Rivian Automotive Inc (RIVN) 2026 Q2 法說會逐字稿

Good afternoon, and thank you for joining us for Rivian's second quarter 2026 earnings call. Today, I'm joined by RJ Scaringe, our CEO and Founder; Claire McDonough, our Chief Financial Officer; and Javier Varela, our Chief Operations Officer.

下午好，感謝各位參加 Rivian 2026 年第二季財報電話會議。今天與我一同出席的有：執行長暨創辦人 RJ Scaringe、財務長 Claire McDonough，以及營運長 Javier Varela。

Before we begin, matters discussed on this call, including comments and responses to questions, reflects management's views as of today. We will also be making statements related to our business operations and financial performance that may be considered forward-looking statements under federal securities law. Such statements involve risks and uncertainties that could cause actual results to differ materially. These risks and uncertainties are described in our SEC filings and the earnings presentation we filed with the SEC today.

在開始之前，本次電話會議中討論的事項（包括評論與對提問的回應）均反映管理層截至今日的觀點。我們也將就公司營運與財務表現發表若干聲明，依據聯邦證券法，這些聲明可能構成前瞻性陳述。此類陳述涉及風險與不確定性，可能導致實際結果出現重大差異。相關風險與不確定性已載於我們向 SEC 提交的文件，以及我們今日向 SEC 提交的財報簡報中。

During this call, we will discuss both GAAP and non-GAAP financial measures. A reconciliation of historical non-GAAP to GAAP financial measures is provided in our earnings presentation and press release. Just before the earnings call, we posted our earnings presentation, which includes an overview of our progress over the recent months.

在本次電話會議中，我們將同時討論 GAAP 與非 GAAP 財務衡量指標。歷史非 GAAP 與 GAAP 財務衡量指標之調節表已提供於我們的財報簡報與新聞稿中。就在財報電話會議開始前，我們已發布財報簡報，其中包含我們近幾個月進展的概覽。

I encourage you to read it for additional details around some of the items we will cover on today's call. Following our prepared remarks, we will be taking questions from sell-side analysts. In the interest of keeping the call to 1 hour, we would ask these analysts to limit any follow-on questions to one.

我鼓勵各位閱讀該簡報，以取得我們今天將涵蓋部分議題的更多細節。在我們的準備發言結束後，我們將接受賣方分析師提問。為了將本次電話會議控制在 1 小時內，我們請分析師將追問問題限制為一題。

With that, I'll turn the call over to RJ.

接下來，我把電話交給 RJ。

Thanks, Chip, and good afternoon, everyone. Thanks for joining us for today's call. In the second quarter, we began deliveries of our two vehicles to external customers. We expect R2 to be a game changer for our customers and a driver of Rivian's long-term growth and profitability.

謝謝你，Chip，各位下午好。感謝各位參加今天的電話會議。在第二季，我們開始向外部客戶交付我們的兩款車型。我們預期 R2 將為客戶帶來顛覆性改變，並成為推動 Rivian 長期成長與獲利能力的動能。

Importantly, in the US automotive marketplace start for high-quality EV choice, I strongly believe R2 is an attractively priced option for everyday ventures that will resonate with a broad set of consumers. Media reviews of R2 are outstanding. One journal wrote, R2 is one of the best new cars I've driven in years. Another praise R2 as among the finest vehicle designs in the world. Early customer feedback of R2 is positive, and I have to say, I love it as my daily driver.

重要的是，在美國汽車市場對高品質電動車選擇的需求起飛之際，我堅信 R2 以具吸引力的定價，成為日常出行與冒險的選項，將引起廣泛消費者共鳴。媒體對 R2 的評測極為出色。有一家期刊寫道：R2 是我多年來開過最棒的新車之一。另有評論稱讚 R2 是全球最出色的車輛設計之一。早期客戶對 R2 的回饋也很正面；我必須說，我很喜歡把它當作我的日常用車。

Strong R2 reviews and increasing brand awareness are directly translating into accelerating customer touch points with our products. In the second quarter, we hosted over 57,000 demo drives, a Rivian record. While it's early in our ramp, we are pleased with the reservation conversion we've seen to date for launch addition R2.

R2 的強勁評測與不斷提升的品牌知名度，正直接轉化為客戶與我們產品接觸點的加速成長。第二季我們舉辦了超過 57,000 次試駕，創下 Rivian 紀錄。雖然我們的爬坡仍在早期階段，但我們對目前看到的 R2 首發版預訂轉換率感到滿意。

Turning to our manufacturing footprint. Our team in normal is focused on the production ramp of R2. As a reminder, R2 production started with a single shift, and we expect to scale to two shifts by the end of the third quarter. We are seeing good progress in new team member training and process improvements for the ramp of the first R2 shift.

接著談談我們的製造版圖。我們在 Normal 的團隊正專注於 R2 的產能爬坡。提醒各位，R2 的生產從單班制開始，我們預期在第三季末擴增至兩班制。我們在新進團隊成員訓練與第一班 R2 產線爬坡的流程改善方面，正看到良好進展。

Additionally, members of our supply chain team are spending time on site with suppliers as we look towards supporting the ramp in the back half of the year and into next year.

