SpaceX 2026 年第二季度财报电话会议：星舰、星链与 AI 雄心

大家下午好。感谢各位今天参加 SpaceX 2026 年第二季度财报电话会议。我是投资者关系负责人安德莉亚·威廉姆斯。今天和我一起的有埃隆、格温和布雷特，我们正在得克萨斯州巴斯特罗普的工厂与大家连线。我们的第二季度财务业绩已于中部时间下午 3 点刚过公布，所有相关材料都已发布在我们的投资者关系网站 ir.spacex.com 上。本次网络直播中，我们的准备发言之后将进行问答环节。我们将先接受股票分析师的现场提问，然后转向投资者通过我们专门的财报问答平台提交的问题。感谢所有提前提交问题的人。在本次网络直播中，我们将讨论我们的业务前景并做出前瞻性陈述。这些陈述基于我们截至今天的预测和预期。实际事件或结果可能因多种风险和不确定性而存在重大差异，包括我们最近提交给美国证券交易委员会的文件中提到的那些，包括我们的最终招股说明书和今天发布的 10-Q 表格。此外，在本次电话会议中，我们可能会讨论新闻稿和提交给美国证券交易委员会的文件中的某些非 GAAP 指标。您可以在其中找到额外的披露，包括与可比 GAAP 指标的所有对账。对于 2026 年第二季度，我们公布了以下财务业绩，与去年同期相比：营收 78 亿美元，较 41 亿美元增长 92%；季度净亏损 5.41 亿美元，改善 4.67 亿美元；调整后 EBITDA 为 35 亿美元，较 12 亿美元增长 191%。现在我将把电话交给埃隆做开场发言。

Good afternoon everyone. Thank you for joining us today for SpaceX's second quarter 2026 earnings conference call. I'm Andrea Williams, head of investor relations. Joining me today are Elon, Gwynne, and Bret, and we are speaking to you from our facility in Bastrop, Texas. Our second quarter financial results were announced just after 3 p.m. Central time, and all relevant materials have been published to our investor relations website at ir.spacex.com. For this webcast, our prepared remarks will be followed by a question and answer session. We will take live Q&A from equity analysts and then move to questions directly from our investors that were submitted on our dedicated earnings Q&A platform. Thank you to everyone who submitted your questions in advance. During this webcast, we will discuss our business outlook and make forward-looking statements. These statements are based on our predictions and expectations as of today. Actual events or results could differ materially due to a number of risks and uncertainties, including those mentioned in our most recent filings with the SEC, including our final prospectus and today's published Form 10-Q. Additionally, during this call, we may discuss certain non-GAAP measures in our press release and our filings with the SEC. You can find additional disclosures, including all reconciliations with comparable GAAP measures. For the second quarter of 2026, we announced the following financial results as compared to the prior year quarter. Revenues of $7.8 billion, up 92% from $4.1 billion; a quarterly net loss of $541 million, an improvement of $467 million; an adjusted EBITDA of $3.5 billion, up 191% from $1.2 billion. Now I will hand the call over to Elon for his opening remarks.

谢谢。到目前为止，SpaceX 又度过了具有里程碑意义的一年。我们正在用 Starship 开发一款完全且快速可重复使用的可靠火箭，并取得了巨大进展。我们正在通过快速可靠的 Starlink 互联网，连接全球 170 多个市场的数百万更多用户。我们相信，我们正在以比任何人都快的速度大规模建设 AI 算力。我们还在显著改进我们的 AI 模型，最引人注目的是上个月发布的 Grok 4.5。现在我将分别深入探讨这些主题。

Thank you. So it's been another milestone year for SpaceX so far. We're making great progress developing a fully and rapidly reusable reliable rocket with Starship. We're connecting millions more people globally across 170 markets with fast, reliable Starlink internet. We're building AI compute capacity at scale faster than anyone else, we believe. And we're significantly improving our AI models, most notably with the release of Grok 4.5 last month. And now I'll dive into each of those topics separately.

关于 Starship，我们在过去 90 天内完成了 Starship B3 的两次成功飞行。第 13 次飞行展示了实现轨道任务并返回 Starbase 进行捕获所需的核心能力。在第 14 次飞行中，我们将首次将我们的第三代 Starlink 卫星（即我们的通信卫星）送入运行轨道。我们今年将尝试实际捕获 Starship V3 的第一级和第二级，我们甚至可能在下次飞行时就捕获飞船。实际上，很难向大多数人解释 Starship 的惊人意义。我认为，也许用“挑战轨道”来描述是理解 Starship 将对文明未来产生巨大影响的好方法。

So with Starship, we completed two successful flights of Starship B3 in the past 90 days. Flight 13 demonstrated core capabilities necessary to achieve an orbital mission and return to Starbase for catch. And on flight 14, we will have our first flight to fly our version three Starlink satellites, our communication satellites, to operational orbit. We will try to actually catch the first and the second stage of Starship V3 this year, and we could possibly catch the ship as soon as the next flight. And it's difficult to actually explain to most people just the incredible significance of Starship. And I think perhaps describing it in terms of challenge to orbit is a good way to appreciate the magnitude of the impact Starship will have on the future of civilization.

目前，我们每年通过 Falcon 向轨道运送约 2500 吨。按照正常标准，这是一个很大的数字。我认为世界其他地区每年运送约 300 吨。因此，我们每年通过 Falcon 运送地球总质量到轨道的 80% 到 90%。现在有了 Starship，我们的愿望——我认为我们会实现这些愿望——是每年向轨道运送远超 100 万吨，最终可能达到每年 1000 万吨。所以这不像你有一张图表来绘制数据。你需要一张非常非常大的桌子来放图表，才能看到图表上的任何竞争对手。而且，即使让我们的下一个最大竞争对手在柱状图上只有一像素高，你仍然需要一张非常大的柱状图来代表 SpaceX 的 Starship 飞行器。

So right now, we deliver roughly 2,500 tons a year to orbit via Falcon. And this is a big number by normal standards. The rest of the world, I think, delivers around 300 tons. So we deliver 80 to 90% of total Earth mass to orbit per year with Falcon. Now with Starship, our aspirations—and I think we will achieve these aspirations—are to deliver well over a million tons to orbit per year, and probably ultimately 10 million tons per year. So this is not like if you had a chart where you plotted things. You would need a very, very big table to put the chart on to see even any of our competitors on that chart. And to make our next biggest competitor one pixel high on the bar chart, you would still need a very big bar chart to represent the SpaceX Starship vehicle.

值得注意的是，我们实际上鼓励竞争对手模仿我们。我们不会做任何事来拖慢他们。我们实际上以公平的价格发射竞争性卫星星座，与我们向所有人收取的价格相同。所以我们确实鼓励竞争对手模仿我们，但模仿像 Starship 这样的东西相当困难。但我们不会利用专利作为地雷来阻止人们模仿。我们积极鼓励模仿。所以无论如何，当你思考这个问题时，你从每年 2500 吨增加到每年 100 万吨，甚至可能 1000 万吨。这真的是一个荒谬而深刻的差异。但这就是我们的计划，我相信我们会实现这个计划。

And it's worth noting that we actually encourage our competitors to copy us. We do nothing to slow them down. We actually launch competing satellite constellations at fair prices, the same price we charge everyone. So we do encourage our competitors to copy us, but it's quite difficult to copy something like Starship. But we don't use patents as a landmine exercise to stop people from copying us. We actively encourage it. So anyways, when you think about it, you go from 2,500 tons a year to a million, potentially 10 million tons per year. It's really a ridiculously profound difference. But that is our plan, and I think we will achieve that plan.

关于 Starlink，同样值得注意的是，Starlink V3 卫星的能力大约比 Starlink V2 卫星高一个数量级，而 V2 已经是非常强大的卫星了。我们预计将发射大约多一个数量级的 Starlink V3 卫星。这意味着交付带宽将增加大约两个数量级。即使我们的每比特变现能力下降 10 倍，这仍然意味着 Starlink 的收入将增加 10 倍。所以我认为人们真的低估了 Starlink。这是一个大事件。人们很难理解这一点，但并非不可能的是，在某个时候，Starlink 将提供世界大部分互联网服务，至少在我们被允许运营的国家，而这些国家是绝大多数国家。所以这一点值得牢记，而且这不是在遥远的未来。它不到 10 年。

Regarding Starlink, it's also worth noting that the Starlink V3 satellite is about an order of magnitude more capable than the Starlink V2 satellite, which was already a very capable satellite. And we expect to launch about an order of magnitude more Starlink V3 satellites. So that would mean a roughly two order of magnitude increase in the delivered bandwidth. Even if our monetization per bit dropped by a factor of 10, that would still mean a 10x increase in the revenue of Starlink. So I think people are really underestimating Starlink here. This is a big deal. It's kind of hard for people to wrap their minds around this, but it's not out of the question that at some point Starlink will deliver a majority of the world's internet, at least in countries where we're allowed to operate, which is the vast majority of countries. So this is important to bear in mind, and it's not in the infinite future. It's less than 10 years.

随着 AI、人形机器人、车辆机器人以及大量机器人的出现，对带宽的需求将比过去大得多。你可以想象一个人每秒消耗或产生几百比特的输出。实际上，普通人一天（即 86,400 秒）的输出不到每秒 1 比特。

And with the advent of AI, humanoid robotics, vehicle robotics, and just a massive number of robots, the appetite for bandwidth will be much greater than it has been in the past. You think of a human consuming or perhaps producing perhaps a few hundred bits per second of output. And actually, the average human is outputting less than one bit per second over the course of a day, which is 86,400 seconds.

