Rivian CEO Says Subsidized Chinese Capital Risks Western EV Failure

Rivian CEO称受补贴中国资本或致西方电动车面临失败 传统汽车制造商若不迅速采用集成式区域软件架构以抗衡中国结构性更低的生产成本，将在2030年代失去市场份额。 在本次讨论中，Rivian创始人RJ Scaringe深入剖析了创办独立汽车初创公司所面临的巨大资金与架构挑战，重点阐述了公司早期与Amazon的合作，以及近期通过整合Gen 2平台实现正毛利率的成果。他介绍了Rivian的高度垂直整合战略、即将推出的大众市场R2 SUV平台，以及包含高利润技术授权和DTC直营生态系统的多支柱商业模式。此外，Scaringe还就自动驾驶中的端到端AI、第四代“Maximus”电机背后的工程设计，以及在全球范围内构成威胁的中国电动汽车复合结构性成本优势分享了技术见解。 传统老牌车企依赖来自不同供应商的50至150个碎片化、孤立的计算机，而Tesla和Rivian则采用集中式架构，支持全车系统协同通信。 (0:00) 创办一家独立汽车公司需要在产生任何初始收入之前，将数十亿美元资金、数千名工程师和庞大生产设施几乎同时协同到位。 (0:00) Amazon在2019年的合作及7亿美元初始融资加速了Rivian的可行性，这主要得益于商用配送车队消除发动机怠速带来的运营与经济优势。 (2:44) 在疫情及随后的2022年半导体危机期间同时推出三款高度复杂的车型（R1T、R1S和Amazon EDV），极大地推高了Rivian的早期运营成本。 (5:51) Rivian通过将其电子电气架构彻底重构成“Gen 2”，把17个离散计算机整合为仅3个区域控制器，在2024年末和2025年初实现了正毛利率。 (12:02)

Rivian founder RJ Scaringe contends that building a viable modern automaker requires abandoning legacy component outsourcing in favor of vertically integrated, software-defined zonal architectures. To sustain profitability, EV manufacturers must pair proprietary hardware-software stacks with aggressive cost reductions and high-volume, mass-market vehicle platforms. Consolidating Electronics into Zonal Architecture (46:35) Traditional automotive manufacturing relies on 50 to 150 separate electronic control units (ECUs) purchased from Tier 1 suppliers, creating isolated software silos that prevent centralized vehicle control. Scaringe characterizes this traditional supplier model as a "field of weeds" with too many abstraction layers between the manufacturer and the code, preventing cohesive vehicle updates. To solve this, Rivian overhauled its vehicle platform for its Gen 2 architecture, consolidating 17 domain computers down to three zonal controllers—East, West, and South—that govern all propulsion, body control, and lighting. This consolidated architecture allows Rivian's engineering teams to deploy fleet-wide software releases within two weeks, in contrast to the rigid, multi-year requirement cycles typical of legacy automakers. The software-defined approach also forms the basis of Rivian's B2B licensing business, exemplified by a $5.8 billion joint venture agreement to license its zonal architecture and software stack to the Volkswagen Group across brands and price tiers. Scaling Production from R1 to Mass-Market R2 (15:40) Launching an automotive brand demands billions of dollars in capital, thousands of engineers, and massive production facilities years before generating initial revenue. Rivian faced acute operational strain in late 2021 by launching three complex vehicles simultaneously—the R1T pickup, R1S SUV, and Amazon commercial delivery van (EDV)—while navigating the 2022 semiconductor shortage. To keep assembly lines running, the company absorbed heavy supplier price premiums, sometimes paying double for critical parts. To establish a sustainable cost structure, Rivian spent 2023 resourcing its bill of materials and redesigning high-voltage components, such as its fourth-generation "Maximus" electric motor, which uses continuous winding to eliminate 95% of stator welds. These revisions helped Rivian achieve positive gross margins in Q4 2024 and Q1 2025. While the company's premium $90,000 R1 platform generates roughly 40,000 sales annually, Rivian plans to unlock volume with the $45,000 R2 platform. Rivian is expanding its Normal, Illinois factory to 215,000 units of total annual capacity (including up to 175,000 R2s) and constructing a Georgia facility to add 400,000 units across two phases. Multimodal Autonomous Systems and Software Control (41:15) Rivian transitioned its autonomous driving platform from a Gen 1 rules-based trajectory planner to an end-to-end foundation model. This system processes sensory data via transformers and relies on a fleet training flywheel that flags discrepancies whenever a human driver's choices diverge from the model's predictions. Scaringe forecasts that "hands-free-eyes-off" autonomy will shift from a luxury feature to an industry expectation within five years. Contrasting his views with camera-only autonomous architectures, Scaringe argues that multimodal sensing is mathematically superior because all sensors have noise. While cameras struggle in poor weather, imaging radar now costs roughly $100 to $125, and LiDAR costs have plummeted from $75,000 fifteen years ago to approximately $200. Rivian applies a similar end-to-end philosophy to its cabin software, deliberately excluding Apple CarPlay to prevent a disconnected interface layer from blocking native vehicle control, while integrating Google Maps to resolve user navigation demands. Chinese Cost Asymmetry and the EV Adoption Gap (61:37) Addressing the global rise of Chinese electric vehicles, Scaringe disputes the idea that Chinese automakers possess proprietary mechanical breakthroughs, noting that competitor vehicle teardowns reveal standard design and consolidation practices. Instead, Scaringe attributes their pricing power to a compounding economic advantage: government-subsidized, near-zero cost of capital combined with lower labor costs across every supplier tier down to raw materials. He states this cost structure cannot be replicated under Western capital conditions, prompting protective tariffs in the United States. Scaringe also rejects claims that slow U.S. electric vehicle adoption—where EVs represent roughly 8% of sales and Tesla holds about 50% market share—is caused by consumer reluctance. Instead, he points to a lack of compelling, affordable choices under $50,000, contrasting this with China's market, where varied choices drive adoption rates 3.5 to 4 times higher. Rivian intends to address this gap with the R2 while utilizing its direct-to-consumer sales model to capture profitable servicing and used-vehicle revenue across a 15-year lifecycle. While Rivian has demonstrated technical validation through its Volkswagen partnership and achieved positive gross margins, its broader viability depends on executing the mass-market R2 ramp. Competing successfully will require managing high Western capital costs against low-cost foreign producers while navigating potential demand volatility from shifting EV tax policies.

Rivian 创始人兼首席执行官 RJ Scaringe 认为，新造车企业的生存核心在于通过高度垂直整合与集中式区域架构打破传统供应链的割裂；公司正以高端 R1 平台的工程积淀转向面向大众市场的 R2 平台，并依托技术对外授权与直营生态重塑汽车产业的盈利模型。 区域电子架构与软硬件全栈自研 (46:35) Scaringe 指出，除 Tesla 与 Rivian 外，目前绝大多数传统车辆仍依赖 50 至 150 个由不同一级供应商提供的独立电脑（ECU），导致车内各系统无法互通，甚至无法从控制台统一调节后视镜。为彻底解决这一碎片化架构，Rivian 在其第二代（Gen 2）平台上将原有的 17 个控制器精简为东、西、南 3 个区域控制器，全面接管动力传动、车身控制与照明等整车功能。这种集成架构使工程团队能在两周内完成新功能的编写、测试与 OTA 推送。 这一架构能力促成了 Rivian 与大众汽车集团达成一项价值 58 亿美元的合作协议，Rivian 将向后者授权其区域架构与软件平台。Scaringe 预测，到 2030 年代初，不具备软件定义架构的主机厂将难以在市场立足。此外，Rivian 拒绝接入 Apple CarPlay，理由是第三方手机映射层无法深度联动车况数据；公司选择原生整合 Google 地图数据，以保证在 AI 语音交互与路线能耗规划上的系统一致性。 从高端制造迈向大众规模化降本 (15:40) Rivian 早期发展受益于亚马逊的战略投资与商用车订单。亚马逊在 Rivian 上市前累计注资约 20 亿美元（始于 2019 年的 7 亿美元融资轮），目前 Rivian 每年生产约 10,000 辆定制电动货车，并已开始向其他商业客户开放销售。尽管在疫情与 2022 年半导体供应链危机期间同时推进三款车型的量产导致物料成本剧增，但 R1 平台仍在 7 万美元以上的高端电动 SUV 领域占据美国约 35% 的市场份额，平均售价超过 9 万美元，年交付量约 40,000 辆。 通过 Gen 2 的电子架构重构、供应链合同重组以及采用第四代“Maximus”连续绕组电机（减少电机约 95% 的焊点），Rivian 在 2024 年第四季度与 2025 年第一季度实现了正向毛利率。以此为基础，Rivian 计划推出起售价 4.5 万美元的紧凑型 SUV“R2”，将其伊利诺伊州工厂年产能扩充至 21.5 万辆（含多达 17.5 万辆 R2），并正在佐治亚州建设规划两期共 40 万辆产能的新生产基地，以此切入平均售价 4.9 万美元的美国主流乘用车市场。 重构传统车企的三重商业支柱 (53:20) 传统汽车工业长期处于重资本、低利润的运行模式，Scaringe 阐述了 Rivian 不同的三大商业支柱：首先是整车制造与销售规模，预计未来以 5 万美元均价销售 20 万辆 R2 即可贡献 100 亿美元营收；其次是高毛利率的企业级技术授权，以大众汽车的交易为代表，该业务能有效摊薄早期的巨额研发开销；第三则是绕过传统经销商体系的直营生态（D2C）。 Scaringe 指出，传统汽车经销商的核心利润来源于售后维修。Rivian 承担了前期构建服务网络的高昂固定成本，但随着保有量扩大，车队售后将转变为强劲的盈利来源。结合深度车辆诊断与软件更新能力，软件定义汽车能够在二手车流转中维持更高的残值。据 Scaringe 测算，一辆汽车在 15 年全生命周期内产生的总收入通常是其初始购买价格的 2 到 2.5 倍，直营模式确保了这部分持续性收益全部留在主机厂内部。 端到端大模型与多传感器融合路线 (37:56) 在自动驾驶技术路径上，Scaringe 介绍了系统从 Gen 1 到 Gen 2 的演进。传统的 Gen 1 依赖规则驱动，由感知模块识别目标并赋予速度矢量后交由手写逻辑规划路径；而 Gen 2 转向端到端的基础大模型，利用 Transformer 架构直接处理高分辨率图像与传感数据。当人类驾驶员的操作与后台运行的模型预测产生偏差时，系统会自动标记并回传数据，依托规模车队形成自我迭代的数据飞轮。Scaringe 预测，未来五年内“脱手且脱眼”（hands-free-eyes-off）的高阶智驾将由小众配置转为行业基准。 针对行业内的传感器争论，Scaringe 从数学理论与噪声控制角度明确支持多模态融合，反对纯视觉路线。他指出，随着近十年供应链演进，激光雷达成本已由数万美元降至约 200 美元，高规格成像毫米波雷达降至 100 至 125 美元。在强光、雨雪和浓雾等恶劣工况下，雷达能够弥补摄像头的物理缺陷，多传感器的交叉输入还能有效加速纯视觉感知算法的自监督训练。 全球成本结构分歧与市场选择困境 (61:37) 对于中国电动汽车的全球崛起，Scaringe 认为其优势并非源于不可告人的机械工艺或秘密设计，车企互相拆解产品早已是行业惯例。中国车企的核心成本壁垒在于从原材料到一、二、三级供应商层面，均享有由政府补贴形成的极低资本成本，叠加相对较低的人工成本，形成了在欧美资本环境下无法直接复制的复合优势。而在技术层面，领先的中国品牌同样从零起步，采用了集中式区域架构和垂直整合模式，从而能以更低成本容纳丰富的功能配置。 针对美国电动汽车仅约 8% 的普及率以及特斯拉独占半数份额的现状，Scaringe 辩称这并非消费者拒绝电动化，而是除特斯拉外市场极度缺乏 5 万美元以下具备竞争力的产品；相比之下，中国消费者在各个价位段拥有充沛的高品质选择，因此渗透率可达美国的 3.5 至 4 倍。他认为，虽然短期内补贴退坡等政策调整会减缓欧美传统车企的转型节奏，但这恰恰为能够提供高性价比优秀车型的独立电动车企留出了市场空间。 Scaringe 总结强调，汽车行业向软件定义与区域集成架构的演进已不可逆转，这是车企在 2030 年代全球竞争中维系存续的基本门槛。虽然跨大西洋安全法规壁垒和地缘政策给全球化带来阻碍，但消除大众市场“供给匮乏”并建立起全生命周期盈利模式，仍是检验造车新势力能否实现可持续商业闭环的关键测试。

[18:38] RJ Scaringe: The comical thing is you can pull up to a stoplight, you look to your left and you see a brand new Ferrari and you're actually quicker than that car. And you've got groceries in the front trunk, your kid's in a car seat and silently it just out accelerates any of those things, and then it can veer off the road and drive over a mountain. [27:38] RJ Scaringe: Trying to do that through a homogenized single layer app that works across many, many brands and by definition in some ways has to play to a lowest common denominator is really challenging. [31:11] RJ Scaringe: If we have a certain pool of dollars we can spend on the car [34:00] RJ Scaringe: a Cybertruck could not work in Europe [46:45] RJ Scaringe: inconceivable [46:45] RJ Scaringe: field of weeds [46:45] RJ Scaringe: car Costco [46:45] RJ Scaringe: the ultimate billboard [53:20] RJ Scaringe: afterburner [61:37] RJ Scaringe: On the mechanical side, there's nothing that's not seen by every car company [64:05] RJ Scaringe: I think every manufacturer should be thinking, 'How do we build a software-defined architecture as quickly as possible?' [65:59] RJ Scaringe: it's bad for the world, it's bad for my kids. [65:59] RJ Scaringe: No, it's because the EV is not that compelling.

