Starlink Mobile Really Just Happened & Elon Can't Contain His Excitement ⚡️

以下是播客内容的摘要翻译： 主持人迪伦·卢米斯（Dylan Loomis）首先赞扬了埃隆·马斯克（Elon Musk），称他为“工业巨头”、“杰出工程师”和“最伟大的科技创始人之一”。他表示，很少有美国人比埃隆为推进美国国家利益做得更多，埃隆的天才将科幻小说变成了现实。埃隆创立了特斯拉，这是75多年来首家成功的新汽车制造商，生产了首款商业上可行的电动汽车，并首次推出了消费者自动驾驶技术。迪伦将埃隆比作“现代爱迪生”，并祝贺他今天被授予国家科学奖章。他认为埃隆在YouTube和X上的“特斯拉泡泡”之外没有得到足够的认可，并相信埃隆的成就“仍处于早期阶段”，他补充说，他喜欢看到埃隆微笑和大笑。 埃隆澄清了TerraFab项目，明确指出SpaceX和特斯拉将建造并运营该工厂，“毫无疑问”。他建议台积电如果愿意，可以转租TerraFab的一部分，但仅此而已，并且只有在TerraFab获得回报的情况下。TerraFab的主要目标是为SpaceX和特斯拉生产芯片，这意味着台积电转租的产出也将用于这些公司。迪伦对埃隆取消台积电拥有和运营TerraFab任何部分的选项感到鼓舞，他将任何潜在的交易比作英特尔作为工艺合作伙伴经理而非所有者-运营商的角色。英特尔提供技术，就设置提供建议，并为其新的14A工艺获得一个大客户。与英特尔不同，台积电已经售罄其所有前沿芯片，这使得埃隆的转租提议成为台积电在美国本土生产新晶圆的更快方式，而无需走“从头开始建设”的路线。台积电将在格莱姆斯县获得电力和水，以换取租金，关键问题是台积电有多少产量将保留给特斯拉和SpaceX。迪伦倾向于SpaceX和特斯拉在走TerraFab路线时尽可能多地保留控制权，他认为战略合作伙伴有利于加速，但警惕进一步的纠缠导致复杂化。 主持人随后宣传了赞助商Bolt.new，他描述了他为自己的梦幻橄榄球联赛构建的一个应用程序。这个应用程序跟踪12场过关赌注的实时下注情况，显示哪些正在进行中，并立即报告输赢，以及历史和当前年度排行榜。他使用Bolt.new构建了它，称这是一种无需大量技术工作即可从口头提示创建应用程序的简单方法，尤其是因为“并非所有这些前沿模型都擅长前端设计之类的东西”。他强调了Bolt Forge，它提供对开放权重模型的访问，提供高达50倍的使用量。他演示了如何选择像Kimmy K3这样的模型，并通过口头提示创建“带有滑动条的RoboTaxi与所有权盈亏平衡模型”，并指出构建有助于匿名训练开放权重模型。他鼓励观众尝试Bolt.new，提供使用代码ELECTRIFIED的30天免费专业账户。 转到Starlink新闻，美国航空宣布将把Starlink扩展到1000多架干线飞机上，安装工作将于2027年初开始。这建立在此前5月份宣布在500多架空客飞机上安装Starlink的基础上，现在覆盖了其1000多架空客和波音窄体及宽体飞机的整个干线机队。迪伦指出，这对于美国航空来说是“完美的时机”，因为达美航空“行动迟缓”。 SpaceX获得了FCC的批准，可以发射和运营15000颗下一代卫星用于Starlink Mobile，该公司去年9月提交了申请。FCC还发布了一项豁免，允许SpaceX通过卫星提供无线服务，而无需与地面移动运营商签订租赁协议，为Starlink Mobile与AT&T、T-Mobile和Verizon竞争创造了监管途径。Starlink Mobile当前的第一版（T-Satellite，与T-Mobile合作）使用大约650颗卫星，估计速度约为4 Mbps，而即将到来的15000颗卫星星座承诺为每位用户提供150 Mbps的5G速度。SpaceX设想同时使用卫星和更小、更具成本效益的地面微蜂窝（femtocells）提供蜂窝覆盖，包括在郊区和城市地区。FCC表示，这个星座将服务于蜂窝通信盲区并增加竞争。该批准允许卫星从335公里下降到326公里（202英里）的轨道运行，略低于第一代（约360公里），并使用从Echo Star收购的更多无线电频谱。SpaceX获得了在美国境外进行蜂窝业务的许可，尽管使用Echo Star频率的测试取决于交易的第二步完成（明年11月）。埃隆评论说，“Starlink Mobile直接到蜂窝的V2卫星令人难以置信”，它能够提供当前V1系统“超过100倍的带宽”。 SpaceX进一步宣布达成协议，收购全国范围内的低频段频谱许可证组合，旨在成为美国主要的移动运营商。埃隆称之为“非常大的交易”。一篇新的博客文章指出，这种“优质低频段频谱”解决了“一个关键的剩余技术空白”，使其能够将卫星到移动星座与先进的地面部署相结合，形成灵活的架构。这种新的低频段频谱将提供一个覆盖层，穿透墙壁等障碍物并在建筑物内提供服务，补充Starlink Mobile用于高带宽的全球2Ghz中频段。大多数现有的移动设备已经支持这个未充分利用的频段。SpaceX的目标是成为第一个将卫星和地面频谱作为单一网络部署的运营商。迪伦强调了T-Mobile首席财务官的公开驳斥（称物理学不允许直接到蜂窝具有竞争力）和AT&T首席执行官约翰·斯坦基（John Stanky）提出的关于未经许可从家中辐射蜂窝信号的问题，称SpaceX随后的举动是“诗意的正义”。亚伦·伯内特（Aaron Burnett，埃隆分享）指出，2027年和2028年的手机将配备芯片组，以原生支持Starlink Mobile的高吞吐量频谱。FCC还授权Starlink V3宽带卫星的回程频段用于现有网关网络。这种低频段频谱被视为室内覆盖的“缺失拼图”。尽管FCC仍需批准800 MHz的使用，但计划是卫星覆盖室外/农村/盲区，而带有800 MHz的地面塔台覆盖室内/密集区域，手机可在两者之间无缝切换。800 MHz许可证之前由Grain Management持有（今年早些时候从T-Mobile获得）。据报道，这笔交易约为60亿美元，仍需FCC批准。SpaceX计划拥有该频谱，而不是借用T-Mobile的。尽管仍需要时间进行V2直接到蜂窝卫星、EchoStar交易的完成（明年11月）和地面网络的建设，迪伦表示，“路径和计划现在都已官方化并清晰”。 迪伦讨论了财务影响，指出美国移动运营商市场每年约为2000亿美元。他预测Starlink Mobile到2030年代初可能获得200亿美元（10%）到400亿美元（20%）的收入，尽管服务最早要到2027年底才能推出。他强调这不仅仅是“金钱问题”，因为像800 MHz频谱这样的稀缺资产至关重要，类似于计算能力。如果与Verizon和AT&T有运营商协议的AST Space Mobile收购了该频谱，Starlink可能不得不从T-Mobile租用频率。T-Mobile已经从其网站上删除了Starlink品牌，现在将其服务称为“T-Satellite”，并与AT&T和Verizon一起成立了自己的卫星电话覆盖合资企业。这是为了避免完全依赖Starlink。凭借自己的许可证，Starlink获得了优势，无需T-Mobile的许可，可以直接或向其他运营商销售盲区覆盖服务。迪伦怀疑Starlink将直接面向客户。盘后交易中，T-Mobile股价下跌超过6%，AT&T下跌超过7%，Verizon下跌近7%。迪伦表示，由于过去的抱怨，他期待在2028年或2029年离开Verizon。他总结说，虽然Starlink Mobile还有很多需要证明的地方，但SpaceX现在拥有无与伦比的硬件、基础设施和频谱组合，并断言“Starlink的接管正在全面展开”。 在快速更新中，迪伦提到明天的视频将重点关注GrokBot和DigitalOptimus，埃隆澄清后者是“实时视频人类仿真”，与GrokBot不同。埃隆最近表示，“人们根本不知道”即将到来的AI阶段所需的“疯狂”计算能力，亚马逊禁止代理就是一个例子，埃隆认为这是竞争对手的机会。埃隆为DigitalOptimus设定的3月12日（6个月）的时间线“已经过去”。Starlink更新了其推荐计划，现在向推荐人和被推荐朋友各提供100美元。40号飞船正在返回发射场进行持续分析。埃隆否认了关于SpaceX放缓设施建设的“虚假报道”，称他们正在“迅速加速”。特斯拉的“辅助驾驶”（TAD）正在欧盟取代FSD，斯洛伐克预计很快会批准。dlumis.com上提供了FSD在欧盟进展的实时仪表板。FSD 14.3 Lite正在向一些Hardware 3所有者推出（小范围推出，与14.2 Lite的发布说明相同）。美国国家公路交通安全管理局（NHTSA）给予特斯拉30天（截止到10月30日）的时间来回应CyberCab审计查询，这比特斯拉要求的8周延期要短，原因是请求的“内容庞大”。迪伦推测特斯拉可能在CyberCab的批准上“拖延时间”。RoboTaxi服务视频显示在暴雨和夜间运行，阿肖克（Ashok）表示“风雨无阻地准备就绪”，坦帕的一名记者指出，Waymo因天气原因暂停服务，而RoboTaxi仍保持活跃。迪伦强调这证明了对处理恶劣天气和夜间操作的强烈信心。当天收盘时，特斯拉股价持平于375美元（下跌0.74%），SpaceX股价为160.57美元（下跌4.19%），纳斯达克指数下跌1.39%。

