
* **Mixed Fleet & Scaling:** Initially, unsupervised vehicles are mixed with a broader fleet using safety monitors, with the ratio expected to increase over time.
* **Chase Vehicles:** Unsupervised vehicles currently have chase Tesla vehicles observing for validation, expected for another 2-3 weeks.
* **Limited Operational Domain:** A rider (Joe Tettmire) was unable to change his destination, suggesting these unsupervised vehicles might operate within a more limited geofence or specific area, likely South Austin where Tesla has the most data.
* **First News Breaker:** The first person to report this news was an ex-Tesla AI engineer.
* **Vehicle Interaction:** The steering wheel is locked for manual input, and the passenger seat is usable. Messing with the wheel triggers an alert and eventual vehicle pullover.
* **Pricing Consistency:** Robotaxi pricing ($4.31 for a 3.31-mile trip) remains the same as with safety monitors, confirming Tesla's long-term strategy to not factor monitor salaries into the model.
* **Tipping Option:** A tipping option exists in the app, but historically can sometimes glitch, showing a "hedgehog notification."
* **Elon's Confirmation:** Elon Musk confirmed the launch on X, congratulating the AI team and inviting others to join Tesla AI for solving real-world AI challenges.
* **Historic Timeline:** It took exactly 7 months from the launch of the Austin Robotaxi service to this first unsupervised paid customer ride, setting a benchmark for future city launches.
* **Camera-Only Achievement:** This is a historic moment for a camera-only autonomous system, with capabilities identical to consumer FSD vehicles.
* **Morgan Stanley Analyst Update:** Andrew, Morgan Stanley's new Tesla stock analyst, acknowledged the pivotal moment of the vision-only approach but provided conservative robotaxi scaling predictions (1,000 by year-end, 30,000 by 2030, 5 million by 2040), which the host believes are too low.
* **FSD Demo Expansion:** Tesla has expanded FSD demo drives across the EU, now including Spain and Finland, with rides booking through Q1.
* **FSD Cannonball Run (LA to NYC):** A 2024 Model S on hardware 4 with FSD 14.2.2.3 completed a 3,081-mile, zero-intervention trip from Los Angeles to New York City in 58 hours and 22 minutes (64 mph average), with 17% of the time spent charging. The FSD system handled extreme winter conditions (snow, ice, slush) with high praise from the driver.
* **Elon's FSD Status:** Elon Musk stated that self-driving is a "solved problem."
* **Global FSD/Robotaxi Rollout:** Tesla expects its robotaxi service to be "very widespread" in the US by the end of this year and hopes for supervised approval in Europe next month and China on a similar timeline.
**Tesla & SpaceX Energy & AI Initiatives:**
* **Elon's WEF Interview:** Elon reiterated that AI cost is declining quickly, and energy will soon be the limiting factor, leading to more chips produced than can be powered by year-end.
* **Large-Scale Solar Production:** Tesla and SpaceX are both now focusing on building large-scale solar, aiming for a combined 100 gigawatts per year in the US within approximately 3 years. This is a significant increase from Tesla's historical deployments and is likely geared towards AI data centers in space.
* **Domestic Solar Production:** Elon emphasized that domestic solar cell production (not just assembly) can and will be done due to high import tariffs, signifying a vertical integration shift, benefiting locations like Buffalo, NY.
**Optimus Humanoid Robot:**
* **Progress:** Optimus is currently performing simple tasks in the factory and is expected to handle more complex industrial tasks later this year.
* **Public Sales Timeline:** Elon predicts Tesla will be selling Optimus to the public by the end of next year, with high confidence in its reliability, safety, and functionality. The host expresses skepticism about this timeline for home settings due to stringent safety and reliability requirements.
**ARK Invest "Big Ideas 2026" Report:**
* **Optimus Difficulty:** ARK estimates Optimus is 200,000 times harder to scale than a robotaxi (compared to Elon's 100 times).
* **Robotaxi Market Share:** Waymo is stuck at 15-20% market share in San Francisco, with Uber remaining dominant.
* **Autonomous Miles:** Tesla is nearing 8 billion cumulative autonomous miles, significantly dwarfing all other companies (<1 billion).
* **Cost Per Mile Advantage:** ARK predicts Tesla's refreshed Model Y is 35% more affordable incrementally per mile than Waymo's Gen 5 (Jaguar I-Pace). By 2030, the CyberCab is projected to be 50% cheaper than Waymo's Gen 6 (Olli), aiming for $0.20/mile, a claim Elon said was "probably true," though the host has doubts about profitability at that level.
**Other Tesla News:**
* **Megapack Deployment:** Copia Power announced that Maricopa Energy in Arizona has installed 594 Tesla Megapacks, comprising 2,200 MWh of battery storage paired with a 550 MW solar project, expected to be operational by year-end.
* **Tesla Semi Refresh ("Atlas"):** A new, updated version of the Tesla Semi, code-named "Atlas," was shown without a fairing. It appears to be slightly shorter, with a different roof height primarily for accommodating various trailer types (flatbeds, tankers) and easier transport, rather than indicating long-haul vs. short-haul range. Updates are expected on the Q4 call.
* **Cybertruck Deliveries:** First Cybertruck deliveries have started in the UAE.
**Competitor News:**
* **Waymo in Miami:** Waymo has begun welcoming public riders into its fully autonomous ride-hailing service in Miami, starting with a 60-square-mile service area and nearly 10,000 sign-ups.