[0:00] Cars used to have cool names. [0:01] You know, you had Mustang, Thunderbird. [0:03] Now we have R1, R2. [0:06] Yeah, so the code name for Gen 2 is Peregrine. [0:08] Okay, beautiful. Which is [0:09] a super fast falcon. [0:10] Would anyone here buy a Rivian Peregrine? [0:12] Yeah, I'm seeing a lot of takers. [0:13] With the exception of two companies, Rivian and Tesla, [0:16] every car on the road has anywhere [0:18] from 50 to maybe 150 little computers that run the car. [0:21] This is why you, in a normal car, [0:23] cannot adjust the, say the side mirrors, [0:26] from the center console. [0:27] It's two or three different companies. [0:28] In 2013, 2014, I would go meet with suppliers, [0:33] CEOs of very large suppliers and say, [0:35] "We're working on an electric SUV and electric truck [0:38] as a flagship product." [0:39] To say, "laughed out of the room" is probably accurate. [0:44] So you know how to pour these things and everything. [0:45] These are actually really easy. [0:46] This is good, this is a new... [0:48] No, you've got to go all the way. [0:49] All the way? It's perfectly sized to fit. [0:51] That is impressively sized. [0:52] RJ Scaringe is the founder and CEO of Rivian. [0:55] Makers of electric trucks and SUVs. [0:58] You've probably started to see Rivians [0:59] out and about on the road. [1:00] It's a delightful product, [1:01] and how they built an independent new car company [1:04] is a really interesting story. [1:05] Welcome, cheers. Cheers, yeah. [1:08] The Rivian story makes no sense to me. [1:12] It shouldn't have worked because you grew up [1:15] passionate about cars, nature enthusiast. [1:18] You did a mechanical engineering PhD [1:21] and Rivian was not only the first company you started, [1:25] it was your first job, or I mean your first career. [1:27] Yeah, yeah. [1:28] Like, hardware companies are startups on hard mode, [1:32] auto companies are hardware companies on hard mode. [1:35] So everything about this just seems [1:38] like it shouldn't have worked out. [1:40] Yeah, yeah, it's really hard. [1:41] Explain yourself. [1:43] Yeah, I mean one of the things [1:43] that makes starting a car company so difficult [1:45] is you need a lot of hard-to-obtain things [1:50] all at the same time. [1:51] So you need billions of dollars of capital, [1:54] you need a team of thousands of engineers. [1:56] You need hundreds of suppliers [1:58] that would want to work with you [1:59] and provide components and parts, [2:01] a manufacturing facility that's enormous [2:03] and takes years to build. [2:05] And if you had all those ingredients at the start, [2:07] it would take maybe four or five years [2:08] to launch the first product. [2:10] But in the beginning you have none of those ingredients, [2:11] and so it really is a challenging exercise. [2:14] And for me, I think the part [2:16] that was probably harder than I expected [2:17] was just the path to go from, we started with no capital, [2:21] so to go from no capital to raising [2:23] a very small amount of capital, [2:25] enough to prove out some semblance of an idea, [2:29] but then to start to raise [2:30] more and more capital to scale it. [2:31] But it was a low probability of success. [2:34] So what were the moments where, [2:35] if you're following the prediction market, [2:36] if you're looking on Polymarket, "Will Rivian succeed?" [2:40] What are the moments where the probability [2:41] would've spiked upwards at some point? [2:44] Probably 2019. [2:45] 2019 Amazon invested in the company, [2:47] so that would be a big spike in the probability. [2:49] Sure, how much did they invest? [2:50] Well, in total they invested several billion dollars [2:52] before we went public. [2:53] You know, today they're our largest shareholder, [2:55] but in addition to them, so they invested at, [2:58] I think around 2 billion prior to our IPO, roughly. [3:01] Because it invested in a few rounds. [3:03] But the round they invested in was a $700 million round. [3:05] It was our big first round of a number of strategic [3:10] and really well-known investors came in at that point. [3:13] And that was after we'd shown the product [3:15] at the LA Auto Show. [3:17] And prior to that we'd raised a few hundred million dollars, [3:19] but we still needed quite a bit more money [3:20] to launch the company. [3:21] But importantly, beyond just them investing, [3:23] we also put a commercial deal together [3:25] where we build commercial vans for them. [3:27] And so you now see them all over the place. [3:29] These are the big electric vans that we build [3:31] with Amazon as a lead customer, [3:32] but there's other non-Amazon customers. [3:33] Why did they care about electric vans? [3:35] Was it their climate commitment? [3:36] Was it something to do with operationally it works better? [3:39] What do they want? Well, there's a true [3:40] cost advantage to running electric vans for deliveries. [3:42] So when you think about a delivery van, [3:45] it spends a lot of its time sitting still. [3:47] And so not having an engine idle, [3:49] not having the same service frequency [3:52] and all the engine maintenance and repairs, oil changes, [3:55] brake replacements, everything just goes away. [3:57] So there's a real cost advantage to running EVs, [3:59] but there weren't any great electric van products [4:02] in the market and certainly nothing that was designed [4:05] with robustness or with their use case in mind. [4:08] Okay, so they thought for their logistics operation, [4:10] having a proprietary source of electric vehicles [4:12] would lead to a long-term operating cost advantage. [4:15] Yeah, I mean, it was wild. [4:17] From the early meetings with Amazon, [4:19] we sketched an idea on the whiteboard. [4:22] And that was in like 2019, early part of 2019, [4:26] we were sketching the ideas for the vehicle [4:29] and by end of 2019 we'd signed a deal [4:33] to supply commercial vans, [4:34] and then exactly two years later we launched the product. [4:38] So it was a extremely fast program [4:40] to go from whiteboard to vans delivering packages [4:44] within about two years. [4:45] Yeah, yeah, yeah. And where are those? [4:47] Where is production of those vans now, [4:49] in terms of volumes or kind of how you work [4:51] with Amazon on it? [4:52] It's about 10,000 or so a year. [4:53] It's going to be growing. [4:55] So we're bullish on the van market in the long term. [4:59] I think in the short term there's been [5:00] a lot of things we didn't plan for. [5:02] We had COVID, supply chain interruption. [5:04] We've seen a number of companies [5:06] just be much more capital constrained [5:07] in how they deployed assets [5:09] to invest in, let's say, electrification [5:11] or electrification infrastructure. [5:12] And so the whole commercial van space [5:14] has moved slower than we thought. [5:15] But the economics are so advantaged [5:17] that we're really clear on it long term being a big space. [5:20] Do you already sell those vans to others? [5:22] Will you sell those vans to others? [5:23] Yeah, we just started about a year ago selling to others. [5:27] Okay. But the ramp up has been slow. [5:30] When you think of large operators of vans, [5:34] very often these are not the companies [5:35] that are going to take the biggest risks on technology [5:37] or be on the front edge of early adoption. [5:40] So we think of it as they take smaller steps with pilots, [5:42] prove the model, prove that they can support [5:44] the charging infrastructure. [5:46] And then once they see the economic advantage, [5:48] then I think many will make the switch. [5:51] Yeah, going back to the Rivian story being a bit unlikely, [5:53] the other thing that seems to me [5:54] is getting reliable production going is really hard. [6:01] And again, it feels like the first cars [6:04] that roll off the line should be super janky [6:06] and unreliable and everything like that. [6:08] In preparation for this episode, [6:09] I've been stealing my wife's Rivian, [6:11] I've been really enjoying it. [6:12] And it is, you're complaining because [6:13] you're making fun of me [6:15] because it's a used version that we bought, [6:17] not a new version, so it's a '23 model, [6:19] which is one of the earliest ones, and it's awesome, [6:22] and it works great and the build quality is incredible [6:25] and things like that. [6:26] And so again, that's the other part to me [6:27] that feels unreliable of the story. [6:29] Just it feels like that's not how [6:32] hardware companies usually go. [6:33] So how did you guys get the quality [6:35] and reliability good out of the gate? [6:35] Yeah, I mean one of the things [6:37] that's so different about building a car company [6:41] versus most software, most technology companies, [6:43] and for that matter, even how venture capital, [6:45] how it's evolved and how how it takes place [6:47] is the first product of most startups, [6:50] like usually a tech or software-centric startup [6:53] is pretty loose, it's there to demonstrate [6:55] that the idea holds promise, [6:57] and it may not have scalability built into it, [7:02] but it shows there's a product market fit. [7:03] And then following that there could be [7:05] more capital invested to grow it, [7:07] to add robustness to the product. [7:09] So the minimum bet size for you is really big. [7:12] Yeah, so in a vehicle, your first dollar of revenue [7:14] takes many billions of dollars of development. [7:17] And the thing you launch needs to be incredibly robust. [7:20] And so that means multiple winters of testing. [7:23] Multiple hot summers where you're testing vehicles [7:25] in the desert, very large durability [7:29] and reliability team set up to run this. [7:32] And there's cultural challenges. [7:34] When you think about building a new company. [7:37] One of the strengths of a new company very often [7:39] and I think what allows startups [7:42] and entrepreneurs to excel [7:43] is they're willing to try new things as businesses, [7:47] it gets embedded into the culture. [7:49] There's a certain move fast and break things mindset. [7:53] And that's great for technology, [7:55] it's great for pushing boundaries, [7:56] but it's very different than the skills [7:58] and the cultural mindset you need [8:00] to run stable operations where let's say [8:04] if you're running a manufacturing plant, [8:05] you need repeatability, [8:07] you need a lot of process orientation. [8:10] And so just for us, [8:12] this step of going from a product development company [8:14] developing new technology, [8:16] making prototypes, which are wonderful. [8:19] And demonstrating that those prototypes are capable [8:21] and then transitioning that into a production system [8:24] that can scale and has high quality [8:27] and robustness built into it. [8:28] That took a lot of work, it was hard. [8:31] But the first version of many hardware products [8:34] have a lot of teething pains. [8:35] Like, did the R1 have a lot of teething pains? [8:38] Did you guys do something different to avoid that? [8:40] Well, so for us, we made a few, [8:44] if I could go back in time, [8:45] I probably would change one thing. [8:46] We launched a truck, an SUV, so the R1T and the R1S, [8:48] so it's a sibling-set of products. [8:51] At the same time we launched a commercial van. [8:52] So we launched three different products at the time [8:56] and we were building out our plants [8:58] starting in the beginning of 2020 and then launched in 2021. [9:02] And so like trying to build out an industrial ecosystem, [9:05] turn on a supply chain and then ramp a product is hard. [9:09] Doing that in the middle of COVID is really hard, [9:11] and doing that across three different product lines [9:13] was extremely hard. [9:15] And so I probably would've inserted more sequencing [9:18] between all those different products. [9:19] But nonetheless, when the R1 launched [9:22] we won vehicle of the year, which was awesome. [9:27] And so MotorTrend does this, a few different organizations. [9:29] We had multiple truck or SUV of the year awards [9:32] and then Consumer Reports does this annual brand survey, [9:35] and they look at customer satisfaction, customer appeal, [9:38] and they look at likelihood to repurchase. [9:40] And so in our first year of production [9:42] with our first product we've ever produced, [9:44] we came out by far number one on that, which was amazing. [9:48] And in our second year we came out by far number one. [9:51] So we beat the next closest by like 15 points. [9:55] So that was amazing. [9:56] It was really just a reflection [9:57] of how much work the teams had put in [10:00] to make all these product-level decisions, [10:02] trade offs around... [10:04] You think of a vehicle, it's many tens of millions [10:07] of trade offs you're making around cost [10:10] versus performance versus a feature. [10:13] And so we'd really done a nice job [10:15] of getting the product market fit right, [10:16] so really connected with consumers. [10:18] But saying all that, the ramp was hard. [10:20] So launching late '21, it was difficult [10:24] because we were coming off the back of COVID [10:27] and then we had this awful supply chain crisis that hit, [10:31] you may remember like 2022 or getting anything [10:34] from drywall to semiconductors was hard. [10:36] But in our case we were last in line [10:39] with many of the suppliers because we were new. [10:42] And so you can imagine we went [10:44] from having a bill of materials [10:46] that was already inflated [10:47] relative to existing large incumbents, [10:49] to something where if we just wanted supply, [10:52] they said, [10:53] "look, you're going to have to pay us an extra 20%." [10:54] And so it was like this across thousands of parts in a car, [10:58] all these discreet decisions [10:59] of do we stop production for this part that should cost $2, [11:04] but we're being asked to pay $4. [11:06] And you do that several thousand times and it adds up. [11:09] So we had a cost structure that was really challenging. [11:12] We chose to actually ramp with some of that cost. [11:15] And I assume you mostly just ate those cost increases [11:18] because it doesn't make sense [11:19] to spike the cost of the vehicle. [11:20] Did you debate whether you should pass along [11:22] the cost increases or eat them? [11:24] Well in parallel to this, [11:26] of course inflation went really wild in 2022 into 2023. [11:31] So the price of vehicles grew quite a bit [11:35] in that same timeframe. [11:36] And so we did add, in reaction to that, [11:40] we did have adjustments to our pricing, [11:42] but we've tried to be really stable in pricing. [11:44] because it's always hard [11:45] if your pricing's moving around a lot, [11:47] it wrecks havoc on the residual market. [11:49] Yeah, you can't plan for car purchase. [11:51] It's just really hard for consumers. [11:53] So we tried to be really methodical with that. [11:56] But saying all that, we had to sort of grow through '23, [12:00] we resourced a lot of our bill of materials. [12:02] We launched what we call Gen 2 of the vehicle, [12:05] which you don't have, but is a meaningful step forward [12:08] in terms of technology, but also a considerable reduction [12:12] in the overall cost of the vehicle. [12:13] And so finally end of '24, beginning of '25, [12:17] we started to hit positive gross margins. [12:19] So Q4 of '24, Q1 of '25 positive gross margin. [12:24] We're extremely bullish on long-term profitability, [12:27] but this business is really tough. [12:29] If you have disruptions to the supply, [12:31] just the fixed costs are really tough. [12:33] What's the Gen 2 pitch? [12:34] What's most exciting from a tech point of view? [12:36] Is it autonomy, is it something else? [12:38] Boy, yeah, there's a lot there. [12:41] We didn't change the exterior of the car much. [12:43] So really, we redid a lot of the interior, [12:46] like the guts of the car. [12:48] And the biggest change was on Gen 1 [12:50] we designed and built all the electronics in the car. [12:54] But we allowed for a number of the compute platforms [12:58] to be specific to an area, like specific to a domain. [13:02] Oh, this the zonal architecture where you move to... [13:04] Yeah, so the reason we did that at the start [13:06] is we needed to move fast. [13:07] So let's say you're on the body controls team, [13:10] your timeline isn't necessarily the same [13:12] as let's say the powertrain controls team [13:14] or the vehicle dynamics team. [13:15] And so we allowed the proliferation [13:18] of a few different domains [13:19] that we knew long term would consolidate. [13:21] But we made the decision, [13:22] because we said we own the whole software stack, [13:24] we own the electronics, [13:24] the consolidation will come in step two. [13:27] And then with Gen 2 we went from 17 down to three. [13:29] And so now we have a, what we call an East-zone controller, [13:32] a West-zone controller, and a South-zone controller. [13:34] When you say controller, these are the the computers [13:37] and the integrated kind of systems management? [13:39] Yeah, so there's three real-time computers in the vehicle [13:42] that run the whole platform. [13:43] So they run everything from propulsion to body control [13:46] to your lighting, all the real-time systems. [13:51] And that was a big lift to move all the software [13:54] into this new architecture. [13:57] But we also used that as a chance [13:59] to completely change out the way we approached autonomy. [14:01] And so that effort started in actually about late 2021 [14:05] where we recognized we needed to move entirely [14:08] to a platform where we owned [14:09] the entirety of the perception stack. [14:11] So all the cameras, in our case also the radars, [14:15] a much more powerful compute platform for autonomy [14:18] and launch that in a way [14:20] such that we can build this large data flywheel [14:22] where we can train the system [14:23] and what we often call now end-to-end training [14:25] where we have cameras, radars that feed in, [14:28] that's building a large parameter, [14:30] multi-billion parameter foundation model [14:33] that governs how the vehicle drives. [14:34] We then deploy that in a distilled state [14:36] back into the vehicle. [14:38] So that's now in the car, and the slope of the curve [14:41] in terms of progress and additional features [14:43] is going to be really steep. [14:45] Your Gen 1 won't have that, but Gen 2s well. [14:48] But that was a huge part of the lift [14:50] and all that laid the network architecture, [14:52] the consolidation of ECUs, this refactored software stack, [14:55] it laid the foundation for us to do a lot of other things. [14:58] So we're about to launch our lower price vehicle, [15:01] which we call R2. [15:02] It's the coolest thing we've ever done. It is awesome. [15:05] It's a smaller SUV, five passenger, [15:06] as I was telling you before, you will love it. [15:09] If you like the R1, you'll really like the R2, [15:12] but it's a starting price of $45,000. [15:16] It has all the essence of a Rivian that you see in R1 [15:18] but squeezed into a slightly— [15:21] And higher volumes, I presume. [15:22] Much higher volumes. [15:23] Can you give us the comparison [15:24] between R1 volumes and R2 volumes? [15:26] So R1, the average selling price is just over $90,000. [15:30] And so it's, in terms of... [15:33] These things are always subject to boundary diagrams. [15:35] But if you draw the boundary diagram [15:37] around electric premium vehicles, [15:40] so like premium SUVs that are over $70,000, [15:44] we're by far the bestselling vehicle [15:47] in that class of vehicles. [15:49] So we have about 35% market share [15:51] as we draw that boundary diagram. [15:52] Said differently we have the bestselling electric SUV [15:56] over $70,000 in the United States. [15:58] Interesting, in California and the state of Washington, [16:00] we have the bestselling SUV electric or non-electric [16:04] in California and the state of Washington. [16:06] That fits with what I see out on the roads. [16:07] Yeah, so if you're here in the Bay area, like everywhere, [16:09] so it is the bestselling premium SUV. [16:11] But the size of the market [16:12] for someone who's buying a $90,000 vehicle [16:14] is just naturally more collapsed. [16:17] So kind of how many R1s do you ship a year? [16:18] This year we'll do, call it 40-ish. [16:21] 40,000, okay. [16:22] And so what are you tooling up for [16:23] in terms of R2 production? [16:25] So R2 we're going to build first in our plant in Illinois, [16:28] just south of Chicago, a town called Normal. [16:30] And there we'll be building up to 175,000 R2s. [16:34] In addition to R1 in addition to our commercial vans. [16:37] But we're also building a plant in Georgia, [16:40] which will be the further expansion [16:42] and the size of the market for R2 and then R2 variants. [16:45] So there's an R3, there's other things [16:47] we haven't yet announced. [16:48] But there's a whole host of products [16:50] that sit on that platform. [16:51] You know, it's many millions of units. [16:53] It's the whole, if you think of [16:55] the whole automotive demand space, [16:57] the average price of new car [16:59] in the United States is $49,000. [17:00] The most popular configuration [17:01] is a five seat SUV or crossover, [17:05] and so it's just a huge space. [17:07] Okay, so you're on the cusp of a very significant ramp up. [17:09] Yeah, so sort of the first step is going from, in total, [17:13] call it sub 50,000 units a year, [17:15] the plant will be capable of 215,000 units [17:18] with just our Normal facility, Normal, Illinois facility. [17:21] And then we add in Georgia across two phases, [17:24] another 400,000 units of capacity. [17:27] We've taken the approach [17:27] of very heavily vertically integrating technology. [17:30] And so everything from the low-voltage systems [17:33] we just talked about, [17:34] so the computers in the car, [17:36] the software that runs in them, the perception stack, [17:38] our self-driving platform, our whole infotainment platform, [17:42] and then into the high-voltage side, batteries, [17:44] gearboxes motors, inverters, DC-DC converters, [17:47] DC-AC converters, all that power electronics, [17:49] all is done in house. [17:51] And so when you're producing less than 50,000 units a year, [17:54] the engineering fixed cost [17:56] and the operational fixed cost to run that is very high. [17:58] So it's a structural cost challenge, but you reach your— [18:01] Sorry, the fixed cost to run which? [18:03] Well just to run an engineering team [18:05] to engineer all that content, [18:07] but it's only finding its way [18:08] into a small number of vehicles. [18:10] You're saying the systems are very complex [18:12] and so you want to amortize that— [18:13] So once we add volume, [18:15] then it becomes this enormous structural cost advantage [18:17] because we don't have... [18:19] Typically you have a lot of tier one suppliers [18:22] that provide all these systems and they add a lot of cost, [18:25] and for that matter also a lot of complexity. [18:28] So because we do it all ourselves, [18:30] all that complexity and cost is wiped away. [18:33] But you need the volume to really have it make sense. [18:38] It seems interesting to me [18:41] that there was a gap in the market for an electric SUV. [18:45] because when you think about it, [18:46] you guys were thinking about starting originally [18:49] with a sports car, that's what you founded the company [18:50] to start with, and my sense is that you said, [18:53] "Huh, the electric car company [18:56] starting with a sports car idea has kind of been taken." [18:58] That's exactly, yeah. "So maybe we'll pivot." [19:01] But I've driven the Tesla Roadster, [19:04] it's not a good idea [19:05] to put a ton of batteries in a sports car. [19:07] It doesn't handle well, whereas an SUV is heavy anyway. [19:11] It's perfectly suited to being an electric vehicle. [19:13] And so why were you guys able to arrive into the market [19:16] and take what feels like the most obvious form factor [19:19] for an electric vehicle? [19:20] Boy, is that a good question? [19:22] It's funny how things change over time. [19:23] So today it seems obvious, but in like 2013, 2014, [19:31] I would go meet with suppliers, [19:33] and CEOs of very large suppliers and say, [19:36] "We're working on an electric SUV and electric truck [19:39] as a flagship product. [19:40] Following that we want to launch [19:42] more affordable electric SUVs." [19:44] And the lack of understanding [19:47] of what an electric vehicle can do in those conditions [19:50] was so extreme that it's hard to even appreciate this. [19:53] I mean, I would get, to say, "laughed out of the room" [19:55] is probably accurate. [19:56] I mean, it would be like, [19:58] "Why would you make an electric SUV?" [19:59] And the questions would be, "Well what if it [20:01] has to go over a bumpy road?" [20:03] Or, "What if it needs to go through water?" [20:06] And so there's this perception [20:07] that you couldn't drive an electric vehicle off road. [20:10] You couldn't go through deep puddles. [20:12] And so born out of that, [20:13] we realized it was