现在的计算机，你知道，可以轻松地每秒处理数十亿比特，持续 24 小时。所以这就是为什么我预计随着 AI 和机器人技术的发展，对带宽的需求会急剧增长。而且我认为，Starlink 是唯一能够真正满足这种带宽需求的东西。

Now computers, you know, could easily do billions of bits per second continuously over a 24-hour period. So this is why I would expect the appetite for bandwidth to grow dramatically with the growth of AI and robotics. And really, I think Starlink is the only thing that can actually service that bandwidth.

关于 AI，我们在 Grok 上取得了快速进展，因为 Grok 4.5 是一个巨大的改进。Grok 4.6 可能下周就会发布，然后 Grok 4.7 大约在 3 到 4 周后发布。我们预计 AI 开发的节奏会大幅加快。

Regarding AI, we're making rapid progress on Grok because Grok 4.5 was a huge improvement. We have Grok 4.6 coming out probably next week, and then Grok 4.7 is about 3 or 4 weeks from today. We expect the cadence of AI development to improve dramatically.

然后，Grok 5 应该会在今年年底前发布，我们将整合 SpaceX 的全部数据。基本上，SpaceX 在四分之一世纪里产生的所有数据，数量巨大，都将纳入 Grok 的训练。我们认为这将使 Grok 成为迄今为止最优秀的工程师。

Then with Grok 5, which should be out before the end of this year, we will be incorporating the entire corpus of SpaceX data. Basically all the data that SpaceX has ever produced, which is a tremendous amount over the course of a quarter century, will be incorporated into Grok training. We think this will make Grok by far the best engineer.

此外，当然，我们也在向其他人提供算力，而且我们建设和部署算力的速度比任何人都快。我们的增长速度肯定比任何人都快，我们的算力部署效率也是最高的。我们预计今年年底将拥有超过 2 吉瓦的算力。到明年年底，我们的累计在线算力可能会高出几倍。可能更接近 10 吉瓦，而不是 5 吉瓦。

In addition, of course, we are providing compute to others, and we are building and deploying compute faster than anyone else. Our rate of growth certainly is faster than anyone else's, and our efficiency of compute deployment is also the highest. We expect to end this year with over two gigawatts of compute. Probably our cumulative compute online by the end of next year will be several times higher. It may be closer to 10 gigawatts than 5 gigawatts.

展望未来，我们决定只使用 Nvidia 的架构，因为我们认为 Blackwell 架构是最好的架构。我们认为它是最好的 AI 计算机，我们非常重视与 Nvidia 在多层面的紧密合作和伙伴关系。所以我们将独家使用 Nvidia。

Going forward, we've decided to build exclusively on Nvidia because we think the Blackwell architecture is the best architecture. We think it's the best AI computer, and we greatly value our close cooperation and partnership on many levels with Nvidia. So we're exclusive to Nvidia.

这些新上线的算力显然将进一步有利于 Grok 的开发，我们收购 Cursor 也会如此，这笔交易应该很快就会完成。我想我们已经通过了几乎所有的监管障碍。

This new capacity coming online will obviously further benefit the development of Grok, as will our Cursor acquisition, which should close quite soon. I think we're through almost all of the regulatory hurdles.

关于 Starmind AI 卫星，它本质上将是一台优化的 Blackwell NVL72 计算机，这不是某种遥远的未来事物。我们预计明年开始发射这些卫星。我们认为 NVL72 VR 计算机的设计比典型的、比如标准机架式设计要好得多。我们预计将在地面和轨道上部署它，因为我们认为这将是对普通 NVL72 机架的根本性简化。成本更低，效率更高。所以如果我们把它放到太空，为什么不也放到地面呢？所以我认为那会非常酷。

With respect to the Starmind AI satellite, which will be essentially an optimized Blackwell NVL72 computer, this is not some sort of far-future distant thing. We expect to start launching these next year. We think the design of the NVL72 VR computer is a much better design than typical, say, having a standard rack-style design. We expect to actually deploy this on the ground as well as in orbit because we think it's going to be a radical simplification of the normal NVL72 rack. It will cost less, it will be more effective. So if we're going to put it in space, why not put it on the ground? So I think that's going to be pretty cool.

总之，SpaceX 团队正在解决人类历史上一些最困难的工程问题。我认为团队之所以成功，是因为我们在很多方面拥有有史以来最聪明、最敬业的团队。能与这样的团队共事，我感到无比自豪，我要感谢团队付出的巨大努力，也感谢一路支持我们的投资者。

In conclusion, the SpaceX team is solving some of the hardest engineering problems in the history of humanity. I think the team is succeeding because we've got, in a lot of ways, the smartest, most dedicated team of humans that has ever existed. I'm incredibly proud to work with such a team, and I'd like to thank the team for their incredible hard work, and also thank the investors who have supported us along the way.

谢谢 Elon。我们第二季度表现卓越，今天我将重点介绍运营和商业亮点，以及我们未来优先推进的举措。

Thanks Elon. We had an exceptional second quarter, and today I'll focus on the operational and commercial highlights as well as the initiatives we are prioritizing going forward.

从我们的太空业务开始，我们仍然是全球领先的发射服务提供商，今年上半年共完成 78 次发射，向轨道运送了 1,041 吨质量，主要分配给我们的内部 Starlink 任务，但也有出色的客户任务。虽然我们的发射能力是支持连接业务的关键，但我们继续拥有非常强劲的商业订单，以及非常强劲且不断增长的美国政府任务需求，这推动了该板块的收入。

Starting with our space business, we remain the leading launch provider globally with 78 total launches and 1,041 tons of mass to orbit delivered in the first half of this year, primarily allocated to our own internal Starlink missions, but with great customer missions as well. While our launch capabilities are key to supporting our connectivity business, we continue to have a very robust commercial manifest and a very robust and growing US government mission demand which drives revenue for the segment.

我们目前正以最高的 Falcon 发射频率进行发射，并且正处于 Starship 投入运营的边缘。正如 Elon 所说，我们 V3 Starship 飞行器的前两次飞行使我们处于有利位置，能够实现近期目标：进入轨道、捕获并重复使用飞船和助推器，以及部署我们的 V3 Starlink 宽带卫星。

We're currently launching at our highest Falcon cadence and are on the precipice of operationalizing Starship. As Elon said, the first two flights of our V3 Starship vehicles put us in an excellent position to achieve our near-term goals of reaching orbit, catching and reusing the ships and boosters, and deploying our V3 Starlink broadband satellites.

说到这个，Starlink 本季度表现突出，得益于消费者、企业和政府业务的持续增长。第二季度，我们在全球净增超过 170 万 Starlink 消费者订阅用户。这是我们迄今为止新增客户最多的一个季度，高于第一季度新增的 140 万。我们通过将 ARPU（即每用户平均收入）稳定在每月 66 美元（与第一季度持平）实现了这一目标。

Speaking of which, Starlink had a standout quarter driven by continued growth across our consumer, enterprise, and government businesses. During the second quarter, we added net more than 1.7 million Starlink subscribers globally consumer. This reflects our best quarter of new customers to date, an increase from the 1.4 million we added during the first quarter. We achieved this by holding ARPU, which is the average revenue per user, stable at $66 per month compared to the first quarter.

截至第二季度末，我们的服务已在 167 个市场可用，此后我们继续开通新市场，很快会有一些非常令人兴奋的公告，甚至可能就在今天。

We ended the second quarter with service availability in 167 markets and have continued to activate new markets since then, with some very exciting announcements coming soon, maybe even as early as today.

截至 6 月 30 日，我们在轨运营的 Starlink 宽带和移动卫星星座已增至约 10,200 颗，其中 9,600 颗宽带卫星提供约 800 太比特每秒的总下行容量。此前，在最近一次 Starship 发射中，我们对 V3 Starlink 卫星进行了成功的快速测试，期间我们通过每条激光链路连接了每颗卫星。

As of June 30th, our constellation of operational Starlink broadband and mobile satellites in orbit grew to roughly 10,200, with our 9,600 broadband satellites delivering roughly 800 terabits per second of total downlink capacity, following a successful speedrun test of the V3 Starlink satellites on the most recent Starship launch, during which we connected every satellite through every laser link.

我们计划在即将到来的 Starship 任务中，将 V3 卫星部署到星座中用于运营。通过 V3 卫星，我们能够为 Starlink 星座增加大量容量，这将使我们能够继续提供更好的服务（我认为现在已经很好了），同时服务全球越来越多的客户。

We intend to deploy V3 satellites into the constellation for operational use on the upcoming Starship missions. The significant amount of capacity we're able to add to the Starlink constellation from the V3 satellites will enable us to continue providing even better service, I think it's pretty great already, but to do so while serving more and more customers over the world.

事实上，在未来几年，我们预计 Starlink 将占据全球互联网流量的重要部分，Elon 也谈到了这一点。

In fact, in the years ahead, we expect Starlink will represent a significant portion of global internet traffic, which Elon also talked about.