**SpaceX IPO & AI Satellites:**
* **IPO Plans:** SpaceX has lined up four banks (Bank of America, Goldman Sachs, JP Morgan, Morgan Stanley) to lead its potential IPO.
* **AI Satellite Scaling:** Elon clarified that the "100 gigawatts per year" goal for SpaceX is for *space AI satellites* launched from Earth within 3-4 years, correcting previous "terawatt" claims. He noted that hundreds of terawatts per year would require moon manufacturing and is 10+ years away, providing context that the average US electricity consumption is only 0.5 terawatts.
**Market Performance:**
* **Tesla Stock:** Tesla stock closed the day up 4.15% at $449.36, outperforming the NDX (up 0.76%).
摘要
My Tesla referral link: https://ts.la/dillon81150 My fitness channel: https://www.youtube.com/@BuiltxBasicz Bandz Fitness IG: ...
GPT-4正在为你翻译摘要中......
中英文字稿 
Welcome to ElectroFied, it's your host, Dylan Lumis. Apologies if my voice is a little off today, dealing with a little bug, but was not going to stop me from making a video today as we have Tesla 99T sharing the first video of a robot taxi without a safety monitor in Austin. Meaning unsupervised customer paid rides are officially live in Austin. Oh, so long real. And yes, these were paid rides. Here's the proof. He shared the screenshot paying $2.22 for one ride in $3.07 for the other.
欢迎来到ElectroFied频道，我是你的主持人Dylan Lumis。如果我的声音有些不对劲，请见谅，因为我正在处理一些小问题，不过这并没有阻止我今天制作视频，因为我们获得了一个重要的消息。Tesla 99T分享了首个在奥斯汀没有安全监控员的机器人出租车视频。这意味着无人监督的付费客户乘车服务在奥斯汀正式上线。确实让人感到兴奋。而且这些乘车确实是付费的。这里有证明。他分享了截图，显示一程花费了$2.22，另一程花费了$3.07。
And we have many Tesla engineers sharing the news. A show said robot taxi rides without any safety monitors are now publicly available in Austin. Starting with a few unsupervised vehicles mixed in with a broader robot taxi fleet with safety monitors and the ratio will increase over time. Joe Tettmire shared a video of his first unsupervised robot taxi ride. And I do think it's important to point out that at least for now, these unsupervised vehicles do have chase Tesla vehicles.
我们有很多特斯拉工程师在分享这个消息。一档节目中提到，现在在奥斯汀，无人监督的机器人出租车已经可以公开试乘了。一开始会有一些无监督的车辆混在有安全监控员的机器人出租车队伍中，随着时间的推移，无监督车辆的比例将会增加。Joe Tettmire 分享了他第一次乘坐无监督机器人出租车的视频。不过我认为有必要指出，至少目前，这些无监督车辆后面还是有特斯拉跟随车的。
And as Joe's vehicle pulled into the parking lot, the chase vehicle just kind of hung back just to make sure everything's going smoothly to validate the new operation. So maybe another two to three weeks of these chase cars just to take that extra step of safety precaution. You know, so they've got a lot of experience and a lot of data for this area of the town as well. So I think that may be one of the reasons why at least for today, the unsupervised vehicles are kind of concentrated here in this part of the site.
当乔的车辆驶入停车场时，跟踪车辆则保持在后面，以确保一切顺利进行，以验证新的操作。因此，可能还需要额外两到三个星期的跟踪车辆来增添一层安全保障。他们在这个城区有丰富的经验和大量的数据。所以我认为这可能是为什么至少今天无人监管的车辆集中在这个区域的原因之一。
So great to see again just how this is continuing to progress. So what I'm going to do now, I think, is I'm going to try to change my routing and go back to where I started from so I can continue with this robot taxi. So there'll be a break here for a moment and then I'll resume in a second. Okay, well, I tried to re-root back to where I started. I was not able to at least at this stage of the robot taxi. So it's not clear exactly why Joe was not able to adjust his destination in this unsupervised ride, but something to note.
很高兴再次看到事情不断进展。接下来我打算尝试改变路线，返回出发地，以便继续使用这辆机器人出租车。所以这里会暂停一会儿，然后我会马上继续。好的，我尝试重新返回到出发地，但在目前这个机器人出租车的阶段无法实现。不太清楚为什么乔在这次无人驾驶旅程中无法调整目的地，但这点值得注意。
And it may be the case that these few unsupervised vehicles do have a more limited operational domain rather than the entire geofence in Austin. I've not yet seen anything about limitations on choosing destinations, but it's likely that you may only get one of these unsupervised rides if you're in a specific part of the geofence, which right now might be limited to South Austin where Tesla launched the service initially and has the most data.
这些无人监控的车辆可能在奥斯汀的运行区域比整个地理围栏更有限。我尚未看到关于目的地选择限制的具体信息，但很可能只有在地理围栏的特定区域内的人才能使用这样的无人驾驶服务。目前，这个区域可能仅限于奥斯汀南部，因为特斯拉最先在这里推出服务，并拥有最多的数据。
I also think it's worth pointing out that the first person to break this news is indeed an X Tesla AI engineer. Due to that, what you will, it's not like he works there anymore, but I thought it was worth pointing out. But clearly, as we just saw Joe Tetmire Godin unsupervised paid ride and he does not work for Tesla now or in the past. No, I mean the passenger seat. I mean the passenger seat now. It's a different car.