actually this amazing opportunity [20:15] to not only say, "Yes, you can make an electric SUV," [20:19] or, "Yes you can make an electric lifestyle pickup," [20:22] but you can make it so much better [20:25] than anything in the combustion world. [20:27] And so we really had fun with that. [20:29] I mean the R1, our new quad motor R1 [20:32] does zero to 60 on the R1T, the truck, [20:35] we'll do zero to 60 in just under two and a half seconds. [20:38] It can manage the most complex off-road trails. [20:42] Like Hell's Revenge or any of the really hard stuff [20:45] where you're going up almost 45 degree angle [20:48] steep rock faces, it can do that no problem. [20:51] And you can use it every day. [20:52] And so like the comical thing [20:53] is you can pull up to a stoplight, [20:55] you look to your left and you see a brand new Ferrari [20:59] and you're actually quicker than that car. [21:02] And you've got groceries in the front trunk, [21:04] your kid's in a car seat [21:05] and silently it just out accelerates any of those things, [21:08] and then it can veer off the road and drive over a mountain. [21:11] So we wanted to make this product [21:14] that's just completely challenged [21:16] those incorrect assumptions around technology. [21:20] So if you rewind to those suppliers in 2013, 2014, [21:23] is what they missed, one, they were maybe thinking [21:25] about the Ford F-150 market, [21:28] which will be slower as adopters, [21:30] but there's a very large just general everyday SUV market, [21:36] and kind of outdoor enthusiast market. [21:38] Is kind of thing one, they kind of misunderstood [21:40] some of the truck markets. [21:41] And then secondly, you can actually make a better vehicle [21:45] when you have all this abundant chips power [21:47] available to you because especially as you get [21:48] into the outdoorsy stuff, [21:50] having just an abundant source of power [21:52] is actually pretty nice [21:52] for a lot of the other functionality. [21:53] Yeah, I think with electrification [21:56] we have found that particularly early on, this has shifted. [21:59] But there was this perception [22:00] of what an electric vehicle buyer was [22:02] and if the description of a Rivian was, "It's electric" [22:07] and then you describe all these compromises, [22:10] that sort of fits the historical, [22:12] call it 2014 description of electric vehicle buyer. [22:15] Buying a vehicle, accepting a bunch of compromises. [22:18] But we have customers today [22:19] that are not buying it because it's electric, [22:21] they're buying it because it's great to drive, [22:23] it's fun, it's capable, it fits their lifestyle. [22:26] And so that was a goal No-one ever got [22:27] super excited about a Chevy Volt. [22:28] They're like, "This is the car I want." [22:30] We wanted a vehicle that was like, [22:31] "If you're cross shopping this with a Porsche Cayenne Turbo" [22:35] or let's say like a Wrangler Rubicon, [22:39] you'd be like, "Wow, this is a better vehicle." [22:41] And it actually does the things that the Rubicon can do [22:44] and it does the things [22:45] that the Cayenne Turbo can do in one car. [22:47] Where did the culture [22:49] of surprise and delight features come from? [22:51] Because that's one thing I associate with Rivian. [22:53] We have here your gear guardian, [22:55] is that what you said this fella's name is? [22:57] The kind of animated fellow [22:59] who kind of makes clear to any ne'er-do-wells [23:03] that they are being monitored [23:05] if they're trying to break into the car. [23:06] You also have things like the flashlight [23:08] is a very cute feature. [23:09] And then I also noticed, [23:11] this is not quite surprise and delight, [23:12] but they're often features that other cars will have [23:14] that I find they're just executed really well in the Rivian. [23:17] And so the functionality where you pair it with your phone [23:19] and then it kind of locks and unlocks [23:20] based on your coming and going. [23:22] Lots of cars claim to have that, [23:24] but unless it works super reliably, [23:26] it's actually more annoying than not having it, right? [23:28] because you walk away from the car and you're like, [23:29] "Huh, did it lock?" or whatever. [23:30] And again, the execution of that feature [23:32] on the Rivian is really, really good. [23:34] It's better on Gen 2. [23:36] I will, I promise that someday I'll upgrade. [23:39] Where did this come from? [23:41] Why have a flashlight in the car? [23:43] I'm a huge car enthusiast [23:44] and I think that's really helpful [23:47] in that I love cars, I've grown up loving them. [23:53] And so as a kid, when you're growing up reading about cars, [23:57] it connects with this emotional, [24:00] almost dreamlike part of you [24:01] where you're imagining and you're dreaming of things. [24:04] I have three kids now [24:06] and I ask them to sketch, "Hey, sketch me [24:08] what an R12 looks like, or an R9, or an R23. [24:12] The ideas they come up with [24:14] are so wonderfully removed [24:16] from the practicalities of what's possible. [24:18] And they have all these like surprising features. [24:20] Sometimes they're like wildly not possible. [24:24] But we wanted to have some of that kid-like joy [24:27] built into the product, [24:28] and even the aesthetics of how the vehicle looks. [24:30] It's one of the fastest cars you could buy today, [24:34] but it doesn't look mean, it looks friendly. [24:37] And so we wanted the vehicle to be lovable. [24:39] We wanted a lot of, when you fall in love with something, [24:43] it makes like the little things nice. [24:45] And so, I mean, my kids and I were at a farm the other night [24:48] and it got dark and we're walking around in the cornfields [24:51] and we needed a flashlight. [24:52] So it was perfect, we walked back to the car, [24:53] pulled a flashlight out of the car [24:55] and everybody that was with us was like, [24:56] "How'd you get a flashlight?" [24:57] I'm like, "Oh, I have one in the car, it's built in." [25:00] And so it's those little features that they just unlock. [25:03] That made the whole night different for myself, [25:06] my kids, and a bunch of other people. [25:08] And so we thought about [25:09] those kinds of features really carefully. [25:11] What are fun features that you left [25:13] on the cutting room floor [25:14] that you would love to have put in? [25:15] Or maybe you'll do in a future car, but... [25:17] On R1? Yeah. [25:19] Oh, I shouldn't say this here [25:20] because everybody's going to say, "Bring it back." [25:21] Up until very, very late in R1 we had two roof options. [25:27] One's the glass roof, [25:28] which is what you guys have in your car, [25:30] and the other was a beautiful carbon fiber roof [25:32] that you could take it off the top [25:36] and it fit perfectly puzzled into the front trunk. [25:38] Oh wow. [25:39] And so you get this, now when you look up with the glass, [25:42] imagine there's no glass there and it's open. [25:44] So you get this really nice open air— [25:45] A Jeep Wrangler kind of... [25:46] Yeah, that was really nice, [25:46] especially in a car like this, to have that capability. [25:50] Future R1, someday we'll bring that back. [25:52] That was really cool, we all lamented losing that, [25:54] but it was just a lot. [25:55] Again, back to complexity and launching, [25:57] we decided just to manage our own complexity in that. [26:00] But that was one. [26:02] Now R1 has just about everything we wanted. [26:04] The new quad we brought back this, [26:07] we call Kick Steer feature, where we can use [26:09] the fact that there's four motors in the quad [26:11] in the four-motor version to introduce torque vectoring [26:14] to an extreme degree where you can pivot— [26:16] Like an F-22 Raptor, you're cross vectoring. [26:18] Yeah, you can pivot it on its axis, [26:21] so it's really fun off road. [26:22] It's not an on road feature [26:24] you're not going to use it on the highway, [26:25] and it's quite useful. [26:26] Not with that attitude. [26:27] Yeah, not with that attitude. [26:29] But, so we brought that back. [26:30] I think as I look at R2, there's some features on it [26:33] that we didn't put in R1, but that really make R2 special. [26:36] So R2 still has a flashlight, [26:38] but the rear glass on the lift gate now drops. [26:42] And I've been using an R2 with the kids to do school [26:43] drop offs. [26:47] And it is so cool to come up with the back glass open [26:50] and the kids just throw their backpacks in [26:51] through the back of the car, so I love that. [26:55] And then we've added a few other [26:57] really thoughtful features around. [27:00] Like all the storage compartments [27:01] are thought about from the perspective of, [27:04] "Do you have kids, do you have pets, [27:06] do you have a backpack, do you have a handbag? [27:08] Do you have something that you need to put somewhere [27:10] and where does it go?" [27:11] And increasingly, I found in a lot of modern cars, [27:14] you just don't have places for all this gear. [27:16] And so we do personas. [27:19] So imagine you're a backpacker, where does all your gear go? [27:22] Imagine you're commuting to the office, [27:24] where does your gear go? [27:25] Imagine you've got two kids [27:26] and a diaper bag, where does that go? [27:28] And so we really have thought about it [27:30] from the perspective of users. [27:33] And that user-centric design just manifests [27:35] into all these things being thought about, really. [27:38] The last time I saw you, [27:38] I was asking you about CarPlay, [27:40] and I was saying how you guys needed that. [27:42] But then I realized afterwards [27:44] that I think when people say they want CarPlay [27:47] or when I said I want CarPlay, [27:49] what I actually want is navigational [27:52] and traffic data from a source that I trust. [27:55] And the inbuilt navigation system in a car, [27:58] like from a Mercedes or BMW or something like that, [28:01] no-one thinks that they have the best traffic data [28:04] and then you guys launch Google Maps and it's awesome. [28:06] Yeah, it's great. How did that come about? [28:09] Well, you just nailed it. [28:11] This is a decision, it's generated... [28:14] I said there's many millions of decisions. [28:17] Many of them will never get noticed [28:19] and they're just under the surface. [28:21] One of those decisions that's been noticed quite a bit [28:23] is the fact that we've intentionally [28:24] not included CarPlay in the vehicle. [28:26] And that's not to say we don't think [28:29] a close partnership with Apple's important. [28:31] So we have Apple Music integration. [28:32] We have a bunch of Apple integrations that are yet to come. [28:34] But, a great relationship with the team at Apple. [28:37] But it was more to say we just felt and continue to feel [28:40] very strongly around creating a consistent, [28:43] fully-integrated digital experience [28:46] where you're not jumping between apps, [28:49] let's say from a CarPlay app back to the vehicle app, [28:52] and it's quite jarring when you don't have, [28:54] let's say vehicle-level controls [28:56] when you're in the CarPlay environment. [28:58] That view we've had since the early, early days. [29:01] I think that's going to become even more important [29:04] and more true in a world of integrated AI. [29:06] So suddenly when you can talk to your vehicle [29:09] and say, "Please open the front trunk. [29:13] Please unlock the doors, lower the windows, [29:14] set the temperature, take me to this place. [29:17] I'm hungry, I feel like tacos, [29:18] find something along the way to the charger." [29:21] All those immersive experiences really need the full breadth [29:25] of context from your mobile device, from your vehicle, [29:27] from the vehicle back to you. [29:28] What's the state of charge, [29:30] and trying to do that through a homogenized single layer app [29:35] that works across many, many brands [29:38] and by definition in some ways [29:40] has to play to a lowest common denominator [29:42] is really challenging. [29:43] And so there's a few areas that by not having CarPlay, [29:45] it generates friction. [29:46] The first is maps, and so until we launched [29:49] this partnership with Google, [29:51] we had built our own mapping platform [29:54] where we took OpenStreetMaps [29:55] and we layered in a routing platform on top of that. [29:58] And we had information that we would buy on traffic, [30:01] but it was not going to be as good as Google. [30:02] Yeah, I don't even know if it's irrational for me [30:05] to not trust it, but I just noticed that I