我们也看到 Starlink 在企业和政府业务方面有显著的增长机会。我们相信，随着我们继续获得市场份额，这些市场的收入有可能达到至少与消费者业务相当，甚至可能超过消费者业务的规模。

We also see significant growth opportunities for Starlink in enterprise and government on the enterprise and government side of the business. We believe revenue from these markets have the potential to reach a scale at least comparable but likely exceed our consumer business as we continue to gain share.

第二季度，我们与美国航空签署了一项重要协议，并启动了与西南航空、维珍大西洋航空、伊比利亚航空和 Heirlingus 等新合作伙伴的服务。我们已成为主要航空公司的首选供应商，客户反馈极佳，但仍有很大的增长空间。

In the second quarter, we signed a major agreement with American Airlines as well as activated our service with new partners including Southwest, Virgin Atlantic, Iberia, and Heirlingus. We have become the provider of choice among major airlines with outstanding customer feedback but still have significant room for growth.

事实上，我们从一家航空公司客户那里听说，令人震惊的是，客户宁愿乘坐较短的支线航班，也不愿乘坐直飞航班，以确保他们乘坐的是 Starlink 激活的航班。他们在业内从未见过这种情况。

In fact, we heard from one of our airline customers, shockingly customers are flying shorter hop flights instead of direct so that they could ensure that they're on a Starlink activated flight. They'd never seen this in the business before.

Gwynne，你能先给我们概述一下本季度的亮点吗？

Gwynne, could you start by giving us an overview of the quarter's highlights?

在政府方面，我们在第二季度赢得了超过 60 亿美元的美国合同，支持了重要的太空军项目，这些项目为我们的国家提供了关键任务的通信和传感能力。我们预计这一领域在未来一年还有更大的增长空间。第二季度，我们还与包括软银、NT、DOKO 和新西兰 Spark 在内的国际运营商启动了新的 Starlink 移动合作伙伴关系。我们的近期优先事项是在明年晚些时候整合 65 兆赫的 Echoar 频谱之前，在 Starship 上发射移动 V2 卫星。这一频谱转移最近获得了 FCC 的批准，代表了 Starlink Mobile 的基础性竞争优势。

On the government side, we won more than $6 billion in US contracts in Q2, supporting major Space Force programs that offer our nation mission-critical communications and sensing capabilities. And we see even more room for growth in this sector in this coming year. In Q2, we also launched the new Starlink mobile partnerships with international carriers including SoftBank, NT, DOKO, and Spark New Zealand. Our near-term priority is launching mobile V2 satellites on Starship ahead of integrating the 65 megahertz of Echoar Spectrum later next year. This spectrum transfer was recently approved by the FCC and represents a foundational competitive advantage for Starlink Mobile.

那 AI 方面呢？算力扩张进展如何？

And what about the AI side? How is the compute expansion going?

在 AI 领域，行业正朝着越来越高的算力需求迈进。我们正在迅速扩大算力容量，以满足我们自身 Grok 扩展的需求，以及谷歌和 Anthropic 等其他领先公司的需求。就我们自己的 Grok 模型而言，Grok 4.5 发布后的企业反馈到目前为止非常积极。7 月发布后，Token 消耗量一开始就翻了三倍，我们继续看到该模型在使用量和变现方面的强劲信号。

In AI, the industry is hurdling towards ever more compute demands. We're rapidly expanding our compute capacity to meet our own needs as Grok expands as well as those for other leading companies such as Google and Anthropic. In terms of our own Grok models, the enterprise feedback on rollout of Grok 4.5 has been very positive so far. Token consumption tripled out of the gate after the July release, and we continue to see strong signals on usage and monetization for the model.

那收购 Cursor 团队呢？

And the Cursor team acquisition?

我们期待欢迎 Cursor 团队加入 SpaceX，整合我们的工程能力，并开始受益于合并后的销售能力。

We're looking forward to welcoming the Cursor team to SpaceX to integrate our engineering and begin to benefit from a combined sales capability.

总体而言，您如何描述公司的势头？

Overall, how would you describe the company's momentum?

回顾来看，我们作为一家企业的势头非同寻常。我们的工程团队正在扩展 Starship 和下一代卫星，而销售团队则在扩大 Starlink 航空、移动和全球市场。与此同时，我们正在增加额外的算力容量，并在更广泛的客户群中扩展和深化我们的 AI 合作伙伴关系。尽管我从事这一行业已有 24 年，但对未来几个月我感到无比兴奋。有这么多激动人心、令人惊叹的事情在发生，感觉我们真的像是重新开始了一样。

In review, our momentum as a business is extraordinary. Our engineering teams are scaling Starship and our next-gen satellites while our sales teams expand Starlink aviation, mobility, and global markets. At the same time, we're bringing on additional compute capacity and expanding and deepening our AI partnerships across a broader set of customers. And though I've been in this business for 24 years, I could not be more excited about the months ahead. With all the exciting and amazing things going on, it really feels like we're just getting started all over again.

谢谢你，Gwynne。现在，Bret，你能给我们介绍一下财务情况吗？

Thank you, Gwynne. Now, Bret, could you walk us through the financials?

谢谢你，Gwynne。我们在第二季度展示了强劲的业绩。每个板块的收入增长都在加速。我们同比将净亏损收窄了近一半，并且调整后 EBITDA 的增长速度明显快于收入。重要的是，我们通过 IPO 和首次投资级债券发行显著加强了资产负债表。我们在全球发射领域的领先地位、Starlink 的经常性收入以及 AI 基础设施的快速增长，这些因素的结合继续推动运营杠杆的改善。现在，我将逐一介绍各板块的财务情况。

Thank you, Gwynne. We demonstrated strong results in Q2. Revenue growth accelerated across every segment. We narrowed our net losses by nearly half on a year-over-year basis, and we grew adjusted EBITDA meaningfully faster than revenue. Importantly, we significantly strengthened the balance sheet through our IPO and inaugural investment grade bond offering. The combination of our global launch leadership, recurring Starlink revenue, and rapid growth in AI infrastructure continues to drive improved operating leverage. Now, I'll walk through the financials for each segment.

对于太空板块，收入环比增长 55%，同比增长 29%，达到 9.62 亿美元，这得益于更多更大的客户发射和有利的客户组合。板块成本和支出同比增长 3.89 亿美元，我们继续加速对 Starship 项目的研发投资。Starship 的目标是将有效载荷能力提高四倍，并将发射成本比我们的 Falcon 9 火箭降低 10 倍，从而在我们所有业务板块中释放出巨大的能力。我们继续在建设支持每年数千次 Starship 发射所需的基础设施方面取得进展，包括加速 Raptor 和运载火箭的生产、建设我们的千兆湾，并在 Starbase 和卡纳维拉尔角的 39A 和 37 号发射台激活多个发射台方面取得重大进展。本季度太空板块调整后 EBITDA 亏损 2.05 亿美元，主要反映了这些较高的研发支出。

For the space segment, revenue grew 55% sequentially and 29% year-over-year to $962 million, driven by a higher number of larger customer launches and a favorable customer mix. Segment costs and expenses rose $389 million year-over-year, and we continue to accelerate R&D investments in our Starship program. Starship aims to quadruple payload capacity and reduce launch costs by 10 times compared to our Falcon 9 rocket, unlocking significant capabilities across all of our business segments. We continue to make progress building the infrastructure required to support thousands of Starship launches per year, including accelerating Raptor and launch vehicle production, the buildout of our gigabays, and making significant progress towards activating multiple launch pads at Starbase and in Cape Canaveral at pad 39A and pad 37. Space adjusted EBITDA was a loss of $205 million for the quarter, primarily reflecting these higher R&D expenses.

接下来是连接板块，收入达到 43 亿美元，环比增长 32%，同比增长 66%。增长得益于 Starlink 订户的创纪录净增以及企业和政府收入的急剧增加。Starlink 订户 ARPU 为 66 美元，与第一季度持平。随着我们继续全球扩张，我们正在采取本地化的市场进入策略，确保我们的产品和服务适应当地需求。虽然我们的地理扩张可能会随着时间的推移拉低混合 ARPU，但我们预计持续的订户增长势头将带来强劲的收入增长。企业和政府收入同比增长 108%，代表了持久的收入来源，有助于整体板块利润率的强劲。正如 Gwyn 提到的，我们看到了这一板块的巨大增长机会，例如，在最大的板块之一——航空领域，我们的渗透率还不到 10%。

Moving to the connectivity segment, revenue was up to $4.3 billion, up 32% sequentially and 66% year-over-year. Growth was driven by record net additions of Starlink subscribers and a sharp increase in enterprise and government revenue. Starlink subscriber ARPU was $66, unchanged compared to Q1. As we continue our global expansion, we're taking a localized go-to-market approach that ensures our product and service offerings fit local needs. While our geographic expansion may drive down blended ARPU over time, we anticipate continued subscriber momentum to deliver strong revenue growth. Enterprise and government revenue grew 108% year-over-year and represents a durable source of revenue contributing to strong segment margins overall. As Gwyn mentioned, we see enormous opportunity for growth in this segment as we remain, for example, less than 10% penetrated in one of the largest segments being aviation.