我想指出的一个值得注意的地方是，最先爆出这个消息的人确实是一位前特斯拉AI工程师。虽然他现在已经不在特斯拉工作了，但我认为这一点值得一提。不过显然，正如我们刚才看到的，Joe Tetmire 体验了一次无人监管的付费试乘，而且他现在和以前都没有在特斯拉工作过。我指的是乘客座位，现在这是另一辆车。
This is a ride one. And what is interesting that you can't control the steering wheel. Yeah, it's now protected, though, but you can now control it. Yeah, so I'm pretty sure that this robot taxi can fit at least three people if not four. I'm not sure if it's allowed to see the gear actually. Yeah, it's so poor. Cool. So the steering wheel is locked for manual input and you are able to sit in the passenger seat.
这真是一段有趣的旅程。值得注意的是，你不能控制方向盘。方向盘现在虽已被保护，但你仍然无法控制它。我确信这辆无人驾驶出租车至少可以坐三个人，甚至四个人。我不确定是否可以看到变速杆。真是有点遗憾。不过，很酷的是，方向盘已经锁定，无法进行手动操作，而你可以坐在乘客座位上。
And if you continually mess with the wheel and alert pops up and it appears to say that if you keep doing it, the vehicle will pull over. Joe Tetmire said his first trip was $4.31 for a 3.31 mile trip, which is exactly the same formula as before. Meaning at least with this data point, the robot taxi pricing model is no different with the safety monitor removed, which if that holds up would confirm what we've been saying here for months, that Tesla is likely to build this pricing algorithm without the safety monitor.
如果你不断摆弄方向盘，会弹出一个警告，看起来像是在说如果你继续这样做，车辆会靠边停下。乔·特米尔（Joe Tetmire）说，他的第一次行程花费了4.31美元，行驶了3.31英里，这与之前的计算公式完全相同。这意味着，至少从这个数据点来看，即使没有安全监控员，机器人出租车的定价模型也没有变化。如果这一点成立，就会证实我们几个月来一直在说的，特斯拉可能在没有安全监控员的情况下建立这个定价算法。
And they would just eat the salaries they're paying to the safety monitor for as long as they need to, but they're not going to build that into the model because long term, the plan is to not have monitors. And that's also why I said I was not expecting Tesla to offer lower rates when the supervisor was removed, one because they're already undercutting the competition significantly so they just don't need to and two because I was already expecting the pricing algorithm had the safety monitors removed.
他们会继续支付安全监督员的工资，只要需要就会这样做，但不会把这部分纳入长期计划中，因为长期计划是取消这些监督员。这也是为什么我说当去掉监督员时，我不指望特斯拉会降低价格的原因之一。首先，因为他们的价格已经比竞争对手低很多，所以没有必要降低，第二，我原本就认为他们的定价算法中已经不包含安全监督员的费用了。
Joe did share this screenshot. You have the option to leave a tip $0.69 for $1.20 or other, but as we've seen in the past, sometimes if you try to leave a tip, you'll just get the hedgehog notification and you're not actually able to. So it still might be Tesla just poking fun at other companies that are looking for tips for everything. And Elon shared it saying just started Tesla Robotaxi drives in Austin with no safety monitor in the car, congrats to the AI team.
Joe确实分享了这个截图。你可以选择给 $1.20 留下 $0.69 或其他金额的小费，但正如我们过去所见，有时如果你尝试留下小费，你只会收到刺猬通知，实际上无法成功。所以，这可能仍然是特斯拉在拿那些对每件事都寻求小费的其他公司开玩笑。Elon 分享时还提到，相继在奥斯汀启动了没有安全监控员的特斯拉自动驾驶出租车，祝贺 AI 团队。
If you're interested in solving real world AI, which is likely to lead to AGI, IMO, join Tesla AI, solving real world AI for Optimus will be 100 times harder than cars. And I do think that's a line we should be repeating to ourselves as we set our expectations for the next few years. Remember how Morgan Stanley got a new Tesla stock analyst since Adam Jonas moved into a new AI centric role? Well, I think it shows with their update today. Andrew started off strong saying Tesla's removal of the safety driver in Austin marks a pivotal moment in proving out its vision only approach to autonomy. The scaling of a passive optical only AV system would seriously challenge the conventional thinking of many in the Robotaxi community.
如果你对解决实际世界中的人工智能问题感兴趣，而这很可能会导致通用人工智能（AGI）的诞生，在我看来，加入特斯拉的AI团队吧。解决用于Optimus的实际世界AI问题将比解决汽车难一百倍。而且，我确实认为我们应该在未来的几年中不断提醒自己这一点。还记得摩根士丹利因为Adam Jonas转入了以AI为中心的新职位后，找了一个新的特斯拉股票分析师吗？好吧，我认为这在他们今天的更新中体现出来了。Andrew开场很有力地说，特斯拉在奥斯丁移除安全驾驶员是证明其纯视觉方法实现自动驾驶的一个关键时刻。仅依靠光学的被动系统的扩展将对许多在自动驾驶出租车领域的传统思维方式形成严峻挑战。
And then it falls apart. We expect Tesla to make significant progress this year with a thousand Robotaxis on road by year end scaling to 30,000 by the end of the decade and 5 million by 2040. Now, yes, I'm on record saying I was overzealous with my Robotaxi estimates when the service launched. But if Tesla doesn't have significantly more than a thousand Robotaxis on the road by the end of the year, I will gladly come out and say something is not going right. And I think the same thing could be said about his prediction for 30,000 by the end of the decade.