don't. [30:07] Yeah, it's better. [30:08] It's a big step forward with Google how we've done it. [30:11] So that I'd say fully eliminated the concern on mapping. [30:16] And then the next is text. [30:18] And voice to text and being able [30:19] to interact with your device. [30:20] And that's equally important for Android [30:22] and for Apple customers. [30:23] And so that's coming very soon. [30:25] Oh, that's cool. [30:26] And then there's really no gap, [30:28] and what we found is working directly with Apple [30:30] on the applications that folks care about, [30:32] like Apple Music, these are the kinds of things [30:34] we'd rather do and give customers [30:36] this more highly-curated choice of different platforms. [30:41] But within the Rivian ecosystem. [30:43] You know, again, I was struck by, [30:44] after we discussed it, [30:46] that again, I don't actually want CarPlay, [30:49] it's kind of funny when you think about it. [30:49] There's the screen with all the icons and the Audible app, [30:52] and no-one wants to go browse through [30:53] their audiobook library on the car screen. [30:56] What they want is navigation data from a source they trust [30:59] and to play audio from their phone. [31:00] Those things they actually want. [31:02] Can I give you my pitches for my suggested Rivian features? [31:06] Yes, bring it on. [31:08] You know, have you seen that episode of "The Simpsons" [31:10] where Homer gets to design his own car? [31:11] Yeah, somebody gave me you a picture of that car. [31:13] Exactly, yeah, yeah, the Homer car. [31:14] But I feel like it's a classic bike shed issue [31:17] where I don't know, no-one's that familiar [31:19] with what Stripe does at the innards level. [31:21] And so they just assume that Stripe [31:23] is doing all the sensible things. [31:25] Whereas everyone drives a car [31:27] and so everyone could be like, [31:28] "Here's what you guys should do." [31:30] So with that, here's what— [31:30] Family members, friends, totally. [31:32] Exactly. Yeah. [31:33] Why do all cars offer memory of the driver, [31:38] either the old-fashioned way that other cars do it [31:40] where you press the position one or two for the seat. [31:44] You guys do it, connects to the phone, it's really nice. [31:47] That moves the seat, that moves the side mirrors. [31:51] No car moves the rear view mirror. [31:54] Why is that the final frontier of adjustability? [31:58] Well, there's no actuator on the mirror. [32:00] Sure, but that's your choice. [32:03] It's definitely our choice. [32:04] We could, it's a cost trade off. [32:08] But you have actuators on the side mirrors. [32:10] Yeah, they're harder to reach though. [32:12] The rear view mirror is harder to reach. [32:13] So the side mirrors, if you go back in time, [32:16] if you go back 25 years. [32:17] Oh, sorry, they're harder for the driver to reach. [32:18] Yeah, yeah, yeah. They used to be [32:19] manually adjusted. Yeah I know, the little... [32:21] And then little knobs. [32:22] And then way, way back [32:25] you'd have to push on the glass to move it around. [32:27] And then we had little knobs. [32:29] The advanced knob technology. [32:31] Yeah, and then we moved to electrically controlled, [32:33] and in those cases we moved to electric controlled [32:35] because it's hard to reach out the window [32:37] or reach across the car. [32:38] In this case it's just there. [32:40] But you guys are the surprise and delight company. [32:41] So here's the question we always have to ask ourselves. [32:43] If we have a certain pool of dollars [32:46] we can spend on the car. [32:49] Essentially we're the arbiters [32:51] of how those dollars get spent, [32:53] and we can't make everyone happy. [32:54] Somebody will maybe wants that, but to do that [32:58] something else either has to fall out [32:59] or the price has to go up. [33:00] Okay, if you're worried about the bond— [33:02] So we end up taking... [33:03] Those are the kinds of things we just— [33:05] You could also do it like, [33:05] you know when you see the really big SUVs, [33:07] they actually just have a video feed projected up there. [33:10] You could also do the video feed, [33:12] which then would not require any adjustment. [33:13] Completely, yes. [33:14] So we actually have had a lot of debate around this. [33:17] This is very actively debated, where you see the mirror, [33:19] it's not actually a mirror, it's just a screen. [33:22] And we've not done that yet. [33:26] If I'm honest, partially, not partially, because of me. [33:29] I've not loved— [33:30] Oh, the team's coming and pitching you on... [33:33] There's a team that would like to do it [33:35] with the mirror as a screen and I've been against it. [33:39] In part because if you have good visibility, [33:41] so if you can see out the back, it's a little disorienting. [33:44] And so my concern is you make 50% of people happier with it, [33:48] but a lot of people would say, "Just give me a mirror." [33:51] And so we've just made the trade to say, [33:52] "We're going to take those dollars [33:54] and spend them elsewhere." [33:56] There's not anything wrong with those as products, [33:57] it's just a decision. [33:59] Okay, next question. [34:00] California mandates booster seats up to the age of eight. [34:04] And I've always found it funny [34:05] that we are boosting the child [34:07] as opposed to moving the third point [34:09] up and down on the seatbelt. [34:10] Where you could just have an adjustable seatbelt [34:13] and then ensure that the child [34:15] is correctly configured where you move the third point down [34:17] and now the seatbelt is appropriately across their lap [34:20] and it would just be much less stuff to be carted around. [34:22] You're not requiring boosters, you just adjust the seat [34:25] to be correctly configured to the child's size. [34:27] And Volvo kind of does this with the XC90 [34:29] where they have an integrated booster seat [34:31] in the middle seat, but it's only one seat [34:33] and it's actually still a booster and everything. [34:35] Why not do adjustable seatbelt points [34:38] as a child-restraining device? [34:39] So we do have the adjustability of the third point, [34:43] we can adjust it, it just only goes down [34:46] to fifth percentile female. [34:48] So it doesn't cover the kid range. [34:51] And the reason for that [34:52] is it's classic unintended consequences. [34:53] So we have to test at different settings [34:56] and so we test across from fifth percentile female [34:59] to 95th percentile male. [35:01] And that's what that range is set up for. [35:02] We also, uniquely in the United States, [35:05] there's no other developed country that has this, [35:07] where we have unbelted tests. [35:10] So we run a lot of our crash tests with no belts on. [35:14] And so the kinematics of a body moving [35:16] in the event of a harsh collision [35:18] are very different if you have your seatbelt on [35:20] or your seatbelt off, [35:21] and so it's a very interesting question. [35:23] There is no clear answer, [35:24] but if you were to say, "We're going to test [35:26] around all occupants being belted," [35:29] you would end up with different solutions [35:32] and different control algorithms for the airbags. [35:34] Oh, that's so interesting. And so we also plan [35:36] for unbelted in a way that I think other countries [35:39] don't contemplate that. [35:41] Sorry, and that is when you say, "We," [35:43] the United States has a certification. [35:45] The United States, yeah. United States, yeah. [35:46] Oh, I didn't know the United States were an outlier there. [35:48] Are there places where if you could change the certification [35:53] and safety rules around cars, you would make changes? [35:58] It'd be really nice if the United States and Europe [36:00] had one set of standards, yeah. [36:03] The biggest difference in terms of safety [36:05] between the US and Europe is in pedestrian protection. [36:08] So when you think about if you hit somebody on the road, [36:11] what you want is you want to have them go up over the car [36:13] but they're likely going to have a hard impact [36:14] with the hood. [36:17] And so in Europe there's very tight [36:19] pedestrian protection standards, [36:21] and so the shape of the bumpers have been designed [36:24] to make sure they get brought over the top of the car [36:27] and then the hood needs to have some give to it. [36:31] And so we've designed our hoods. [36:34] Like, a big part of the hood design [36:36] is to contemplate a head, [36:37] like a human head hitting the hood. [36:39] You know, there's certain cars that you see here [36:40] in the United States that could never pass [36:42] a pedestrian protection in Europe. [36:44] Like for example, a Cybertruck could not work in Europe. [36:47] Interesting. [36:51] You wouldn't expect a car maker like Rivian to use Stripe. [36:53] But one reason they do [36:54] is they need to accept a huge range of payment values. [36:57] So when you're putting down a $1,000 deposit on the car, [36:59] you probably want to use your credit card. [37:01] But when you're actually ready to buy [37:03] and you're spending $90,000, [37:04] you might run into a few issues with that. [37:06] So Rivian uses Stripe [37:07] to securely accept bank payments using ACH. [37:11] The right payment method for the situation, it just depends. [37:14] For Rivian, they want bank payments. [37:16] And many businesses on Stripe, [37:17] they want to accept the blossoming variety [37:19] of international payment methods. [37:22] If you're expanding to Brazil, for example, you need Pix, [37:24] which is now how most people pay online there. [37:27] In Poland offering the local option BLIK [37:29] can lift conversion by over 40%. [37:31] And globally, customers increasingly expect [37:33] point of sale credit using services like Klarna. [37:36] Stripe's Optimized Checkout Suite [37:38] handles all these payment methods [37:39] and their differences for you. [37:42] We use AI to analyze the customer and the context they're in [37:45] and then we dynamically surface the payment methods, [37:47] over 75 different options, most likely to convert [37:50] for that specific customer. [37:52] Learn more at stripe.com/payments. [37:56] Are accident rates going up because of texting and driving? [38:00] Yeah, for sure. Okay. [38:01] But countering that is as the vehicle [38:03] increasingly drives itself. [38:04] Sure, that's where I was going to go. [38:06] There's a battle between people are becoming worse drivers, [38:09] which is kind of bad and scary but true [38:11] because they're just increasingly on their phone. [38:15] And we were walking through the city at one point recently [38:17] and a car kind of blew past the pedestrian crossing [38:19] right as people were kind of in the process of crossing [38:23] and this old guy next to us [38:25] as we were crossing the road said, [38:28] "You have to be on your phone to be driving like that." [38:31] I thought that was an apt description [38:32] of much driving these days. [38:33] But how much are the improvements in semi-autonomy, [38:38] like the fender bender detection [38:39] and stuff like this helping? [38:41] I mean, so I have a 9-year-old, [38:42] a 7-year-old, and a 6-year-old. [38:44] And I truly think that by the time [38:47] my oldest is able to drive, [38:49] the environment's going to be so different [38:51] than what we see today. [38:52] And we're in this... [38:53] The next five years [38:56] we're going to see autonomy features [38:58] where the idea of hands-free, [38:59] or hands-free-eyes-off go from being something [39:03] that a very small percentage of customers [39:05] are looking for or using [39:07] to something that's essentially expected. [39:10] And so for that reason, [39:11] it's by far the biggest investment area for us [39:14] in our business. [39:15] It's the most critical area of technology development. [39:19] But what's changed so much that has me believing deeply [39:22] that the next five years will look very different [39:23] than the last five years is just the way [39:25] that AI is now part of this system. [39:27] So historically the way self-driving systems [39:30] were created was you'd have a perception stack, [39:32] it would see the world, it would identify all the objects, [39:35] it would classify the objects like, "Person, car, tree," [39:38] assign vectors to those objects. [39:40] So acceleration, velocity for each of those, X, Y, and Z. [39:44] And then it would hand all of those objects [39:47] and their classifications to a planner [39:48] that would be rules-based. [39:49] So a planner written