连接板块的总成本和支出增加了 9.7 亿美元，同比增长约 58%。这一增长是由于支持我们收入增长的支出增加，包括卫星星座的增长、对下一代版本 3 卫星的研发投资增加，以及额外的营销支出。正如 Elon 提到的，我们的版本 3 卫星预计将提供比当前一代多 10 倍的宽带容量和数据密度，我们认为这已经相当惊人，并且对于支持持续的订户增长和更高价值的服务层级至关重要。连接板块的运营收入同比增长 79%，达到 17 亿美元，超过了收入增长，推动了近三个百分点的运营利润率扩张。连接板块调整后 EBITDA 同比增长 64%，达到 26 亿美元。

Total costs and expenses for the connectivity segment increased $970 million or approximately 58% year-over-year. The increase was driven by higher spend to support our revenue growth, including growth in our satellite constellation, increased R&D investment for our next-gen version 3 satellites, and incremental marketing spend. As Elon mentioned, our version 3 satellites are expected to deliver 10 times the amount of broadband capacity and data density from our current generations, which we believe are already pretty amazing and which are essential to supporting continued subscriber growth and higher value service tiers. Income from operations from the connectivity segment increased 79% year-over-year to $1.7 billion, outpacing revenue growth and driving nearly three points of operating margin expansion. Connectivity segment adjusted EBITDA rose 64% year-over-year to $2.6 billion.

AI 板块收入为 26 亿美元，环比增长 213%，同比增长 247%。这一增长主要得益于新的云服务协议以及 Grok 和 X 订阅收入的增长。广告收入环比增长 7%，因为我们彻底改革了广告平台技术，以实现 AI 功能并增强所有客户类型的自助服务。在第二季度，我们签订了云服务协议，提供对我们 Colossus 和 Colossus 2 站点算力容量的访问。这些协议的初始爬坡为本季度贡献了 16 亿美元的增量 AI 基础设施收入。

AI segment revenue was $2.6 billion, up 213% sequentially and 247% year-over-year. The increase was driven primarily by new cloud services agreements as well as growth in Grok and X subscription revenue. Advertising revenue grew 7% sequentially as we overhauled our advertising platform technology to enable AI functionality and enhance self-service for all customer types. In Q2, we entered into cloud services agreements providing access to compute capacity at our Colossus and Colossus 2 sites. The initial ramp from these agreements contributed $1.6 billion of incremental AI infrastructure revenue in the quarter.

AI 板块的总成本和支出同比增长 16 亿美元，反映了研发支出的增加，主要由于我们加速算力部署导致基础设施支出增加。我们显著收窄了 AI 板块的净运营亏损至 13 亿美元，并在第二季度实现了该板块调整后 EBITDA 转正，达到 11 亿美元。新托管协议带来的增量收入产生了高增量 EBITDA 利润率，因为我们变现了可用的算力容量。我们在第二季度末拥有 1.4 吉瓦的铭牌算力，高于第一季度的 1 吉瓦和一年前的 400 兆瓦。我们预计今年年底将超过 2 吉瓦的算力容量。

Total costs and expenses for the AI segment rose $1.6 billion year-over-year, reflecting increased R&D primarily due to higher infrastructure spend as we accelerated compute deployment. We meaningfully narrowed our AI segment net operating loss to $1.3 billion and turned adjusted EBITDA positive for the segment in the second quarter at $1.1 billion. The incremental revenue from new hosting deals generated high incremental EBITDA margins as we monetized available compute capacity. We ended the second quarter with 1.4 gigawatts of nameplate compute, up from 1 gigawatt in Q1 and 400 megawatts a year earlier. We expect to end this year at over 2 gigawatts of compute capacity.

第二季度公司总资本支出约为 184 亿美元，其中约 158 亿美元用于支持 AI 算力基础设施。其余部分用于资助持续的 Starship 和发射基础设施、卫星生产以及全球地面站扩展。

Total company capital expenditures in the second quarter were approximately $18.4 billion, of which roughly $15.8 billion supported AI compute infrastructure. The remainder funded ongoing Starship and launch infrastructure, satellite production, and global ground station expansion.

谢谢你，Bret。这是一个全面的概述。

Thank you, Bret. That's a comprehensive overview.

此外，本季度我们根据与待完成的 Echoar 交易相关的频谱信贷协议支付了 8.56 亿美元。我们的资本配置优先级依然清晰且纪律严明。我们计划继续在三个我们认为能带来最高长期回报的领域积极投资：Starship 开发与生产规模、下一代 Starlink 宽带与移动星座，以及 AI 算力基础设施。在这三项投资中，我们始终关注资本效率、投资回报以及维持强大的投资级资产负债表。从资本市场角度看，第二季度具有变革性。我们完成了首次公开募股，净募资约 857 亿美元，随后又发行了 250 亿美元的首批投资级优先票据，部分用于偿还 200 亿美元的过桥贷款。这些优先票据分五个批次发行，加权平均利率为 5.855%，平均期限为 11.7 年。本季度末，我们持有 1000 亿美元的现金、现金等价物和有价证券，以及 475 亿美元的积压订单。展望未来，我们三个业务板块的需求依然强劲，尤其是云服务安排。我们签署的每份协议都带来了越来越有利的经济效益。正如 Elon 所提到的，我们预计算力市场的供需失衡将持续。当前的经济效益已使我们在算力上的新增资本部署回收期不到一年。例如，在第三季度的头几周，我们已经额外签约了 67 亿美元的云服务收入，该收入将在今年 10 月开始爬坡的六个月内产生。我们相信，这使我们有望在今年年底前达到 1000 亿美元的 ARR（年化收入运行率），这基于我们今年 12 月的预期收入，并包括 Cursor 的贡献。至此，我将把电话交还给 Andrea。

Additionally, we paid out 856 million during the quarter under our spectrum credit agreement related to our pending Echoar transaction. Our capital allocation priorities remain clear and disciplined. We plan to continue to invest aggressively in the three areas we believe generate the highest long-term returns: Starship development and production scale, next-generation Starlink broadband and mobile constellations, and AI compute infrastructure. Across all three of these investments, we remain focused on capital efficiency, return on investment, and preservation of a strong investment-grade balance sheet. The second quarter was transformative from a capital markets perspective. We completed our initial public offering, raising approximately $85.7 billion in net proceeds, and followed that with a $25 billion inaugural investment-grade senior notes offering that was partially used to repay our $20 billion bridge loan. The senior notes were issued across five tranches with a weighted average interest rate of 5.855% and average maturity of 11.7 years. We ended the quarter with $100 billion of cash, cash equivalents, and marketable securities and $47.5 billion in backlog. Looking ahead, we continue to see robust demand in all three of our business segments, but in particular in our cloud services arrangements. We see increasingly favorable economics with each agreement we sign. And as Elon mentioned, we expect the supply-demand imbalance in the compute market to continue. The current economics have translated into a less-than-one-year payback on our new capital deployments for compute. For example, in the first few weeks of the third quarter, we've already contracted an additional $6.7 billion of cloud services revenue over a six-month period that begins ramping starting in October of this year. We believe this puts us on a trajectory, including contribution from Cursor, to reach a $100 billion of ARR, or annualized revenue run rate, by the end of this year based on our expected revenue in the month of December of this year. With that, I'll turn the call back to Andrea.

谢谢 Elon、Gwynne 和 Bret。我们今天的准备发言到此结束，接下来进入问答环节。我们将首先接受股票分析师的问题。请按星号键 1 加入提问队列，并请将问题限制在一个。接线员，请继续队列。

Thank you, Elon, Gwynne, and Bret. This concludes our prepared remarks for today's Q&A session. We will take our first questions from the equity analysts. Please press star one to join the question queue and also please limit yourself to one question. Operator, please proceed with the queue.

我们现在开始问答环节。如果您想提问，请按星号键 1 举手。要撤回问题，请再次按星号键 1。我们要求您在提问时拿起听筒，以确保最佳音质。如果您在本地静音，请记得取消设备的静音。请稍候，我们正在整理问答名单。

We will now begin the question and answer session. If you would like to ask a question, please press star one to raise your hand. To withdraw your question, press star one again. We ask that you pick up your handset when asking a question to allow for optimum sound quality. If you are muted locally, please remember to unmute your device. Please stand by while we compile the Q&A roster.

您的第一个问题来自高盛的 Eric Sheridan。您的线路已接通，请讲。

Your first question comes from the line of Eric Sheridan with Goldman Sachs. Your line is open. Please go ahead.

非常感谢您回答问题，也感谢准备发言中的所有细节。也许就一个关于连接板块的问题。您能谈谈该业务板块中企业和政府合同的积压情况，以及我们如何看待这些积压转化为收入，这对未来几年的增长速度和节奏可能意味着什么？非常感谢。

Thanks so much for taking the question and thanks for all the details in the prepared remarks. Maybe just one on the connectivity segment. Can you discuss the backlog of enterprise and government contracts in that segment of the business and how we think about that backlog converting to revenue and what that might mean for the pace and cadence of growth in the next couple years? Thanks so much.