Here's a summary of the podcast: The host, Dylan Loomis, began by celebrating Elon Musk, calling him an "industrial titan," a "brilliant engineer," and one of the "greatest technology founders." He stated that few Americans have done more to advance America's national interest than Elon, whose genius has turned science fiction into reality. Elon created Tesla, the first successful new car manufacturer in over 75 years, producing the first commercially viable electric cars with the first consumer self-driving technology. Dylan likened Elon to a "modern day Thomas Edison" and congratulated him on being awarded the National Medal of Science today. He opined that Elon doesn't get enough credit outside the "Tesla bubble" on YouTube and X, and believes Elon is still in the "somewhat early days" of his achievements, adding that he enjoys seeing Elon smiling and laughing. Elon clarified the TerraFab project, stating unequivocally that SpaceX and Tesla will build and run the fab, with "zero doubt." He suggested that TSMC might sublease a portion of the TerraFab if they desire, but nothing more, and only if TerraFab gains something in return. The primary objective for TerraFab is chip output dedicated to SpaceX and Tesla, implying any TSMC sublease output would also be for these companies. Dylan was encouraged by Elon removing the option for TSMC to own and operate any part of the TerraFab, comparing any potential deal to Intel's role as a process partner manager rather than an owner-operator. Intel supplies technology, advises on setup, and gains a massive customer for its new 14A process. Unlike Intel, TSMC is already selling all its frontier chips, making Elon's sublease offer a faster way for TSMC to produce new wafers on American soil without the "greenfield route." TSMC would get power and water in Grimes County in exchange for rent, with a key question being how much of TSMC's output would be reserved for Tesla and SpaceX. Dylan prefers SpaceX and Tesla to retain as much control as possible if they pursue the TerraFab route, seeing strategic partners as beneficial for acceleration but wary of further entanglement leading to complications. The host then promoted Bolt.new, the sponsor, by describing an app he built for his fantasy football league. This app tracks live bets for a 12-leg parlay, showing which are on track, and immediately reports wins or losses, along with an all-time and current year leaderboard. He built it using Bolt.new, calling it an easy way to create an app from a verbal prompt without extensive technical work, especially since "not all of these Frontier models are good with things like front-end design." He highlighted Bolt Forge, which offers access to open-weight models, providing up to 50 times the usage. He demonstrated how he could select a model like Kimmy K3 and verbally prompt the creation of a "RoboTaxi vs. Ownership breakeven model with sliders," noting that builds help train open-weight models anonymously. He encouraged viewers to try