by humans that would describe [39:53] the rules of the road and for which [39:54] would make a whole host of decisions [39:56] around what the trajectory should be for the vehicle. [39:58] That whole approach has been completely replaced [40:02] with what I would call Gen 2 of autonomous vehicles, [40:04] which is you still perceive the whole world, [40:07] but you've used a transformer to then call all the images. [40:10] You then feed it into a very large model [40:12] and you train it in the way that you train LLMs, [40:15] you're building this very humanistic [40:17] understanding of the world. [40:18] And if you have a car park that's large enough, [40:20] like what we do, you can create [40:22] this amazing training flywheel [40:23] and the rate of progress is so much faster [40:27] than trying to program how the world works. [40:30] because you just have cars going around in the parking lot. [40:33] Well not yours, Gen 1, but all the Gen 2 vehicles. [40:35] Ah, they're gathering training data for this. [40:37] Yeah, so the obvious ones [40:38] are if you're in self-driving mode, [40:40] if you're on the highway, for example, in our vehicle, [40:41] and it disengages, that's a piece of information [40:44] we train off that. [40:45] More interesting is when you're driving, you as the human, [40:50] the model's running in parallel. [40:52] And when you do something different than the model predicts, [40:55] that's interesting. [40:56] It's like if you change lanes [40:57] and the model predicted you would not change lanes [41:00] or vice versa, if the model predicts [41:02] we should change lanes now [41:03] and you don't change lanes. [41:04] On a singular basis, okay, sort of interesting. [41:06] But you multiply that [41:07] across thousands and thousands of vehicles, [41:09] it becomes a really powerful method [41:11] for teaching a digital driver how to drive. [41:15] Yeah, I feel like the big debate that plays out [41:18] in the public realm on this stuff [41:20] is basically the Waymo versus Tesla approaches [41:23] of more sensors on Waymo's versus fewer sensors on Tesla [41:27] and Elon's being quite dismissive of LiDAR. [41:30] Where do you stand on this sensor debate? [41:32] I think first is, I'd say there's alignment [41:35] amongst Rivian, Tesla, Waymo around the use of AI to train. [41:40] And so the view that you're building [41:43] a large scale foundation model, [41:44] multi-billion parameter model, [41:46] you're building that with the data flywheel [41:49] of the vehicles that are deployed. [41:51] There's consistency there. [41:52] I think for some reason this has been [41:56] a lightning rod issue in ways that I wouldn't have expected. [41:59] You know, if you look at sensor theory [42:01] and even if you have noisy sensors, [42:03] just mathematically it proves [42:04] that having more sensors is a better approach. [42:07] because it always seemed nonsensical to me. [42:08] Yeah, more sensors obviously better and we know how to— [42:11] So that's the reason you have more than one camera. [42:13] So even moving beyond more than one camera, [42:16] we have many cameras around the vehicle, [42:18] but then saying we want additional modalities [42:20] that have a non-overlapping set [42:22] of strengths and weaknesses with a camera. [42:25] And there's no arbitration of the camera versus the radar, [42:29] or camera versus the LiDAR, because it's just not how [42:31] the models are built anymore. [42:32] They have a worldview and then the addition [42:35] of different modalities, it actually helps balance out [42:38] some of the noise and the signals. [42:40] because every signal's going to have noise, [42:41] every camera has noise, every LiDAR has noise, every radar. [42:44] So you just want as many sensors as you can afford [42:46] in the car then combined. [42:48] Yeah, and then it becomes as many as you can afford [42:50] and as many as you can process, [42:52] which ties to how much you can afford. [42:53] And so a few things have happened [42:56] in the last five years that have changed things. [42:57] One, as we move to a neural net-based approach. [43:00] That's what we talked about already. [43:02] The second is that the cost of sensors is... [43:06] Six, seven years ago a LiDAR was $20,000. [43:10] 15 years ago LiDAR was $75,000. [43:12] Today a LiDAR is like 200 bucks. [43:14] So it's a very low-cost sensor [43:16] and it solves certain things like very bright light, [43:19] very low light, extremely well. [43:21] And similarly, a radar is 100 bucks. [43:24] Depending on which one, you can buy the crappy radar [43:26] for 25 bucks, you can buy a great imaging radar [43:29] for 100 bucks, 125 bucks. [43:32] And it gives you rain, snow, fog, [43:35] in ways that cameras obviously don't perform very well. [43:37] And if you've ever driven in really thick fog, [43:39] you probably thought to yourself, [43:39] "Boy, wouldn't it be great if I had a radar [43:41] on my forehead?" [43:42] That's what planes have. Which I think all the time. [43:44] Yeah, but you would navigate those situations [43:47] with a much higher degree of safety. [43:48] But then the last piece I'd call out [43:50] is when you have multiple modalities, [43:53] it allows you to train your perception stack better. [43:56] And so for example, if you're driving in the fog, [44:00] there are subtle little things your eyes will spot [44:03] and you're hunting for, like you're looking [44:04] for the little red dots of the car [44:06] in front of you from their taillights. [44:07] But you have to hunt for them. Yeah. [44:09] Imagine your brain had the assistance in training [44:11] its neural net to immediately know [44:13] where those objects are vis-a-vis radar. [44:16] So your brain can more quickly identify [44:18] the visual characteristics [44:20] and you build this much more robust neural net [44:23] to be able to see in hard situations. [44:26] Bright lights and other really... [44:27] When you're, let's say coming down the 101, [44:31] you've got that sun right in your eyes. [44:33] It's really helpful to have something [44:35] that that doesn't affect. [44:37] But so I think it's strange it became a debate topic. [44:40] I think it's not really... [44:42] Everyone's actually in agreement on the high level points. [44:43] "No, I like that." [44:44] I think everybody's actually in agreement [44:45] on kind of the overall AI approach and then yeah. [44:47] Actually there isn't that much disagreement [44:50] on the sensor stuff, or maybe less than it would appear. [44:54] You're describing some of the progressing components. [44:56] So you're describing how LiDAR and radar [44:58] have really changed in becoming a lot cheaper. [45:01] What else is happening in the land of components [45:03] that you work with? [45:04] Are tires getting better? [45:05] Batteries, tires, windshields, anything like that? [45:08] I think on the electric side [45:09] we're seeing the motors become lower cost. [45:13] So I'll illustrate the point. [45:15] On what's going into R2, [45:16] it's our fourth-generation motor that we've designed. [45:19] And so previous generations used [45:21] what's called a hairpin winding in the stator. [45:22] So you take copper wire, you bend it into a unique shape [45:26] and you've seen these before, [45:27] I'm sure there's a whole ring of wires, [45:30] and then to bring them together, [45:32] depending on how you designed the electric mechanics on it, [45:36] you're welding all these little tips together. [45:38] That was state of the art, and that's what essentially [45:41] 99% of the cars or electric vehicles [45:42] on the road are now doing. [45:44] On R2, on the drive unit that we called Maximus, [45:47] which is my middle son's name, [45:49] which is a great name for a motor. [45:51] We have a continuous winding, so it's one long wire [45:53] that we wind, it's sort of like woven into a mat [45:56] and then we fold it on each other. [45:58] And what it does, it takes the number of welds in the motor [46:01] down by something like 95%. [46:04] And so it's slightly more efficient, [46:06] but importantly it's a lot less costly. [46:09] And so does it materially drive different? No. [46:13] But does it allow us to continually drive the price [46:16] of electric vehicles down, yes. [46:17] Okay, so a lot of components that are pretty important [46:20] for electric vehicles is a meaningful cost curve [46:23] of improvement happening on them. [46:24] Yeah, yeah, yeah. [46:26] And then I think if you take the complexities [46:28] of designing real-time operating systems [46:32] and a compute platform across the vehicle, [46:35] these are things that car companies [46:36] don't historically do very well. [46:38] We've got, as an industry, [46:39] there's this massive supplier base of tier ones [46:43] that provide functions and computers alongside of them. [46:46] So you buy a seat, it historically has come [46:48] with a little computer. [46:50] You buy a car with power windows, [46:51] it has a power window computer. [46:53] And so with the exception [46:55] of two companies, Rivian and Tesla, [46:58] every car on the road has, [47:00] depending on the amount of content, [47:01] anywhere from 50 to maybe 150 [47:03] little computers that run the car. [47:05] This is why you, in a normal car, [47:07] could not adjust the, say, the side mirrors [47:10] from the center console. [47:12] It's two or three different companies. [47:15] Companies and different software. [47:16] It's islands of software written by a little supplier [47:18] or on little ECUs that go back to suppliers. [47:21] And so that actually underpins, [47:23] we did a a $5.8 billion deal with Volkswagen [47:26] where we're providing them, [47:27] this is the second largest car company in the world [47:29] where we're providing them a zonal architecture. [47:32] So a platform that allows us to run the whole car [47:36] with a very small number of computers, [47:37] depends on the size of the vehicle, [47:39] and then a software platform. [47:41] I think it's inconceivable that by, [47:45] call it early 2030s, that a car company can exist at scale [47:50] and maintain their market share [47:52] and not have a software-defined architecture [47:55] and be on an architecture for which they have. [47:58] I think of the software architectures [48:00] that are in cars today, it's like a field of weeds. [48:01] It's like you just have all these little ECUs [48:03] all over the place, they're all little islands of software. [48:05] You can't make updates to them. [48:07] None of them are written by yourself. [48:08] There's so many abstraction layers [48:10] between the actual code and the manufacturer. [48:13] I think that must go away [48:15] for you to be competitive in a world of AI [48:17] where you want deep contextual understanding [48:19] of what's happening across the vehicle. [48:21] And being able to create these highly-immersive, [48:24] highly-evolving experiences [48:25] that get better and better over time. [48:27] And so I think every car company's either going to go [48:29] try to develop it themselves, which is hard [48:30] because they don't typically have those skill sets. [48:32] Try to source it from suppliers, that's very hard [48:34] because those companies are precisely the ones [48:36] that don't want to see all their little computers go away, [48:39] or work with us. [48:42] Obviously I have a biased point of view, [48:44] but we think of the Volkswagen partnership [48:47] as the ultimate billboard for the idea. [48:50] It's to be able to deploy this platform [48:53] in such a wide range of vehicles from flagship Porsches [48:57] down to like a $22,000 European Volkswagen [49:02] demonstrates the scalability of this architecture. [49:04] And so I think that's going to be a huge [49:05] change that must come. [49:08] And if it doesn't, sort of like if manufacturers [49:10] don't make that change, [49:11] they're just going to lose market share, [49:12] and the ones that do have that technology [49:14] are going to gain a lot of market [49:15] share in the next 10 years. [49:16] Yeah, I think people don't realize how much, [49:18] like with traditional car manufacturing, [49:21] the brands are in the car assembly business. [49:24] And you're going to car Costco [49:26] and you're filling your cart with, [49:27] "We need a backup proximity sensor [49:30] and we need a rear view mirror adjust." [49:32] We're spending a lot of time [49:33] on our side mirror adjustment system and things like that. [49:37] But they're kind of assembling all these components [49:39] from sub manufacturers that are not