是的，我们对企业业务非常乐观。我之前提到，我们从政府那里获得了超过 60 亿美元的合同。这些可以看作是分批的，也是我们将上线的新能力的开端。其中一些我们将不得不参与竞争。但从根本上说，我们已经签订了合同，以满足许多应用场景，包括我们最近宣布的以及未来将开展的其他场景。企业收入非常稳定。我们从未失去过企业客户。他们对目前 Starlink 的能力非常满意。Bret 提到，我们在航空业的渗透率仅为 10%，这无疑为未来的收入增长提供了很大的空间。在海事方面，我认为 Starlink 极大地打开了 TAM（总可寻址市场）。Starlink 提供了目前许多船只无法获得的能力，因为它们负担不起 VSAT。我们基本上可以用 Starlink Maritime 为这些船只配备设备。所以我对企业业务非常乐观，对政府业务也非常乐观，并且对我们在这些市场中看到的客户粘性感到非常满意。

Yeah, we are quite bullish on our enterprise activities. I mentioned earlier that we received from the government over $6 billion in contracts. Those are kind of tranches and beginnings of additional capability that we will bring online. Some of which we will have to compete for. But fundamentally, we have the contracts in place to fulfill many phenomenologies, both the ones we've recently announced as well as additional ones that we'll be working on going forward. Enterprise revenue is quite sticky. We have never lost an enterprise customer. They're quite happy with the Starlink capability that we have right now. Bret mentioned we are only 10% penetrated in the aviation industry, which definitely represents a lot of headroom for revenue going forward. Maritime, I think, Starlink opens up the TAM dramatically. Starlink offers the capability that ships that don't really have much connectivity right now because they can't afford VSATs. We can basically outfit those ships with Starlink Maritime. So I'm quite bullish on enterprise, very bullish on government, and quite happy with the stickiness that we've seen in those markets.

是的，我稍微详细说明一下。显然，为了获得大量企业收入，你必须展示出非常高的可靠性和非常高的正常运行时间。在早期，Starlink 的正常运行时间并不高。而且有很多企业客户，我想说大多数企业客户，仍然有那种早期 Starlink 的体验，当时服务有点不稳定。所以我们实际上正在组建一个大型企业销售团队，去与企业客户（主要是企业客户和政府客户）会面，向他们展示今天 Starlink 系统令人难以置信的正常运行时间和低延迟。这样他们就会意识到，他们实际上可以把 Starlink 视为主要提供商，而不是备份提供商。但这需要走出去与人交谈，教育他们了解 Starlink 系统的当前状态及其未来发展方向。我预计企业收入将大幅超过消费者收入。

Yeah, I'll just elaborate a little bit on that, which is that obviously in order to secure a large amount of enterprise revenue, you have to demonstrate very high reliability, very high uptime. And in the early days, Starlink did not have high uptime. And there's a lot of enterprise customers, most enterprise customers, I'd say, that still have that maybe early Starlink experience where it was a little patchy. So we actually are building up a large enterprise sales team just to go out there, meet with enterprise, basically corporate and government customers, show them the incredible uptime and low latency of the Starlink system today. So they realize that they can actually treat Starlink as a primary provider, not sort of as a backup provider. But this requires just going out there talking to people, educating them about the current state of the Starlink system and where it's going. And I would expect enterprise revenue to substantially exceed consumer revenue.

下一个问题，接线员。

Next question, operator.

您的下一个问题来自摩根士丹利的 Adam Jonas。您的线路已接通，请讲。

Your next question comes from the line of Adam Jonas with Morgan Stanley. Your line is open. Please go ahead.

谢谢 Elon。第一个问题问您。只是想确认一下，您说过预计到 2027 年底名义算力将接近 10 吉瓦，更接近 10 而不是 5。您对此有清晰的展望吗？在许可、芯片和涡轮机方面，您的展望有多确定？然后我还有一个后续问题问 Gwynne。谢谢。

Thanks Elon. First one for you. Just confirming you said you expect to be closer to 10 gigawatts of nameplate compute by end of '27, closer to 10 than five. Do you have line of sight, how confident is your line of sight on permitting chips and turbines for that? And then I have a follow-up for Gwynne. Thanks.

所以我们实际上目标是在电力上线、电力冷却和电气设备方面远远超过那个吉瓦数。所以我们的初步目标是在明年年底前，在电力和冷却层面实际达到 20 吉瓦。现在，我不认为我们能实现 20 吉瓦，但我们希望有一系列项目，到明年年底累计达到 20 吉瓦。其中一些可能不会完全按时完成，但我预计在电厂层面，我们可能仍会有接近 15 吉瓦的规模。所以假设大约四分之一的项目比预期耗时更长。我们的目标是拥有比 GPU 多得多的电力、冷却和电气设备。

So we're actually aiming to far exceed that gigawatt number in terms of power online, power cooling, and electrical equipment. So our tentative target is to actually have 20 gigawatts at the power and cooling level online by the end of next year. Now, I don't think we're going to achieve 20 gigawatts, but we want to have a series of projects that cumulatively come to 20 gigawatts by the end of next year. Some of them won't pan out exactly on time, but I would expect that we still probably have, at the power plant level, something close to 15 gigawatts. So assuming that maybe a quarter of the projects take longer than expected. So our goal is to have far more power, cooling, and electrical equipment than we have GPUs.

考虑到 GPU 相对于系统其余部分的成本，这是合乎逻辑的做法。SpaceX 作为一家公司，在硬件方面非常出色。事实上，我认为可能除了中国以外，SpaceX 和特斯拉是地球上最擅长硬件的两家公司。甚至可能包括中国在内。所以我们把用于火箭和卫星的一小部分专业知识，应用到扩展地面数据中心上。我们发现，即使是将建造火箭（这极其困难）所学到的一小部分应用到数据中心，也能带来巨大的收益。所以我们预计将拥有远超所需的电力和冷却能力。而且我们与英伟达的共识是，明年我们将获得他们 GPU 的很大一部分。

That's the logical thing to do given the relative expense of GPUs versus balance of system. SpaceX as a company is incredibly good at hardware. In fact, I'd say probably outside of China, SpaceX and Tesla are the two best companies on Earth at hardware. And perhaps even including China. So we're taking a small amount of the expertise that we use for rockets and satellites and applying that to scaling terrestrial data centers. We're finding that even a small amount of what we've learned building rockets, which are incredibly difficult, applied to data centers yields tremendous benefits. So we expect to have far in excess of the power and cooling that's needed. And our understanding with Nvidia is that we will receive a very significant percentage of their GPUs next year.

好的，谢谢，埃隆。格温，我有个后续问题。你最近被问到 SpaceX 投资故事中哪些部分被低估了，你说是机器人技术。你能详细说明一下为什么这么认为吗？机器人如何融入 SpaceX 的计划？

Okay. Thanks, Elon. Just a follow-up, Gwynne. You were recently asked what is one of the more underappreciated parts of the SpaceX investment story, and you said robotics. Can you elaborate on why you believe that? How do robots fit into the SpaceX plan?

让我试着为你整理一下。我想我可能说过的是，我们未来所需的连接量被低估了。我认为机器人技术，尤其是与 AI 相关的机器人技术，比如用于汽车和人形机器人的，将极大地增加对连接的需求。埃隆在他之前的发言中谈到过这一点。

Let me try to put something together for you. I think what I might have said is underappreciating the amount of connectivity that we will need going forward. I think robotics, especially robotics tied to AI like for autos and humanoid robots, will dramatically increase the demand for connectivity. Elon talked about that in his early remarks.

你可以把我们的卫星看作是机器人。它们是自主的，不需要维护。然后，在看似遥远的未来（但会比你想的来得更快），在月球上使用机器人来扩大月球上的制造业——我知道现在听起来像超级科幻，但这将会发生——将使我们能够在月球上建造质量加速器。如果你在月球上有质量加速器，并且在月球上有太阳能和散热器生产，那么你或许可以将送入太空的智能规模扩大到地球经济的一千倍。甚至可能是一百万倍。所以我们将在月球上着陆大量货物。我们将在月球上建造工厂。机器人对此会很有帮助。然后，随着我们成长为二型文明，问题就变成了：你想要多少智能？

You can think of our satellites as being robots. They are autonomous and do not need servicing. And then, in what may seem the far future but will come faster than you think, using robots on the moon to scale up manufacturing on the moon—which I know sounds like super sci-fi right now but it's going to happen—will enable us to build the mass accelerator on the moon. And if you have a mass accelerator on the moon and you have solar and radiator production on the moon, you can probably scale to a thousand times the economy of Earth in terms of intelligence launch to space. Maybe even a million times. So we are going to land a lot of tonnage on the moon. We're going to build the factories on the moon. The robots will be helpful with that. And then it's like, well, how much intelligence would you like, as we grow towards being a Type II scale civilization?

这回答了你的问题吗，亚当？

Did that get that for you, Adam?

也许他问完了。

Maybe he's done.

下一个问题来自 Croup 的 John Goden。您的线路已接通，请讲。

Your next question comes from the line of John Goden with Croup. Your line is open. Please go ahead.

嘿，谢谢您回答我的问题。SpaceX 迄今为止的一个特点是，组织在其发展过程中非常高效地使用了资本，但同样众所周知的是，为了支持您从这里开始的愿景，资本需求是巨大的。显然，公司发展很快。所以，我希望您能多谈谈各细分领域的资本支出展望。谈谈哪些领域购买而非建造可能会加速增长。显然，移动战略在这里浮现在脑海中，以及关于下一步极端垂直整合（如 terafab）的资本需求的任何相关想法，以及支持未来增长所需的任何东西。谢谢。

Hey, thanks for taking my question. One of the hallmarks of SpaceX to date is how efficiently the organization has used capital so far in its journey, but it's also no secret that the capital needs are significant to support your vision from here. Obviously, the company's moving fast. So, I was hoping you could talk a bit more about the capex outlook across the segments. Touch upon any areas where buying versus building might accelerate growth. The mobile strategy obviously comes to mind there, and any related thoughts on capital needs for next steps in extreme vertical integration like terafab, anything that's needed to support future growth. Thanks.