然后情况就不妙了。我们预计特斯拉今年会有显著进展，计划在年末前让一千辆自动驾驶出租车上路，并在本世纪末扩展到三万辆，到2040年达到五百万辆。是的，我曾公开表示在服务推出时对自动驾驶出租车的预估过于乐观。但是如果特斯拉到年底时还没有远远超过一千辆的自动驾驶出租车上路，我会坦言某些事情没有按预期发展。我认为，对于他预测本世纪末达到三万辆的情况，也可以这么说。
I'd also be comfortable saying if Tesla can't average 8,000 Robotaxis each year for the next four years, then something is wrong. And for those counting, it was exactly seven months from the launch of the Robotaxi service in Austin to this first unsupervised paid customer ride. So for me, that will be the timeline to beat for each new launch in a new city. So this really is historic. Wemo may have beat Tesla to the punch, but Tesla pulled it off with cameras only. And outside of some GPS hardware to communicate with the remote operators, as we know, anyone can go buy this vehicle and own it themselves.
如果特斯拉在接下来的四年里每年不能平均推出8000台自动驾驶出租车，那就说明出现了问题。对于那些在意时间的人，从自动驾驶出租车服务在奥斯汀启动到首次无人监督的付费用户乘坐，历时正好七个月。所以，对于我而言，这将是每个新城市推出服务时需要超越的时间记录。这真的是历史性的。虽然有些公司可能比特斯拉更早推出类似服务，但特斯拉只依靠摄像头就实现了这一点。除了用来与远程操作员沟通的GPS硬件，我们知道，任何人都可以买到这款车并拥有它。
The capabilities are exactly the same, the camera suite is the same, it's just a matter of regulations and Tesla unlocking unsupervised. So of course, we'll watch how quickly Tesla adds unsupervised Robotaxis to the mix and how soon the Bay Area follows suit. But today, I hope everyone can sit back and take it in and celebrate this moment. Many of us have waited years for this. And I really can't even imagine what it's like for the AI team to watch their work manifest in the real world like this. And to everybody who said Tesla would never pull it off, I'd simply say don't doubt Elon.
能力完全相同，相机系统也一样，区别只是监管法规和特斯拉开放无人监督功能的问题。所以，我们当然会关注特斯拉多快会增加无人监督的自动驾驶出租车，以及湾区多快会跟进。但是今天，我希望大家能放松下来，享受并庆祝这个时刻。我们中的许多人已经等待了多年。我甚至无法想象，对于AI团队来说，看到他们的工作以这种方式在现实世界中实现，是怎样的感受。对于那些曾说特斯拉做不到的人，我只想说，不要低估埃隆·马斯克。
As we've seen time and time again, he makes the impossible merely late. And if you thought the clip we shared of Devon the other day was a fluke, well here's one more to say that it's not. Elon gave an interview today at the World Economic Forum, which I found to be interesting. But not just because of what Elon said, but what he said about the WEF in the past. He said it's increasingly becoming an unelected world government that the people never asked for and don't want. In the past, he said I was invited to WEF, but declined.
正如我们一再见到的，他总是能将不可能的事情推迟一点时间去完成。而如果你以为我们前几天分享的Devon的视频只是个巧合，那么这里还有一个例子来证明并不是。今天，埃隆在世界经济论坛上接受了一次采访，我觉得很有趣。这不仅因为埃隆说的话，还因为他过去对世界经济论坛的评价。他说，这个论坛正越来越像一个未经选举产生的世界政府，人民从未要求过，也不想要。他还提到，过去他曾被邀请参加世界经济论坛，但他拒绝了。
He said by misinformation, WEF means anything that conflicts with its agenda. He said it would be great if someone could compile a game contest of who said the craziest stuff between 4chan and WEF my money is on the latter. And one more, there are some really funny WEF parody accounts, although it's legit hard to tell WEF truth from WEF parody sometimes. There was not a ton of new information here, it was a lot of repeated themes like the cost of AI continually declining quickly, nearly monthly at this point. The limiting factor really is energy.
他说，世界经济论坛（WEF）所指的"错误信息"就是任何与其议程相悖的内容。他还提到，如果有人能够举办一场比赛，比较4chan和WEF谁说的离谱话更多，那会很有趣，而他认为WEF更甚。此外，还有一些非常搞笑的WEF恶搞账号，尽管有时候很难分辨WEF的真实信息和恶搞内容。这里并没有太多新信息，大部分是反复提到的一些主题，比如人工智能成本正在快速下降，几乎达到每月都有变化的地步。真正的限制因素实际上是能源。
Elon said it'll likely be the case that we'll be producing more chips than we can actually turn on thanks to those electricity bottlenecks by later this year. One new thing we did learn is that Tesla and SpaceX are both now focused on building large scale solar. Elon said they're both working separately to build 100 gigawatts per year of solar in the US, and the way Elon phrased it, he did make it sound like that would be 100 gigawatts for both Tesla and SpaceX. Elon seems to think it'll take around 3 years for the companies to get there.