integrated. [49:42] I think your point is like in a traditional car, [49:45] the backup beeping sensor, "Oh, it can only beep. [49:47] That's the only thing it does." [49:48] And so you can't integrate it [49:50] with the rest of the car and the systems, [49:51] and has that approach. [49:52] Whereas you think the integrated approach [49:54] is just definitely going to win. [49:55] What's also, it's an interesting moment in time [49:57] because if you were to build a spreadsheet [49:59] that looked at all the features [50:01] of a classic network architecture versus us, [50:05] you'd say there's actually, [50:06] "Looks like you have all the same features." [50:07] But then you look at how they're created. [50:09] And in our case we have, on a relative scale, [50:13] a small team of software developers that control everything [50:16] and we can make changes very quickly. [50:18] If you have an idea, I can put it into the next release [50:20] and it'll be in the car in two weeks. [50:21] And you compare that with something [50:23] that's very, very requirements-driven. [50:24] So the manufacturer writes requirements sometimes years out, [50:27] and it may be things like the seat will have six presets [50:31] and that gets hardwired in, [50:33] and it's very hard to change things or evolve things. [50:37] And so I think that again, it's a moment in time [50:42] where there's a huge amount of effort [50:43] to try to have a similar set of features to what we do. [50:46] But then they can't do cool things like we do. [50:48] So we have a couple of things we do every year [50:50] that try to flex this [50:52] into a very easy-to-understand consumer— [50:54] And the simplest version of it. [50:55] Sorry, you're probably going to say something like this, [50:57] but the simplest version I see is I'll walk out to the car [51:00] and it unlocks based on phone proximity [51:02] and adjust the seats to me, again, [51:05] based on knowing that it's me. [51:06] It's minor but it's really nice. [51:08] But that's like 15 different companies. [51:10] So you have to do HVAC adjustments, seat adjustment, [51:12] lighting adjustment, exterior sound, interior sound, [51:14] vehicle state, battery state. [51:17] But here's where it's exciting [51:18] is we can do something like a Halloween mode. [51:20] So we do one every year and the cars get dressed up. [51:22] So last year we had a "Night Rider" version, [51:24] we had a "Back to the Future" version. [51:26] And when you walk up to your car [51:28] and it's in that mode, it does a totally different sequence. [51:31] It plays, in the case of "Night Rider," [51:34] the light on the front goes... [51:36] You get in the inside of the car [51:37] is different lighting sequences, different sounds. [51:39] And it's just for Halloween. [51:41] But that's only possible if you control all those systems. [51:45] And so we're at the tip of the iceberg [51:48] of being able to do these very immersive, [51:51] very hard to recreate modes [51:54] that you can't do with a traditional architecture. [51:57] "Night Rider" was early [51:58] to having a lot of on-device compute. [52:01] The value of a zonal architecture and systems integration. [52:04] Speaking of actually, there was a Pontiac Trans Am [52:06] if I recall, cars used to have cool names. [52:09] You know, you have Mustang, Thunderbird. [52:11] All these really nice names, El Dorado, very evocative. [52:16] Now we have R1, R2, Model 3, X5. [52:21] Why? Yeah, have we lost [52:23] a bit of the sense of adventure? [52:25] Could we have some— [52:26] Yeah, could be, it's a good question. [52:28] Evocative names? Alpha numeric versus name. [52:30] We decided to just keep things simple [52:32] and we had, it's a structured name. [52:34] Yeah, but you're a surprise and delight company. [52:36] You have a "Night Rider" mode. [52:38] Yeah, I don't, maybe we should have. [52:39] Is that something that's befitting of an R1 or of a— [52:41] Yeah, we had those debates. [52:43] You guys are super outdoorsy. [52:45] Apple's taken the Yosemite, El Capitan naming scheme, [52:49] but again, I feel like there's— [52:49] So all of our cars end up having code names. [52:51] Yeah, can we just use the code names as the public names? [52:52] So the code name for Gen 2 is Peregrine. [52:54] Okay, beautiful. Which is a super fast Falcon. [52:57] Would anyone hear buy a Rivian Peregrine? [52:59] Yeah, I'm seeing a lot of takers. [53:00] But you'll laugh. [53:01] So the first batch of prototypes we built [53:03] then the first one of those vehicles of R1, [53:06] like production representative, [53:08] the code name for that vehicle is avocado. [53:11] Not as good as Peregrine. Less of a saleable name. [53:13] Okay, I can help you with the names. [53:15] Again, back to the bike shed, [53:16] but I'm happy to advise on that. [53:20] as we talk about the Rivian business. [53:22] Think about a car business 20 years ago, [53:24] say a US manufacturer. [53:26] It's, I think regarded as a hard business. [53:28] Capital intensive, it's about getting [53:31] to a significant number of units in market share, [53:33] being disciplined on price, [53:34] generally trying to move people [53:35] up from sedans to higher-value vehicles, [53:38] is at least for the American car makers [53:40] how they try to improve the business. [53:43] How should investors think [53:44] about what the Rivian business is today? [53:47] Where there's, again, the classic car business [53:48] where you're trying to sell a lot of units, [53:50] trying to get volumes up and you do other stuff [53:52] like you do insurance and lending and things like that. [53:54] But car makers have always done that to drive more volumes. [53:58] Or is the business actually very meaningfully different [54:02] from a classic automaker where you have [54:03] this big Volkswagen deal [54:05] that sounds like a capital-light, [54:07] more kind of B2B side of the business. [54:10] Obviously people talk about how ride sharing [54:11] is going to change things. [54:12] And so I'm curious how much Rivian [54:13] is a classic car business done well [54:16] where it's a P times Q, we're going to sell a lot of cars [54:19] and make a margin on them, and you have the volume ramp. [54:22] Versus how much is it, "Oh this is actually [54:25] a totally different kind of business. [54:26] And we're going to rent cars to people [54:28] in the ride sharing world [54:28] or we're going to sell software to other car manufacturers?" [54:31] I would characterize there's three parts of the business. [54:33] There's the volume of selling vehicles. [54:35] And the revenue associated that is quite large. [54:38] And so even with each, today it's $90,000 for every sale, [54:44] but let's say it's half that, [54:47] it still just creates a tremendous amount of revenue [54:49] and helps cover a lot of the fixed costs. [54:51] And allows you to develop technology [54:53] that even if a small percentage of that [54:56] is being put back into technology, [54:57] allows you to develop a lot of technology. [55:00] And so that's core to what we believe, [55:03] and we built a brand that people really love. [55:04] So we just need to get to a higher volume of products [55:08] to pull that together. [55:09] Like 200,000 R2s times 50K per is $10 billion of revenue. [55:16] So you can really start to move up the revenue. [55:19] So last year I think we were just around $5 billion. [55:22] So we'll see our revenue really start to climb. [55:25] And so that is absolutely a part of the business. [55:27] Then there is technology sales [55:30] and there's a lot of different ways to accomplish that. [55:32] But Volkswagen's an example of that. [55:33] But we look at what we're doing across our technology stack. [55:36] There's a bunch of areas that we believe [55:38] are things that can be monetized [55:40] to generate incremental revenue and often revenue [55:43] that has a higher margin profile [55:44] than let's say the vehicle sales. [55:46] Vehicle sales, let's say that comfortably sitting at, [55:49] when we get to the right steady state, [55:51] 20% to 25% gross margin. [55:54] And then there's a third piece [55:55] which is different business models. [55:57] And a big unlock here is actually just being D2C, [56:02] being Direct-to-Consumer. [56:04] And so, with the exception of us and Tesla, [56:07] every manufacturer today sells through distributors [56:09] and distributors then onto dealers [56:11] or in a few cases Direct-to-Dealers. [56:14] But nonetheless there's a middle layer. [56:16] I think there's two middle layers. [56:17] You have to sell to distributors who sell to dealers. [56:19] Yeah, often it's a distributor then to a dealer. [56:21] Yeah, yeah, yeah. That often draws [56:22] the attention to the sale, [56:24] saying, "Oh, you're paying a third party [56:25] 10% to 15% to sell your car." [56:28] That's true, that's a far less efficient way [56:32] to sell the vehicle than having a direct relationship. [56:34] But I think that the unappreciated part of it [56:36] is you lose out on all the business [56:39] that happens post-sale. [56:41] And so the service business actually becomes [56:43] a very large part of our revenue generation. [56:46] And if you look at car dealers and you look at, [56:49] if you know any friends that own large car dealer groups, [56:52] they're very profitable. [56:54] And where they make the vast majority [56:55] of their money is on service. [56:58] The thing you have to wrap your head around though [56:59] is in the beginning, service is not a profitable business [57:01] because you have to build, so you're investing [57:03] a very large amount in building capacity. [57:05] So service infrastructure, so a ton of Cap-Ex. [57:08] You have to build up a large team of service technicians. [57:10] This is a huge spend category for us. [57:13] And in the beginning we're just performing warranty work. [57:16] We don't really have paying service yet. [57:18] But it suddenly goes from this thing that's a cost center [57:21] and either feeding in as part of a warranty expense [57:25] or feeding in as Op-Ex [57:27] and suddenly it becomes cogs and profitable cogs [57:30] and it's like an afterburner to the business. [57:31] because it's the car park's then at a big enough size [57:34] where it just is throwing off [57:35] billions of dollars a year in profit. [57:38] That's the goal. [57:39] That's the services side of the business [57:41] we think of as quite interesting in the long term [57:43] and something we protect. [57:45] And it becomes even more valuable [57:47] if you have a software-defined vehicle [57:48] because there's an opportunity for me, [57:50] think about used cars. [57:52] Used car's another huge profit center for the auto industry. [57:54] And today when you buy a used car, [57:56] if you're a business that's buying a used car [57:58] to then sell to someone else, [58:00] you buy it, you maybe do a 25-point check, [58:02] maybe you do a 50-point check, [58:04] you maybe put some sort of a stamp on it [58:06] to say this has got an additional warranty, [58:09] but you actually have very little understanding [58:10] of what the health of the asset is, [58:12] and you have no ability to incrementally improve the asset [58:16] with software improvements, with a software update. [58:18] We control that so we can decide, [58:21] "Hey, we want to invest in making software improvements [58:25] that will raise the value of these vehicles." [58:28] And so it allows us to play very differently [58:30] in the residual space and uniquely control [58:33] and have profit built into the used car business. [58:36] And so that's another arm that today, [58:39] it represents a very small percentage of our revenue. [58:41] There's not a lot of used Rivian transactions. [58:43] We participate in a lot of them. [58:44] We buy Rivians and sell them. [58:45] Okay. Yeah, yeah. [58:46] But in the long term it becomes a very big business. [58:49] Okay, this is very interesting. [58:50] So traditional car maker, part of why the business is hard [58:53] is because you sell units and the business [58:55] is kind of the least recurring revenue [58:57] you can imagine where you're— [58:58] All the recurring revenue somebody else gets. [58:59] Exactly, and so you're always in this dog fight [59:01] for next year's sales and the following year's sales. [59:03] Whereas you're saying you guys are selling cars [59:06] and then you have monthly active users of Rivian out there [59:11] and you have a direct service business [59:12] because you have direct relationships with those customers. [59:14] And that really changes the direct