是的，不，我感谢这个问题。我认为我们为多年来在资本配置方面的管理感到非常自豪，当然。而且我认为我们正在继续沿着这条路走下去。如果你看看，我们迄今为止非常高效，而且我认为将继续如此。你知道，如果你看看今年剩余时间，具体回答资本支出，我认为你应该认为接下来的两个季度在资本支出水平上与当前季度非常相似。但我确实认为，从长期来看，我认为资本支出，从我们的角度来看，所有的资本支出都不一样。我们现在看到的情况是，我们能够部署，特别是在 AI 计算方面，我们能够以不到一年回本的方式部署资本。这几乎就像你的资本变成了本来会放在销售成本中的项目，因为它在货币化和转化为收入方面进展如此之快。这与 SpaceX 历史上的情况不同，在 SpaceX，我们建造的资产将在未来几十年内从发射场中受益。因此，我们当然始终从优化资本投资回报率的角度来管理。

Yeah, no, I appreciate the question. I think we are very proud of how we have managed capital allocation over the years, for sure. And I think we are continuing down that path. If you look at it, we have been very efficient to date, and I think will continue to be. You know, if you look at the rest of this year, to answer specifically on capex, I think you should probably think that the next two quarters are very similar to the current quarter from a capex level perspective. But I do think, as you look longer term, I think capex, really for our perspective, all capex is not the same. What we're seeing right now is a situation where we're able to deploy, specifically on the AI compute side, we're able to deploy capital in such a way that we're getting less than a one-year payback. It's almost like your capital is becoming almost an item that you would put otherwise in COGS because it's moving so quickly as far as monetization and moving into revenue. That's different than, I would say, historically on the SpaceX level, where we're building assets that are going to benefit us from launch sites for the next couple of decades. And so we certainly manage for optimizing ROI on our capital all the time from that perspective.

是的，我来补充一点。正如 Bret 提到的，我们预计今年 12 月将达到 1000 亿以上的年化经常性收入。可能还值得一提的是，我们内部对达到一万亿美元收入（不是 ARR，而是收入）的预测已从 2031 年提前到 2030 年。所以，在 IPO 之前，我们的财务预测是在 2031 年达到一万亿美元的收入。我们现在预计将在 2030 年实现。而且有非零概率是在 2029 年。

Yeah, I'll add a little bit to that. As Bret mentioned, we are expecting to reach 100 billion plus ARR in December of this year. It's probably also worth mentioning that our internal projections for reaching a trillion dollars in revenue, not ARR, but revenue, have moved up from 2031 to 2030. So, prior to the IPO, the financial projections we had were reaching a trillion dollars in revenue in 2031. We now expect that to be in 2030. And there's a non-zero chance of that being in 2029.

下一个问题来自美国银行的 Ronald Epstein。您的线路已接通，请讲。

Your next question comes from the line of Ronald Epstein with Bank of America. Your line is open. Please go ahead.

是的。嘿，嘿，大家晚上好。您能谈谈您在星舰第 13 次发射中学到了什么吗？哪些比预期好？哪些更糟？然后我们如何看待这在实现快速可重复使用（如隔热罩等）的旅程中的意义？

Yeah. Hey, hey, good evening everybody. Could you walk through what you learned on Starship launch 13? What was better than expected? What was worse? And then how do we think about this in the journey to kind of rapid reusability, the heat shields, so on and so forth.

当然。所以，实际上，第 13 次飞行，我有点担心那次飞行，因为你会想到阿波罗 13 号，那次飞行出了问题。但这次飞行实际上进行得非常顺利。隔热罩看起来非常坚固。飞船仍然漂浮在海洋中，我们预计将回收飞船并带回进行分析。所以我会说情况看起来非常好。这就是为什么，假设我们获得监管部门的批准，我们将在下一次飞行中尝试用塔架捕获飞船，暂定于本月底。我们预计飞行频率将迅速增加。可能一年后，我们将每天至少进行一次飞行，甚至更多。

Sure. So, actually, flight 13, I was a little worried about that flight 13 because you think of Apollo 13, things went wrong on that flight. But this flight actually went incredibly well. The heat shield appears to be very robust. The ship is still floating in the ocean, and we expect to recover the ship and bring it back for analysis. So I would say things look very good. And that's why, assuming we receive regulatory approval to do so, we will attempt to catch the ship with the tower on the next flight, which is tentatively scheduled for the end of this month. We expect the cadence of flights to be increasing rapidly. And probably a year from now, we will be doing at least one flight a day, possibly more.

下一个问题来自摩根大通的 Doug Anmouth。您的线路已接通，请讲。

Your next question comes from the line of Doug Anmouth with JP Morgan. Your line is open. Please go ahead.

非常感谢您回答问题。

Thanks so much for taking the questions.

呃，埃隆，你能详细讲讲利用你的火箭和卫星专业知识来加速算力建设的关键解锁点吗？另外，非常感谢新增的 67 亿美元算力交易。你能再多谈谈你预计供需环境将如何持续，以及你是否能保持这类交易的溢价定价和可持续性吗？谢谢。

Um Elon, can you elaborate on the key unlocks from leveraging your rockets and satellite expertise just in accelerating the pace of compute build out and then um totally appreciate the $6.7 billion in incremental compute deals? Um can you talk more about just how you expect the uh supply demand environment to persist going forward and and whether you'll be able to keep this kind of premium pricing for these deals and sustainability. Thanks.

当然。嗯，俗话说得好，人们常说“这又不是火箭科学”。其实火箭科学正是我们的日常业务。嗯，火箭科学是形容极端技术难度的惯用语，这是有原因的。让我告诉你，火箭拼命想做什么——每次飞行，它们都拼命想把自己炸成碎片。嗯，而工程上的挣扎就是说服火箭不要把自己炸成碎片，而是真正把载荷送入轨道。嗯，但你会注意到，我们的一些竞争对手确实把火箭炸了。嗯，这并不反映他们无能，并不意味着他们不行。事实上，我认为我们的竞争对手都是非常聪明的人。但尽管非常聪明，火箭还是爆炸了。嗯，所以如果你把用于制造巨型火箭定期可靠入轨的一小部分工程技能应用到地面数据中心上，这简直就像纽约洋基队去打小联盟球队，坦率说有点荒谬。嗯，地面数据中心与制造巨型可重复使用且频繁发射的火箭相比，简直是小菜一碟。嗯，而且星链卫星也是外星级别的技术。嗯，仍然没有什么能比得上我们的星链 V2 卫星，而我们的 V3 卫星比 V2 卫星真的强一个数量级以上。所以，你把那项工程的一小部分应用到地面数据中心，就能得到惊人的结果。嗯，这就是我们看到的。嗯，是的。

Sure. Um well you know as the saying goes you know long they say like they would say like this ain't rocket science. Well actually rocket science is literally our daily business. Um and uh you know rocket science is a idiomatic expression for extreme uh technological difficulty and and there's a reason for it because the the thing let me tell you what what rockets desperately want to do uh every flight they desperately want to blow themselves into tiny pieces. Um and uh then the engineering struggle is to u convince the rocket not to blow itself into tiny pieces and actually deliver payload to orbit. Um but you'll note that you know some of our competitors do in fact blow their rockets up. Um this is not a reflection on their uh does not mean they're they're incapable. In fact, I think our competitors are very smart people. But despite being very smart, the rockets still blow up. Um so now if you apply a small amount of the uh engineering skill that goes into making gigantic rockets uh reach orbit on a regular basis uh reliably to terrestrial data centers. This is like, frankly, the Yankees, New York Yankees going in and playing a little league team. And it's it's it's kind of ridiculous, frankly. Um this like the terrestrial data centers are a trivial problem compared to making gigantic reusable rockets and which are launched frequently. Um and the stalling satellites also are uh like alien level technology. Um there there's still nothing that compares to our Starlink V2 satellites and our V3 satellites are really more than an order of magnitude better than our V2 satellites. So So you take a small percentage of that engineering applied to terrestrial data centers and you get an amazing outcome. Um so that's what we're seeing. Um yeah.