Elon 说，到今年年底，由于电力瓶颈问题，我们可能会生产出超过我们实际能够使用的芯片。我们了解到的新情况是，特斯拉和SpaceX现在都专注于建设大规模太阳能项目。Elon 说，他们两家公司分别在美国推进每年100吉瓦的太阳能建设，听起来这个目标是特斯拉和SpaceX各自要实现的。Elon 认为，这需要大约三年的时间才能实现。
And just to give you a sense of scale here, Tesla in its lifetime has likely deployed less than 10 gigawatts of solar. The last time Tesla broke out, its solar deployments on a quarterly call was back in 2023, and at the time they were doing like 50 megawatts per quarter. And it's actually likely that Tesla has deployed less than 5 gigawatts of solar in its lifetime. So even if that 100 gigawatts number Elon talked about was combined for Tesla and SpaceX, that would still be 50 gigawatts for Tesla, which is likely to be an order of magnitude more than Tesla has deployed in its lifetime.
为了让你有个规模感，这里提到的特斯拉：在其整个发展历程中，特斯拉可能部署的太阳能不到10吉瓦。特斯拉上一次在季度电话会议中详细披露其太阳能部署情况是在2023年，当时他们每季度的部署量大约是50兆瓦。实际上，特斯拉在其整个发展历程中可能部署的太阳能甚至不到5吉瓦。因此，即便埃隆提到的100吉瓦是特斯拉和SpaceX的合计数字，那其中50吉瓦归于特斯拉的可能性，也是特斯拉截至目前部署量的十倍。
And now, Elon's talking about Tesla building that much every year. So maybe Tesla has seen an uptick in demand with a new lease offering with a new Tesla solar panel, but I think the safer assumption would be that both of these companies are going to be scaling up for these AI data centers in space. That's because also in this interview, Elon said that the lowest cost place to deploy AI in about 2-3 years will be space. And given Elon is talking about a 3-year timeline to build up to that 100 gigawatt number of solar, this announcement is likely Tesla and SpaceX gearing up for that endeavor. And I'm sure Buffalo will love to hear this announcement as that's where Tesla is making and assembling its solar panels.
现在，埃隆谈到特斯拉每年建造这么多。因此，也许由于新的租赁方案和新的特斯拉太阳能电池板，特斯拉的需求有所上升。但我认为，更安全的假设是，这两家公司都在为太空中的这些人工智能数据中心扩展规模。这是因为在这次采访中，埃隆表示，大约在2-3年后，太空将是部署人工智能成本最低的地方。考虑到埃隆谈论的是在3年内达到100吉瓦太阳能的规模，这个公告可能意味着特斯拉和SpaceX正在为此做准备。而且我相信Buffalo会很高兴听到这个消息，因为特斯拉正在那里生产和组装太阳能电池板。
And Elon did say the importation of solar cells is very expensive due to tariffs, but production domestically can and will be done. And this is actually a very important point because really to date, Tesla has been importing the cells from other places around the world and then just assembling them into solar panels in New York. It's always been very unclear to what level and to what degree Tesla solar is probably the most opaque division that it has, but very clearly today, Elon is talking about specifically producing the cells domestically.
Elon提到，由于关税原因，进口太阳能电池价格很高，但在国内生产既可以做到，也将会做到。这其实是个非常重要的点，因为到目前为止，特斯拉一直在从世界其他地方进口电池，然后在纽约将其组装成太阳能板。一直以来，特斯拉的太阳能业务是公司中最不透明的部分之一，但今天Elon明确表示，这是关于在国内生产这些电池的具体计划。
So this is likely a shift not just in scale, but also when it comes to the level of vertical integration as well. Elon said humanoid will advance quickly, Optimus is doing simple tasks in the factory now, and probably later this year will be doing more complex tasks. But they will still at that point be deployed in an industrial environment. And then he said probably by the end of next year, Tesla will be selling Optimus to the public. Elon said he's confident that Optimus will have high reliability and safety and range of functionality at that point, meaning you would be able to ask it to do anything you'd like.
所以，这不仅可能是规模上的转变，也涉及到垂直整合水平的变化。埃隆提到，类人机器人将快速发展，现在Optimus已经在工厂里执行简单任务，可能在今年晚些时候会开始执行更复杂的任务。但即便如此，它们仍将被部署于工业环境中。他还表示，可能到明年底，特斯拉将会向公众销售Optimus。埃隆相信，届时Optimus会有很高的可靠性和安全性，并具备广泛的功能，也就是说，你将能够让它做任何你希望的事情。
I will continue to say I have my doubts about that timeline. I won't be surprised if we don't see Optimus in a home setting until closer to 2030 because at that point with children and pets and security risks and fire risks, the reliability is going to have to be near perfect. And maybe more importantly, Tesla will have zero problem selling Optimus in an industrial setting really for the next decade plus. So Tesla should not be in any rush to sell Optimus to the home, although we all want it. It is going to be a safer path to keep it in an industrial setting for as long as it takes to work out the kinks.
我对这个时间表一直持怀疑态度。我不会感到惊讶，如果我们直到接近2030年才会在家庭环境中看到Optimus。因为在那时，考虑到孩子、宠物、安全风险和火灾风险，Optimus的可靠性必须达到几近完美。此外，也许更重要的是，未来十多年中，特斯拉在工业环境中销售Optimus将毫无问题。因此，特斯拉不必急于将Optimus推向家庭市场，尽管我们都希望如此。在工业环境中继续完善它将是一个更安全的选择，直到解决所有问题。
And ultimately to develop all of these generalized capabilities. And Elon did say self-driving is a solved problem at this point. We have heard that before. And he said Tesla's robot taxi service should be very widespread by the end of this year in the US. Tesla is hoping to get supervised approval in Europe next month and China on a similar timeline. I'll have the full interview linked below. ARK Invest put out its big ideas report for 2026.