Bolt.new, offering a free 30-day pro account with code ELECTRIFIED. Moving to Starlink news, American Airlines announced it would expand Starlink to over 1,000 mainline aircraft, with installations starting in early 2027. This builds on a previous May announcement to install Starlink on over 500 Airbus aircraft, now covering their entire mainline fleet of more than 1,000 Airbus and Boeing narrowbody and widebody aircraft. Dylan noted this was "perfect timing" for American as Delta was "dragging its feet." SpaceX secured FCC approval to launch and operate 15,000 next-gen SATs for Starlink Mobile, having applied last September. The FCC also issued a waiver, allowing SpaceX to offer wireless services via SATs without a ground-based mobile operator leasing agreement, creating a regulatory path for Starlink Mobile to compete with AT&T, T-Mobile, and Verizon. Current version 1 of Starlink Mobile (T-Satellite, with T-Mobile) uses about 650 SATs with estimated speeds around 4 Mbps, while the upcoming 15,000 SAT constellation promises 5G speeds at 150 Mbps per user. SpaceX envisions using both SATs and smaller, cost-effective ground-based femtocells for cellular coverage, including in suburban and urban areas. The FCC stated this constellation would serve cellular dead zones and increase competition. The grant allows SATs to orbit from 335 km down to 326 km (202 miles), marginally lower than Gen 1 (around 360 km), and use more radio spectrum acquired from Echo Star. SpaceX was cleared for cellular operations outside the US, though testing with Echo Star frequencies is contingent on the transaction's step two consummation (November next year). Elon commented that "Starlink Mobile direct-to-cell V2 satellites are incredible," enabling "more than 100 times the bandwidth" of the current V1 system. SpaceX further announced an agreement to acquire a nationwide low-band spectrum license portfolio, aiming to become a major US mobile carrier. Elon called this a "Very big deal." A new blog post indicated this "prime low-band spectrum" addresses a "key remaining technical gap," enabling a flexible architecture combining its satellite-to-mobile constellation with an advanced terrestrial deployment. This new low-band spectrum will provide a coverage layer to penetrate obstacles like walls and provide service inside buildings, complementing Starlink Mobile's global 2Ghz mid-band for high bandwidth. Most existing mobile devices already support this underused band. SpaceX aims to be the first operator to deploy both satellite and terrestrial spectrum as one network. Dylan highlighted public dismissals from T-Mobile's CFO (stating physics don't allow direct-to-cell to be competitive) and AT&T's CEO John Stanky (raising issues about radiating cellular signals from homes without permission), calling SpaceX's subsequent move "poetic justice." Aaron Burnett (shared by Elon) noted that phones in 2027 and 2028 will be equipped with chipsets to handle Starlink Mobile's high-throughput spectrum natively. The FCC also authorized Starlink's V3 broadband satellites' backhaul bands for existing gateway networks. This low-band spectrum is seen as the "missing puzzle piece" for indoor coverage. While the FCC still needs to approve the 800 MHz usage, the plan is for satellites to cover outdoors/rural/dead zones, and ground towers with 800 MHz to cover indoors/dense areas, with phones switching seamlessly. The 800 MHz license was previously held by