business. [59:17] Nevermind the B2B stuff. [59:18] It depends on the brand, but if you look at it like this, [59:20] for every car that's sold, [59:23] whatever its sale price is, [59:24] the revenue over the life of the vehicle [59:26] is generally around two to two and a half X [59:29] the initial sale price. [59:31] But that's over a 15 year time horizon. [59:34] But there's a lot of revenue [59:34] that happens associated with the vehicle. [59:36] And of course we know that, think of all the businesses [59:38] that exist that are not car companies [59:40] that exist because of those cars. [59:43] And so some of those businesses [59:45] have less relevance in a EV world. [59:49] So do I need to go to a Jiffy Lube to change the oil? [59:53] No, but what I like to have a software update? Definitely. [59:58] That's still a complex thing. [59:58] There's still components that wear out. [60:00] For sure. Yeah, yeah, [60:01] my car's telling me to rotate my tires. [60:02] Why do I need to rotate my tires? [60:04] In your car? [60:05] Because we're biasing the front tires [60:07] when you're in efficiency mode, [60:08] so it's wearing the front tires [60:09] a little bit faster than the rear tires. [60:11] And then Gen 2 it does this dynamically? [60:13] Gen 2, yeah, it can, if you put it [60:16] in what we call the All Purpose Mode, [60:18] and then R2 is even better. [60:20] Because the way we do it dynamically is we... [60:23] I'll go deep here for just one second. [60:25] So it's a permanent magnet motor [60:26] and so a permanent magnet motor [60:27] generates losses if it's spinning. [60:30] And so what we do is when you're in conserve mode, [60:31] it automatically disconnects the rear axle, [60:34] and so you're using just the front. [60:35] On Gen 2 with the dual motor, [60:39] we dynamically disconnect the rear axle even in All Purpose. [60:42] And then R2 we've been able to make it so fast [60:45] that it connects and disconnects [60:46] and you don't even know it's happening. [60:48] So it feels really sporty, [60:48] because when you hit the throttle, [60:52] it all very, very quickly connects everything. [60:54] But you have to mechanically disconnect the rear axle. [60:56] So you think of like, "Can you do that in 140 milliseconds?" [61:01] It's a lot of stuff to do in 140 milliseconds. [61:04] And going to two-wheel drive rather than four-wheel drive, [61:06] that's an efficiency thing? [61:07] Yeah, because you turn off, [61:09] you completely disconnect the motors. [61:10] So there's no losses associated [61:12] with spinning the permanent magnet machine. [61:13] Interesting. Okay. [61:15] The big EV story of the past 10 years [61:17] has been the China ramp up. [61:21] Why, obviously the Chinese EVs are cheap, [61:24] but they're also very good. [61:26] How did China get so good at EVs [61:28] and how does the US [61:29] continue to be competitive with the global market? [61:31] Obviously the US is protected as a market, [61:34] but as you think about winning other countries, say Europe. [61:37] Europe in particular, yeah. [61:39] So there's two pieces of the China story [61:40] and I think the one that we often are drawn to [61:45] is the cost story. [61:47] And so one of the things about the car industry [61:51] is everybody buys everybody's products. [61:53] And so you can go buy, if it's on the market, [61:56] you can buy it, so everyone buys them, they take them apart. [61:59] There's Jim Farley talking open about EVs. [62:01] On the mechanical side, there's nothing that's not seen [62:04] by every car company. [62:05] So when we launch R2, everyone will have it. [62:07] Everyone will take it apart. [62:08] And so what I call out is that there's not any magic [62:11] that's happening in terms of how Chinese vehicles [62:13] are manufactured, the processes, [62:16] the design of the components and the systems. [62:19] There's high degrees of overlap in some cases, [62:23] very similar to what we have here in the US [62:26] and very similar to some of the things [62:27] we've done within Rivian. [62:28] Or if you look at comparisons to Tesla or others. [62:31] There's obviously clean designs. [62:33] So they have great part consolidation, part elimination, [62:35] but there's not something that you'd look at and say, [62:38] "Oh, aha, this is how the cost has come down." [62:40] And what's driven the cost down [62:42] is that you have the compounding benefits [62:44] of much lower cost of capital [62:46] across not just the manufacturer, [62:48] the vehicle manufacturer, [62:49] but across their whole supply chain. [62:50] So tier one, tier two, tier three. [62:52] Everyone's cost of capital is in some cases zero [62:55] or close to zero because a lot of that capital [62:57] is provided by the local government. [62:59] And in some of the larger companies cases, [63:02] the federal government, [63:03] or their equivalent of the federal government. [63:04] And then you have a much lower cost of labor, [63:06] and you put those two things together [63:08] and you multiply it across every part, [63:12] all the way down to every source of raw material. [63:15] It just ends up with a cost structure [63:16] that's very advantaged relative to the western world. [63:19] Particularly very advantaged relative to the United States. [63:22] And that doesn't translate. [63:23] Meaning if you took a car, the exact same design [63:26] of a Chinese car and you produce that in the United States [63:28] with a US or North American-centric supply chain [63:32] in North American cost of capital, [63:34] the cost would be much, much higher. [63:37] And so we're seeing tariffs [63:39] in the case of the United States go in place [63:40] that try to take up the fact [63:41] that there's difference in labor cost, [63:43] difference in cost of capital. [63:45] I think Europe's a big question as to what happens there. [63:47] But depending on how those regulations fall, [63:50] if there's zero trade friction between, [63:55] let's say a country in China, [63:57] more production in the fullness of time [63:59] will shift to China because of those advantages. [64:01] The lower cost of labor, lower cost of capital. [64:03] But is it not just the cost structure? [64:05] because you've seen the videos of the Neo, [64:07] the champagne glasses on the— [64:09] So that's the second part. [64:10] So that's the second thing. [64:11] And this is the part that I think, [64:12] I'm glad you touched on this. [64:14] This is the part that actually [64:17] everyone should be taking note of, [64:18] which is the technology is very good. [64:21] And this isn't true for every Chinese car, [64:23] but there's a subset of the Chinese brands [64:25] that have done an excellent job [64:27] in many ways for the same reasons [64:29] that Tesla and Rivian have developed [64:31] our technology different than the incumbent manufacturers, [64:34] that they started with a clean sheet. [64:36] And so if you start with a clean sheet, [64:38] you're going to naturally look at this and say, [64:39] "Well let's design a zonal architecture [64:41] with as few computers as possible. [64:43] Let's own our software stack. [64:44] We're not going to have it owned by 50 different suppliers [64:48] and abstracted across 100 different islands of software. [64:51] Let's vertically integrate around the high-voltage systems." [64:54] And you end up making a lot of these [64:55] very logical first principles decisions [64:58] in ways that are not too different from a Rivian [65:01] or not too different from a Tesla. [65:02] And then because the cost structure's such, [65:04] they've been able to say yes in some cases to more features. [65:09] They haven't done the powered... [65:12] Even they have not done the powered rear view mirror. [65:15] But they adjust, but you could see [65:18] they've said yes to more features. [65:20] Because they have the cost structure to enable that. [65:23] And so I think that if I'm a western manufacturer, [65:28] that would be my biggest concern [65:30] is you have to very rapidly go build a software, [65:34] a software-centric architecture [65:37] and a world-class electronic stack to support that. [65:40] It's sort of obvious, but I think that's a big part [65:42] of what led to our deal with Volkswagen [65:44] is we offered an ability to very quickly get to that state. [65:49] I think every manufacturer should be thinking, [65:51] "How do we build a software-defined architecture [65:53] as quickly as possible?" [65:55] Because to be competitive on a world stage, [65:57] you need to have it. [65:58] Yeah, I can buy that. [65:59] How do you feel about the phasing out of the EV credits? [66:02] In the short term I think it's [66:02] going to create some headwinds [66:04] and in the longer term, I say this often, [66:10] I think it's hard to argue that it's going to create a... [66:13] It absolutely creates a less-competitive environment. [66:15] So you see a lot of manufacturers [66:17] pulling back on their EV investment. [66:19] I think that's, myopically, I'd say it's good for Rivian, [66:22] but I'd say it's bad for the world, it's bad for my kids. [66:25] I think it's unfortunate [66:27] that we don't have more companies leaning in. [66:29] You know, the best measure of, I think, [66:31] the underinvestment in this space [66:33] and the lack of highly compelling product [66:35] is that there's a single company, Tesla, [66:37] that still has around 50% market share in the United States. [66:40] The United States is at around 8% [66:41] of new vehicle sales are electric. [66:43] Isn't that a surprisingly low number? [66:45] 8% of new vehicle sales in the United States are electric. [66:47] Just if you're to ask someone to guess, [66:48] I feel like they guess a higher number. [66:50] But here's why. [66:50] If our products that Rivian sells today [66:53] is our flagship products. [66:56] I really believe that R2 will expand the 8%. [66:59] But if I'm buying a vehicle [67:00] and I want to spend less than 50K, [67:03] There's hundreds of choices in the ICE world. [67:06] I can get sedans, I can get coupes, [67:08] I can get convertibles, I can get minivans, [67:11] I can get crossovers, SUVs, [67:12] all these different choices, different brands. [67:15] If I'm buying an EV, I think there's... [67:18] You think it's just a choice issue? [67:20] I think there's two choices. [67:21] I think it's a Model 3 or Model Y today, [67:23] and everything else is... [67:25] There's maybe a few that are getting close, [67:26] but everyone clearly likes crossovers [67:27] judging from the internal combustion world. [67:29] And so I think for the first time with R2, [67:31] you'll have a choice that's highly compelling, [67:33] that's not a Tesla, that's not to say [67:35] anything negative about Teslas. [67:37] Well the R2 is the first affordable crossover? [67:40] There's others, I'd say it's going to be [67:41] the first highly compelling, affordable. [67:44] That's the key part is you need it to be highly compelling. [67:46] And so there's been a number of EV launches [67:49] from existing manufacturers that haven't gone that well. [67:52] And I think we get the causality backwards. [67:54] We say, "Oh, it's because customers [67:55] aren't ready to buy EVs." [67:56] And I say, "Well no, it's because the EV [67:57] is not that compelling." [67:58] And the existence proof that if you have [68:00] a highly-compelling product, [68:01] there will be a lot of demand is Tesla. [68:03] The Model Y is one of the bestselling cars in the world. [68:06] Now we just need to have 20 other choices [68:09] in order to start to really make a dent. [68:11] And then if you compare this to China [68:12] where there is a lot more choice, [68:15] China's a three and a half, four times higher level [68:18] of adoption of electrification than we are, [68:20] and the slope of the curve is very steep. [68:22] We have a choice issue here in the US, is our view. [68:25] Last question, where did the name Rivian come from? [68:28] So Rivian is a play on words. [68:29] I grew up on the Indian River [68:32] and so it's the first few letters of the word "river," [68:34] and the last three letters of the word "Indian," I-A-N. [68:37] We came up with all these different ideas, [68:39] but we wanted something that was phonetically smooth [68:41] and nice, but didn't mean anything in any language. [68:44] Now it's so natural, but when we first started using it, [68:46] it was like, "Is that right?" [68:47] It's a great name, it sounds really nice, yeah. [68:49] All right, well, RJ, thank you. [68:50] Yeah, thanks. Thanks for having me. [68:51] Super fun. Yeah, it was fun.