嗯，从变现的角度来看，如果问我每瓦特变现的最佳猜测，比如对 Reuben 来说，我的猜测是大概在 30 到 50 美元之间。嗯，这只是猜测，我就随便猜猜。嗯，但我们显然看到的是，AI 的每瓦特智能正在快速提升。嗯，这意味着每瓦特的实用性也在快速提升。嗯，我认为任何密切关注 AI 领域的人都会开始看到越来越小的模型做出越来越惊人的事情。嗯，有一种观点认为，我们可能会达到即使 H100 也能提供与人类工程师相当的有用 AI 的程度。嗯，基于当前趋势，我认为这不是不可能的结果。那么这会有多有用呢？我认为那将极其有用。嗯，我的意思是，想想 12 个月前的 AI 在哪里？12 个月前是去年 8 月，那是在 Claude 4.5 出现之前。我认为 Claude 4.5 应该被视为一个里程碑，归功于 Anthropic。我要说，4.5 的优秀程度对系统来说是一个冲击。嗯，那基本上是 10 个月前的事。嗯，如果你说两年前的 AI 是什么样，我觉得两年前的 AI 感觉就像应该放在博物馆里的东西，你会说：“你不会相信，但两年前，这就是 AI。”嗯，这个遗物。嗯，所以我认为，到明年年底，基于目前的改进速度，我不确定是否还有数字领域至少是 AI 无法做到的事情。

Um now from a monetization standpoint um the if you say like what's my best guess for monetization per watt um of say Reuben's uh my guess is uh would be somewhere between 30 and $50. Um this is just a guess. I'm just you know take a guess as a guess. Um but but what we're obviously what we're seeing with AI is that the intelligence per what is increasing rapidly. Um so that means the usefulness per what is increasing rapidly. Um and I think anyone who's a close follower of the AI arena um you you you start seeing smaller and smaller models do more and more amazing things. Um there's an argument that um we may get to the point where even an H100 uh is able to deliver useful AI that's comparable to um a human engineer. Um if you based on current trends I think that's that's not an impossible outcome. So then it's like how useful would that be? I think the usefulness that would be is going to be extremely high. Um I mean if you think of like where was AI 12 months ago? I mean this is 12 months ago would have been August last year. This is that would have been before the advent of claude 4.5 which I think is should be considered one of the the milestones and a credit to entropic of you know was it was I would say a shock to the system was how good 4.5 was um and uh that was you know 10 months ago basically um and and if you say what about an AI two years ago I think well an AI two years ago feels like something that should be in a museum okay where you're like, you're not going to believe this, but two years ago, this was AI. Um, this relic. Um, so I I think, you know, by the end of next year, it's not clear to me that that there's anything that I that digital at least that AI won't be able to do um based on the current rate of improvement.

嗯，再看看逻辑和内存的生产速度。嗯，必须始终考虑这里的限制因素。目前的限制因素是内存。内存产出每年增长约 20%。现在，对于任何大型成熟行业来说，这通常都是极快且惊人的。但问问自己，需求每年增长 20% 吗？不，需求每年增长 200%，甚至更高。所以，如果需求增长远快于供应，那么经济学入门知识会告诉你，价格会上涨，而不是下降。

Um then look at the rate at which uh logic and memory is being produced. Um and one must always consider the limiting factor here. Limiting factor currently is memory. The memory memory output is increasing um by around 20% per year. Now, normally that would be uh fantastically fast uh and and and amazing uh for any large mature industry. But but ask yourself, is the demand increasing by 20% a year? No, the demand is increasing by 200% a year, maybe higher. So um so if you've got demand increasing much faster than supply then economics 101 would suggest that the the price increases. It does not decrease.

下一个问题来自德意志银行的 Edison U。您的线路已接通，请讲。

Your next question comes from the line of Edison U with Deutsche Bank. Your line is open. Please go ahead.

嘿，晚上好。感谢你回答我们的问题。想回到关于 1000 亿美元年化收入（AR）的评论上。你能稍微拆解一下吗？也许谈谈要达到这个目标需要什么？然后与此相关，埃隆，我知道你提到你把收入目标提前了一年。你能指出几件让你更有信心的事情吗？谢谢。

Hey, good evening. Thank you for for taking our question. wanted to to come back on the the comment about the 100 billion in AR. Is there any way you can can unpack that a little bit and and perhaps kind of what it takes to to get there? And then in relation to that, Elon, I know you mentioned you you pulled the the revenue target forward by a year. Can you maybe call out maybe a few things that that gave you this this extra confidence? Thank you.

关于拆解，我认为关键在于三个业务部门都在增长，但最大的部分是我们已经谈到的额外业务。我们在 7 月的前几周已经在 AI 云服务方面完成了交易。当然，如你所知，我们已经宣布了与 Google 和 Anthropic 的先前交易，这些交易将在本季度晚些时候或 10 月开始提升，所以业务方面有很多动力，同时我们也看到自家模型和 Cursor 的加入。所以确实是三个部门在推动，但云服务部分贡献最大。

Well, on the unpacking, I think that the key there is uh it's really growth in all three of the business segments, but the biggest piece is and we already talked about uh you know, additional business. We've already closed in the the first couple of weeks of of July on the uh the AI uh cloud services side. But certainly um as you know we we've already announced some previous deals with Google and Anthropic that start to ramp um either later this quarter or uh you know in October and so there's a lot of momentum from that side of the business as well as we're seeing uh you know strength in our own models and um and cursor uh coming into the fold as well. So it's really all all all three segments that are that are driving that but the uh cloud services piece being the largest contributor of that piece.

呃，是的，要澄清一下，12 月的 1000 亿美元年化收入不是问号，那是我们基本上什么都不做就能达到的数字。所以，嗯，你知道，我认为可能会更高，很可能更高。嗯，关于收入构成的相对百分比，我认为可能与我们 IPO 前分享的类似。嗯，只是提前了一年发生。嗯，你知道，我会回到我之前说的，以及 Gwynne 和 Bret 早些时候提到的，我们确实预计来自星链通信的 V3 卫星收入将快速增长。因为就像我说的，它的能力是 V2 卫星的 10 倍以上，我们将发射 10 倍的数量，而且我们还有直连手机星座正在部署。

Uh yes like to be clear the the 100red billion AR in December is not a question mark that's that's what we would achieve if we basically did nothing. So um like you know so it I think it may be higher than that. It probably will be higher than that. Um the uh you know I think in terms of the rel relative percentage of you know breakdown of like what revenue generates what um I think the the breakdown's probably similar to what we we shared preIPO. Um it's just happening a year a year sooner. Um and um you know I will go back to what I said and and what Gwynne and Bret mentioned earlier which is that we we do expect a very rapid increase in uh revenue from Starlink communications from the V3 satellite. um because like like I said, it's more than 10 times the capability of the V2 satellite and we'll be launching 10 times as many and and we have the direct to cell constellation going up as well.

下一个问题来自 UBS 的 John Hodelik。请讲。

Your next question comes from the line of John Hodelik with UBS. Your line is open. Please go ahead.

好的，谢谢。两件事。首先，Elon，接着刚才关于 Starlink Mobile 的话题。您预计什么时候开始发射 V2 卫星？能谈谈在 Echoar 交易后，您对额外频谱的需求吗？还有，您认为移动业务的机会有多大？如果能和宽带业务做个比较，那就太好了。其次，快速跟进一下算力问题。看起来 2027 年算力会有大幅提升。我们应该如何看待这部分算力是用于租赁还是用于训练 Grok？换句话说，有多少必须留作训练 Grok，有多少可以租给第三方？谢谢。

Great. Thank you. Uh, two things. Uh, first, Elon, just following up on that comment on Starlink Mobile. Uh, when do you expect to start launching the V2s? Um, can you talk about your appetite for additional spectrum after the Echoar deal? And then, um, just how big do you think that the opportunity is in mobile and if you could compare that to the broadband business, that would be great. And then secondly, just a quick follow up on the compute. Looks like you're going to have a big ramp in compute coming in 2027. Um, how should we think of that capacity being leased versus used for training Grok? I guess in other words, how much of that has to be withheld to train Grok versus that can be leased out to third parties? Thanks.

我来回答 Starlink Mobile 的问题，John。我们明年将开始发射下一代 Starlink Mobile 卫星——说起来有点绕，它其实是第二代，而 V3 是下一代宽带卫星。明年开始发射。需要说明的是，目前我们运营的系统大约使用 5 MHz 带宽，通过本地电信运营商来运营。通过 Echoar 频谱，我们可用带宽达到 65 MHz，能力大幅提升。换个角度看，我们可能还会把卫星数量增加 10 倍。简单来说，结合下一代卫星和 EchoStar 频谱，Starlink Mobile 的能力将是现在的 100 倍——10 乘 10 的提升。而且现在的服务已经相当不错了，可以通过 Signal 和 WhatsApp 进行语音和视频通话。所以我们预计这项能力会获得非常强劲的采用。至于收入，我做一个非常粗略的顶层估算。美国三大运营商——AT&T、Verizon 和 T-Mobile——合计年收入约 6000 亿美元。我预计我们能从他们那里获取相当多的客户，因为我认为我们的服务会更好。借助轨道卫星，我们将消除信号盲区。在自然灾害期间也会表现更好，因为令人惊讶的是，尽管太空电影把太空描绘得非常危险，但实际上太空是一个相当平静的环境。所以我对 Starlink Mobile 的未来非常兴奋。我们明年开始发射卫星，后年底开始提供服务。

I'll go ahead and take the Starlink mobile questions, John. Um, we're going to start flying the next generation Starlink mobile satellite—confusingly, it's actually the second version, as opposed to the V3, which is the broadband next generation broadband satellite. We're going to start flying those next year. Um, and just so you understand, currently the system that we're operating uses about five megahertz of bandwidth. Um, and we operate that through local telco providers. We have 65 megahertz of bandwidth available to us through the Echoar spectrum, which gives a massive increase in capability. Another way to look at it is we will also probably 10x the number of satellites. So, simplifying, you could look at the Starlink Mobile leveraging this next generation satellite and the EchoStar spectrum as being a hundred times better—10 times 10 better—than what we've got right now. And it's a pretty decent service. You can do voice and video calls over Signal and WhatsApp. Um, so we anticipate really strong uptake of that capability. As far as revenue, I'm just going to take a very high level, top-down view. The big three in the United States—AT&T, Verizon, and T-Mobile—roughly between them, $600 billion a year. Um, and I anticipate us to be able to acquire quite a few of their customers because I think our service will be better. We will eliminate dead zones by leveraging the satellites in orbit. It'll be better during any sort of natural disaster because, surprisingly, even though space movies make space look super dangerous, it's a pretty quiescent environment. Um, and so I'm quite excited about Starlink Mobile going forward. We will start to fly the satellites next year. Um, and we will start providing service end of next year.