Grain Management (from T-Mobile earlier this year). The deal, reportedly around $6 billion, still requires FCC approval. SpaceX plans to own the spectrum, not borrow T-Mobile's. While still needing time for V2 direct-to-cell sats, EchoStar deal closure (Nov next year), and ground network build-out, Dylan stated the "path and the plan are now both official and clear." Dylan discussed the financial implications, noting the US mobile carrier market is about $200 billion annually. He projected Starlink Mobile could capture $20 billion (10%) to $40 billion (20%) in revenue by the early 2030s, though service isn't expected until late 2027 at the earliest. He emphasized it's not "just about dollars and cents," as scarce assets like the 800 MHz spectrum are critical, similar to compute. Had AST Space Mobile, which has carrier deals with Verizon and AT&T, acquired the spectrum, Starlink might have been relegated to renting frequencies from T-Mobile. T-Mobile has already removed Starlink branding from its website, now calling its offering "T-Satellite," and along with AT&T and Verizon, formed its own satellite phone coverage joint venture. This was to avoid sole reliance on Starlink. With its own license, Starlink gains leverage, not needing T-Mobile's permission, and can sell dead zone coverage directly or to other carriers. Dylan suspects Starlink will go direct-to-customer. After-hours trading saw T-Mobile stock down over 6%, AT&T down over 7%, and Verizon just under 7%. Dylan expressed anticipation for leaving Verizon by 2028 or 2029 due to past complaints. He concluded that while Starlink Mobile has much to prove, SpaceX now possesses an unmatched combination of hardware, infrastructure, and spectrum, asserting that the "Starlink takeover is in full effect." In quick hits, Dylan mentioned tomorrow's video would focus on GrokBot and DigitalOptimus, which Elon clarified as "real-time video human emulation," distinct from GrokBot. Elon recently said "people have no idea" about the "insane" compute needed for the coming AI stage, exemplified by Amazon banning agents, which Elon sees as an opportunity for competitors. Elon's March 12th timeline for DigitalOptimus (6 months) "came and went." Starlink updated its referral program, now offering $100 to both referrer and referred friend. Ship 40 is returning to the launch site for continued analysis. Elon denied a "false report" about SpaceX slowing facility builds, stating they are "accelerating rapidly." Tesla's "Assisted Driving" (TAD) is replacing FSD in the EU, with Slovakia expected to approve it soon. A live dashboard for FSD in EU progress is available at dlumis.com. FSD 14.3 Lite is rolling out to some Hardware 3 owners (small rollout, same release notes as 14.2 Lite). NHTSA granted Tesla 30 more days (until October 30th) to respond to a CyberCab audit query, a shorter extension than Tesla's requested 8 weeks due to the "voluminous nature" of the request. Dylan speculated Tesla might be "kicking the can down the road" with CyberCab approval. RoboTaxi service videos showed operation in heavy rain and at night, with Ashok stating "ready for service rain or shine," and a Tampa reporter noting Waymo paused service due to weather while RoboTaxi remained active. Dylan highlighted this as evidence of serious confidence in handling inclement weather and night operations. The day closed with Tesla at $375 flat (down 0.74%), SpaceX at $160.57 (down 4.19%), and the NDX down 1.39%.