为了最终开发出所有这些通用能力，埃隆·马斯克说，现在自动驾驶已经不是问题了。我们以前也听过类似的话。他还提到，特斯拉的机器人出租车服务应该在今年年底前在美国得到广泛应用。特斯拉希望下个月能够在欧洲获得监管批准，而中国的时间表也差不多。完整的采访链接在下面。ARK Invest发布了其面向2026年的重要理念报告。
But parts of this report that pertain to the channel we've already covered in months past. Like the one where ARK is predicting that Optimus will be 200,000 times harder to scale than a robot taxi. Many of you will remember this chart, but it really is somewhat in line with what Elon just said that Optimus is going to be 100 times harder than robot taxi. On the robot taxi front we already highlighted in Pierre's report how Waymo is really stuck at about 15 to 20% market share in San Francisco. It has certainly taken some market share from Lyft, but Uber has remained relatively steady with the majority of the market.
这份报告中的一些部分我们之前已经讨论过数月。比如，报告中指出ARK预测Optimus的扩展难度是机器人出租车的20万倍。很多人可能还记得这个图表，这实际上与Elon最近说的Optimus比机器人出租车难100倍相符。在机器人出租车方面，我们在Pierre的报告中已经提到，Waymo在旧金山的市场份额大约维持在15%到20%左右。虽然它确实从Lyft那里抢了一些市场份额，但Uber依然稳固地占据市场的大部分份额。
And the discrepancy between these two charts on the right is perfect. Most Tesla skeptics will focus on the chart to the left, saying that Waymo, Baidu, Pony AI, and We Ride have all done more actual robot taxi miles than Tesla. But the Tesla Bulls will look at the chart on the right, which is cumulative autonomous miles, where Tesla is dwarfing every other company. Tesla will be eclipsing 8 billion autonomous miles in the next few months, and no other company is even sniffing 1 billion. And hopefully this is the year where this chart on the right starts to filter over to this chart on the left.
这两张右侧图表之间的差异非常明显。很多对特斯拉持怀疑态度的人会集中注意左边的图表，指出Waymo、百度、Pony AI和文远知行的实际机器人出租车行驶里程都比特斯拉多。然而，特斯拉的支持者则会关注右边的累积自动驾驶里程图表，特斯拉在这方面遥遥领先于其他公司。特斯拉在接下来的几个月中将超越80亿英里的自动驾驶里程，而没有其他公司能接近10亿英里。希望今年右边图表上的数据能够反映到左边的图表上。
On this chart, Ark shows how Tesla's refreshed Model Y is about 35% more affordable when it comes to incremental cost per mile, and the bars are broken down by each different part of the business. And specifically that's 35% cheaper than Waymo's fifth generation vehicle, which is the Jaguar I-Pace, not the new OHI. And then on the right we have at scale by 2030, Waymo's Gen 6, which is the OHI compared to the CyberCab, where Ark is expecting the CyberCab to be 50% cheaper than the Waymo.
在这张图表中，Ark展示了特斯拉新款的Model Y在每英里增量成本方面如何更加实惠，约便宜35%。图表中的柱状图显示了业务中每个不同部分的成本。具体来说，这比Waymo的第五代车辆（捷豹I-Pace，不是新的OHI）便宜35%。图表右侧则预测到2030年大规模生产时，Waymo的第六代车辆OHI与CyberCab相比，Ark预计CyberCab会比Waymo便宜50%。
And they're expecting Tesla to be in the neighborhood of 20 cents per mile. And Elon responded to that data saying, probably true. Personally, I have some doubts that Tesla could maintain profitability at a cost per mile that low. So if in the future you want me to do a section on why I have those doubts, I'm happy to do it, but usually I hesitate because there's so many other people in the Tesla space already doing in depth modeling. I really tried to avoid being redundant, but if you'd want to see something like that, just let me know. I will have the full report linked below.
他们预计特斯拉的成本大约是每英里20美分。Elon对这个数据的回应是："大概是真的"。不过，我个人对特斯拉是否能在如此低的每英里成本下保持盈利持有一些怀疑。如果未来你们希望我专门讨论一下我为什么会有这些怀疑，我很乐意这么做，但我通常会犹豫，因为在特斯拉领域已经有很多人在做深入的模型分析。我尽量避免重复，但如果你们想了解这方面的内容，请告诉我。我会在下面链接完整的报告。
On LinkedIn, we have Copia Power saying that Mericopa Energy in Arizona has now installed 594 Tesla Megapax, representing one of the biggest Megapax deployments to date. This is a 550 megawatt solar project paired with 2,200 megawatt hours of battery storage. The project is still under construction, but the word is it should be operational toward the end of this year.