我想我应该详细说明一下——显然这里不止一个问题。如果有人想建立财务模型，有些问题你得填入一些数字。对于 Grok 训练，我们确实预计内部用于 Grok 训练的算力占比会随时间下降，而更多的算力将用于推理，或者租给他人用于训练和推理。所以，我预计大约 10% 的算力会用于 Grok 训练，大概这样。

I guess I should elaborate on—we obviously have more than one question being asked here. Um, obviously if somebody's trying to create a financial model, some of these questions you have to plug in some number here. So for Grok training, we do expect the percentage of compute that we use internally for Grok training to decrease over time and to have far more of the compute used for inference, or to rent training and inference compute to others. Um, so I would expect maybe 10% of our compute to be used for Grok training, something like that.

下一个问题来自 Evercore ISI 的 Catgun Marl。请讲。

Your next question comes from the line of Catgun Marl with Evercore ISI. Your line is open. Please go ahead.

下午好，感谢您回答问题。我想跟进一下 Starlink Mobile，特别是您在美国市场的雄心。我想我们很多人都想更好地了解美国的直接面向消费者的服务会是什么样子。要成为真正的美国第四大运营商，意味着长期需要超过 1000 亿美元的移动资本支出，其中大部分是拍卖频谱和新建基础设施。那么，您如何权衡这种建设计划与 MVNO 或收购路径？您内部的基本假设是，您会把这一雄心推进到什么程度？还是说现在谈这个还为时过早？谢谢。

Good afternoon and thanks for taking the question. I had a follow up on Starlink Mobile and your ambitions in the US market in particular. I think a lot of us are trying to better understand what a US direct-to-consumer service could look like. You know, to become a true fourth US carrier would imply well over $100 billion of mobile capex over time, much of it auction spectrum and greenfield infrastructure. So, how are you weighing that potential build plan against an MVNO or acquisition path? And what's your internal base case on how far you'll take this ambition, or is this all a little bit too premature at this point? Thank you.

我们从 EchoStar 购买的频谱确实包含地面部分。所以我们肯定打算建设地面部分。这样你不仅拥有卫星本身的容量，还会有地面基础设施的建设——基本上是提供真正移动服务所需的硬件和系统，完全符合你的期望。关于资本支出，我不打算讨论。嗯，有点——我不想说不成熟，我只是说现在不谈这个。但让我给点提示：你可以把基站——蜂窝基站——放在 Starlink 宽带天线的支架上。这样你可以在全国部署这些小型飞蜂窝，按需部署。你不一定需要在部署系统之前一次性投入 100 亿或数十亿美元的低频段频谱。所以这就是为什么我不想谈地面建设的资本支出，因为我们有很多很棒的新想法来实现这一点，而且我认为这会非常节省资本支出。

Uh, the spectrum that we purchased from EchoStar does have terrestrial components. So we definitely intend to build out the terrestrial component. So you will have not only the capacity from the satellites themselves, but you will have a buildout of the terrestrial—basically the hardware and systems necessary to make a true mobile service exactly what you want it to be. Um, I'm not going to talk about the capex for that. Um, it's a little—well, I don't want to say it's immature. I'm just going to say I'm not going to talk about the capex for that right now. Um, but let me give a little hint, and that is you could put a base station—a cellular base station—on the gear that holds a Starlink broadband dish. So you can have these little femtocells around the country, and you deploy that as you need it. You don't have to necessarily spend 10 or many billions of dollars in low-band spectrum all upfront before you deploy your system. Um, so that's why I don't want to talk about the capex for building out the terrestrial, because we have a lot of really great and new ideas for how we're going to do that, and I think it will be quite capex efficient.

是的，这一点值得多说几句，因为这是一个非常重要的点——与其部署那些昂贵且难以选址的大型蜂窝基站，我们相当有信心可以部署大量小型站点，这些站点本质上就是 Starlink 天线，同时提供移动频谱段的连接，并且遍布各处。Starlink 天线位于房屋和企业的屋顶上，因此它们有清晰的视野，可以直接为地面上的手机提供连接。事实上，我们提供的连接可能比现有蜂窝运营商提供的更好、带宽更高。这是一个重要的点。

Yeah, that point is probably worth a few more words, which is that—because this is quite an important point—instead of having to deploy these very expensive and difficult-to-locate large cellular base stations, we feel reasonably confident that we can deploy a large number of small stations that are essentially just Starlink dishes that also provide connectivity in the mobile spectrum bands, and have them be all over the place. And the Starlink antennas are located on the roofs of houses and businesses, so they have a clear view to provide connectivity directly to cell phones on the ground. And in fact, providing connectivity that we think is probably better and higher bandwidth than what is currently available from cellular providers. This is an important point.

谢谢。在剩余时间里，我们想回答一些个人股东的问题。我会大声念出来。我们收到了很多关于 Starship 载人着陆系统进展的问题。能提供一些 Starship 的最新情况吗？

Thank you. With the remaining time, we would like to cover some of the questions from individual shareholders. I will read them aloud. We got a lot of questions on the progress for Starship human landing system. Can we provide any updates for Starship?

嗯，显然，在载人之前，我们需要让 Starship 在卫星任务上达到超高可靠性。

Well, obviously we need to make Starship super reliable for satellites before we put any people on board.

但我们预计会实现非常高的发射频率，从而很快达到人类所需的安全水平，可能到明年年底。

But we are expecting to have a very high launch rate and thus achieve a level of safety needed for humans very quickly, probably by the end of next year.

就里程碑而言，在轨推进剂转移对我们 SpaceX 内部以及 HLS 的目标都至关重要。Artemis 3 任务在明年，届时我们将与猎户座飞船对接。随后我们将进行一次无人驾驶的月球直接货运任务。然后我们希望在 2028 年将靴子踏上月球。

As far as the milestones go, propellant transfer in orbit is critical to both our internal SpaceX as well as our HLS ambitions. Artemis 3 mission is next year. That's where we will dock with the Orion spacecraft. We will follow that up with a direct to lunar cargo mission uncrewed. And then we want to put boots on the ground, boots on the moon in 2028.

关于收购 Cursor 一事，有几个问题问到合并后的产品路线图。我想我们会把这个问题留到财报电话会议之外，可能会发布公告或进行讨论。但我们不想在监管机构批准收购之前操之过急。所以我们正在努力尽快完成收购。我认为我们离目标很近了，但我们必须警惕在监管批准上操之过急。

For the Cursor acquisition, there were several questions asking about the combined product roadmap for the companies once it closes. I think we'll leave that to, we'll probably make some announcement or discussion of that outside of the earnings call. But we don't want to jump the gun on regulators for the closing of the acquisition. So we're trying to get the acquisition closed as quickly as possible. I think we're pretty close to that, but we have to be wary of jumping the gun on regulatory closures.

下一个问题来自个人股东。一旦你们开始发射宽带 V3 卫星，它们能多快转化为 Starlink 服务的改善？

Next question from the individual shareholders. Once you start launching broadband V3 satellites, how quickly will they translate into improved Starlink service?

嗯，我们需要一定数量的 V3 卫星，可能至少需要大约一千颗。但大概，你知道，大约明年第二季度，我想我们会达到那个点。

Well, we would need some critical mass of V3 satellites, probably at least on the order of about a thousand. But probably, you know, approximately second quarter next year would be, I think, where we get to that point.

我们最后一个问题是，隔热罩是星舰最难解决的问题吗？

And our last question is the heat shield the hardest remaining problem for Starship.

听着，我不想说大话或什么的，但我认为目前隔热罩问题已经解决了。现在我们想仔细看看目前漂浮在海上的那艘飞船。但所有数据和目视检查的迹象都表明我们已经解决了隔热罩问题，这可以说是最大的一个问题。这并不意味着我们不会对隔热罩进行改进。当然，我们会继续改进，但我想说，目前我们看不到任何技术障碍阻碍实现完全和快速的重复使用。

Look, I don't want to jinx it or anything, but I think I'd consider the heat shield problem solved at this point. Now we want to take a close look at the ship that is currently floating in the ocean. But all indications from data and visual inspection suggest that we have solved the heat shield problem, which is arguably the single biggest problem. That doesn't mean we won't make improvements to the heat shield. Of course, we'll continue to make improvements, but I would say that we do not see any technical obstacles at this point to achieving full and rapid reusability.

谢谢。今天的电话会议到此结束。您可以通过 ir.spacex.com 联系我们，我们下个季度再见。干杯。

Thank you. This concludes today's call. You can reach us at ir.spacex.com and we will see you next quarter. Cheers.

今天的电话会议到此结束。感谢您的参与。您现在可以挂断了。

This concludes today's call. Thank you for attending. You may now disconnect.