在 LinkedIn 上，Copia Power 表示，位于亚利桑那州的 Mericopa Energy 现已安装了 594 个特斯拉 Megapack，这代表了迄今为止最大的 Megapack 部署之一。这是一个 550 兆瓦的太阳能项目，配备了 2200 兆瓦时的电池储能系统。该项目仍在建设中，但消息称，它预计将在今年年底投入运营。
Tesla has now expanded the FSD demo drives across the EU now to Spain as well as Finland. Both are currently booking rides through the end of Q1. Heinrich Zane shared a new video of what we believe to be the latest updated edition of the Tesla semi. The code name for the new semi is Atlas, kind of like the new model Y was called Juniper. So this is only one of two atlas semis and I think that the atlas is the perfect name because the atlas can carry the freight of the world on its shoulders.
特斯拉现在已经在整个欧盟扩展了其全自动驾驶（FSD）演示驾驶项目，最新增加的国家包括西班牙和芬兰。目前，这两个国家的体验预约都已开放至第一季度末。海因里希·泽恩发布了一段新视频，我们相信这展示了特斯拉半挂卡车的最新版本。这个新半挂卡车的代号是Atlas，就像新的Model Y被称为Juniper一样。因此，目前只有两辆名为Atlas的半挂卡车。我认为Atlas这个名字非常合适，因为在神话中，Atlas能够肩负起整个世界的货物。
So this has no fairing. This is what the Tesla semi looks like with no fairing. Another thing I'd like to talk about is complete speculation, but this atlas refreshed semi appears to be slightly shorter and maybe even less tall than the original semi. And again, it's hard, but there's the regular, there's the old version of the Tesla semi. Now we have to go back into the archives to figure out what's going on with the flat top.
这辆车没有整流罩。这就是没有整流罩的特斯拉半挂卡车的样子。我还想谈谈一个纯属猜测的事情，那就是这个 atlas 版的改进型半挂卡车看起来比原版略短，甚至可能不那么高。不过，请注意，这只是观察推测。我们需要回到存档资料中，弄清楚这个平顶设计是怎么回事。
Assi Robinson replied to that saying there are two types, long haul and short haul, and then Dan priestly chimed in saying the roof height variance is independent range variance. The one in the video is a long range semi just with a shorter roof meant for lower profile trailers like flat beds, drop decks, tankers, etc. Better arrow for those setups. Also easier to transport truck to and from test sites. So what that means is the difference in roof height does not dictate whether it's long haul or short haul. That's what Dan meant by saying it's independent of range variance.
Assi Robinson 回应说，有两种类型的卡车：长途和短途。然后，Dan Priestly 补充说，车顶高度的变化与续航里程的变化是独立的。视频中的卡车是一种长途卡车，只是车顶较低，适合较低轮廓的拖车，例如平板车、降载车、油罐车等。这样的设计对这些类型的车更具空气动力学优势，也更方便卡车在测试场地之间运输。因此，车顶高度的不同并不决定卡车是长途还是短途的。这就是Dan所说的它与续航里程无关的意思。
So the real difference for the shorter roof is for different use cases and it's easier to transport. And as you saw in Zangler's video, it looks like that truck with the shorter roof is being prepared for transport. And this is the year the Tesla semi enters production, so hopefully we get an update on all of this on the Q4 call next week. And I know these angles can play all kinds of tricks on our brains when it comes to sizing, but it does look like that one with the shorter roof is indeed a bit smaller and shorter.
因此，更短的车顶在不同的使用场景下有其真正的差异，它更容易运输。正如你在Zangler的视频中看到的那样，那辆车顶较短的卡车似乎正在准备运输。今年是特斯拉半挂卡车投产的年份，所以希望我们能在下周的第四季度电话会议上得到所有这些的更新。我知道这些不同的角度会让我们在判断尺寸时产生各种错觉，但看起来那个车顶较短的卡车确实更小更矮一些。
So it is true that there will be different versions of the semi, a long haul and a short haul, but the difference in roof type is not going to be the giveaway for which is which. Tesla Europe in Middle East said the first Cybertruck deliveries have finally started in the UAE and it really does just look and feel like the Cybertruck belongs in the UAE.
确实，Tesla Semi 会有不同的版本，包括长途和短途的车型，但车顶类型的不同并不能作为区分它们的标志。特斯拉欧洲中东分部表示，首批 Cybertruck 已经在阿联酋开始交付，而 Cybertruck 的外观和感觉确实让人觉得它很适合阿联酋。
And today the drive just reported that a small team in a 2024 Model S just completed a trip from Los Angeles to New York City and what they're calling the first zero intervention FSD Cannonball run. After multiple detours resulting from both human error and winter weather, the Model S completed the 3,081 mile trip entirely without human intervention. It took 58 hours and 22 minutes at an average speed of 64 miles per hour.
今天有消息称，一个小团队驾驶2024款Model S完成了一次从洛杉矶到纽约市的旅程，他们称之为首次"零干预"全自动驾驶"炮弹飞车"挑战赛。尽管途中由于人为失误和冬季天气原因多次绕道，这辆Model S在无人干预的情况下顺利完成了总长3,081英里的旅程。全程耗时58小时22分钟，平均速度为每小时64英里。
The car was on hardware 4 with FSD 14.2.2.3 and the team spent 10 hours and 11 minutes of the trip waiting for the car to charge. So that's about 17% of the time for the trip charging. And we know David Moss did something similar going from LA to Myrtle Beach, but that's not the trip that Elon talked about. He did say LA to New York. Remember back in 2017 as we shared when David Moss pulled off his feet.
这辆车配备了硬件4，并运行FSD版本14.2.2.3。在旅途中，团队花了10小时11分钟等待汽车充电，这相当于旅程时间的17%。我们知道，大卫·莫斯曾经从洛杉矶开到默特尔海滩，做了类似的事情，但这不是埃隆提到的旅行。他确实提到过从洛杉矶到纽约的行程。请记得，我们分享过2017年大卫·莫斯完成这项壮举的消息。
And just like David, Alex and his team were fully hands off for the entirety of the trip including charging stops. The article did say there was only one disengagement the entire trip and Alex said it was his fault he accidentally touched the wheel but they said in the article that Alex is no Tesla loyalist. And Alex said craziest events in the snow but FSD did it. Holy smokes Roy said in one update snow performance and recovery is unreal. The video will be crazy.
就像David一样，Alex和他的团队在整个旅程中都是全程放手的，包括充电站的停车。他们在文章中提到，在整个旅程中只发生了一次中断，Alex说那是因为他不小心碰到了方向盘。不过，文章中也提到Alex并不是特斯拉的忠实粉丝。Alex说，在雪地里发生了一些最疯狂的事情，但全自动驾驶功能（FSD）完成了任务。Roy在其中一次更新中表示，雪地的表现和恢复能力简直让人难以置信。这个视频将会非常精彩。
And Alex said on X in the middle of winter through extreme cold snow ice slush and rain FSD drove 100% of the 3,081 miles of our journey. Elon was right once an autonomous vehicle is mature most human input is an error. A comedy of human errors added hours and hundreds of miles but FSD stunned us with its consistent and comfortable behavior. So these are certainly big achievements now but soon enough these will just be par for the course.
亚历克斯在X上说，在寒冬中，面对极端的寒冷、雪、冰、融雪和雨，FSD在我们3,081英里的旅程中完成了100%的驾驶。埃隆是对的，一旦自动驾驶汽车成熟，大多数人为操作都会成为错误。人类一连串的错误增加了行程时间和数百英里的距离，但FSD以其稳定和舒适的表现让我们惊讶。虽然这些现在无疑是重大成就，但很快这些也将成为一种常态。
Waymo said starting today it's welcoming the first public riders into their fully autonomous ride healing service in Miami. With nearly 10,000 residents already signed up we will be inviting new riders on a rolling basis. They're starting with a 60 square mile service area.
Waymo表示，从今天开始，他们将在迈阿密迎接首批使用其完全自动驾驶出租车服务的公众乘客。目前已有近10,000名居民注册，我们将会陆续邀请新乘客加入。服务区域最初覆盖60平方英里。
It's looking more and more like SpaceX will IPO via the traditional route. Today we learned SpaceX has lined up four banks to lead its IPO. Those being bank of America, Goldman Sachs, JP Morgan and Morgan Stanley.
看起来 SpaceX 很可能会通过传统方式进行首次公开募股（IPO）。今天我们了解到，SpaceX 已经确定了四家银行来主导这次 IPO，它们分别是美国银行、高盛、摩根大通和摩根士丹利。
Nick said Elon has essentially confirmed SpaceX is going public this year and will use the funds to build solar powered AI satellites scaling to hundreds of terawatts in three years. And hopefully even without Elon's response here the alarm bells would be going off that is not at all what Elon said hundreds of terawatts is obviously a long long term play.
Nick表示，埃隆基本上确认了SpaceX将在今年上市，并利用筹集的资金建造太阳能供电的AI卫星，计划在三年内达到数百太瓦的规模。而且，即使没有埃隆在这里的回应，希望大家也能意识到这并非埃隆真正的意思，因为数百太瓦显然是一个非常长期的目标。
It was the 100 gigawatts of solar annually for Tesla in SpaceX that he's expecting in three years. But luckily Elon corrected it saying no the probable case meaning a 50th percental guess is reaching an annualized rate of 100 gigawatts per year of space AI satellites launched from Earth in three to four years.
他预计特斯拉在SpaceX的太阳能业务将在三年内每年达到100吉瓦。但幸运的是，伊隆纠正道，这只是可能的情况，意味着有50%的概率预计在三到四年内，每年100吉瓦的太空AI卫星将从地球发射。
100 terawatts per year requires manufacturing satellites on the moon at massive scale that are shot into deep space with a mass driver which is 10 plus years away. It's worth noting the average US electricity consumption is only 0.5 terawatts so that would mean launching 200 times the current electricity output of Earth.
每年生产100太瓦特的电力，需要在月球上大规模制造卫星，然后用质量加速器将其发射到深空。这种技术预计还需要大约10年以上的时间才能实现。值得注意的是，美国的平均电力消耗仅为0.5太瓦特，这意味着要将当前地球电力产量的200倍送上太空。
And I have officially used up all of my energy and vocal strength for the day. Tesla stock closed the day at $449.36 up 4.15% while the NDX was up 0.76%. The volume came in effectively right on the average.
我今天已经彻底用光了所有的精力和声音。特斯拉的股票收盘时价格为449.36美元，上涨了4.15%，而纳斯达克100指数上涨了0.76%。成交量基本与平均水平相符。
So again hopefully you all can celebrate this incredible milestone for Tesla and another successful coast to coast FSD intervention free drive. Hope you have a wonderful day and a huge thank you to all of my Patreon supporters.
希望大家能够庆祝这个特斯拉的重大里程碑，以及另一次成功的从东海岸到西海岸的全自动驾驶无干预行驶。祝愿你们度过美好的一天，同时非常感谢我所有的Patreon支持者。