此外，隨著我們著眼於支援今年下半年以及明年的產能爬坡，我們供應鏈團隊的成員也正花時間駐點於供應商現場。

In Georgia, construction of our next manufacturing site is progressing well. The combination of our normal and Georgia facilities provides Rivian the path to scale to 515,000 units of annual capacity with plenty of room for future expansion.

在喬治亞州，我們下一座製造基地的建設進展順利。Normal 與喬治亞兩座工廠的組合，為 Rivian 提供了將年產能擴大至 515,000 輛的路徑，且仍保有充足的未來擴張空間。

As an update on our Amazon partnership, Amazon now has over 40,000 Rivian electric delivery vans active in its fleet, delivering packages across thousands of cities in North America. In addition to the EDV standard pack variant already on the road, we're developing new variants with a larger battery pack and all-wheel drive to support Amazon's needs. In the second quarter, we surpassed 1 billion miles driven on the Rivian commercial van platform.

關於我們與 Amazon 的合作夥伴關係更新：Amazon 目前已有超過 40,000 輛 Rivian 電動配送貨車在其車隊中投入運行，於北美數千座城市配送包裹。除了已在路上行駛的 EDV 標準電池版本外，我們也正在開發搭載更大電池組與四輪驅動的新版本，以支援 Amazon 的需求。第二季我們在 Rivian 商用貨車平台上的累計行駛里程突破 10 億英里。

Turning to our technology road map, Autonomy remains one of the most important areas of investment at Rivian. In the not-too-distant future, we believe advanced assisted driving capabilities will be a key differentiator for our customers and a substantial driver of market share. Our autonomy development is on track, and we expect to begin rolling out point-to-point capabilities by the end of this year.

接著談談我們的技術路線圖，自動駕駛仍是 Rivian 最重要的投資領域之一。我們相信在不久的將來，先進駕駛輔助能力將成為客戶選擇的關鍵差異化因素，並大幅推動市占率成長。我們的自動駕駛開發進度如期，預期在今年底前開始推出點到點能力。

We believe our rapidly expanding car park with R2 will accelerate our data flywheel to enable delivering hands-off, eyes off capabilities in 2027 and ultimately, Level 4 capabilities in consumer and robotaxi variants of R2.

我們相信，隨著 R2 使我們快速擴大的車隊規模，將加速我們的資料飛輪，讓我們在 2027 年實現免手動、免視線（hands-off、eyes-off）能力，並最終在 R2 的消費者版本與 robotaxi 版本中達成 Level 4 能力。

Since we began monetizing Autonomy us in April, take rates are trending positively for the service, and we believe further future releases will drive continued increased uptake of Autonomy Plus. Later this year, we will host another autonomy and AI Day where we'll provide more updates around our exciting autonomy and AI progress.

自我們在 4 月開始將 Autonomy 服務商業化以來，該服務的選購率（take rate）呈現正向趨勢；我們相信未來進一步的版本釋出，將帶動 Autonomy Plus 的持續提升採用。今年稍晚，我們將再度舉辦自動駕駛與 AI Day，屆時會提供更多關於我們令人振奮的自動駕駛與 AI 進展更新。

At the end of 2026, we are targeting the launch of our third-generation autonomy hardware -- this includes RAP1, our first in-house design silicon and the addition of LiDAR. The development of our RAP1 chip is on track, and we've begun final testing phases of production silicon. Finally, in May, we rolled out our Rivian assistant on all R1 vehicles and plan to launch the feature on R2 with an over-the-year update later this year. Rivian Assistant is our new AI-powered voice assistant that is built to be a digital co-pilot with integration into the vehicle ecosystem and other external apps. Customers love this new future and its intuitive functionality across applications like Spotify and Google.

在 2026 年底，我們目標推出第三代自動駕駛硬體——其中包含 RAP1（我們首款自研晶片）以及 LiDAR 的加入。我們的 RAP1 晶片開發進度如期，且已開始進入量產晶片的最終測試階段。最後，在 5 月，我們已在所有 R1 車款上推出 Rivian Assistant，並計畫於今年稍晚透過年度 OTA 更新將該功能導入 R2。Rivian Assistant 是我們全新的 AI 驅動語音助理，旨在成為數位副駕駛，並整合車輛生態系與其他外部應用程式。客戶很喜歡這項新功能，以及它在 Spotify、Google 等應用上的直覺式操作體驗。

In closing, I want to thank our team for their continued execution this quarter. We are laying the foundations for a future that we believe will be fully electric, autonomous and AI defined. The reviews for R2 affirm that our category-defining brand and vertically integrated and extensible technology resonates deeply with customers. I couldn't be more excited about the upturn ahead.

最後，我要感謝團隊在本季持續落實執行。我們正在為一個我們相信將全面電動化、自動化並由 AI 定義的未來奠定基礎。R2 的評測印證了我們具類別定義性的品牌，以及垂直整合且可延展的技術，確實深深引起客戶共鳴。我對接下來的上行趨勢感到無比振奮。

With that, I'll pass the call over to Claire to discuss our financial results.

接下來，我把電話交給 Claire，請她說明我們的財務結果。

Thank you, RJ, and good afternoon, everyone. As RJ shared, we started external customer deliveries of R2 in June. My family took a road trip in R2 earlier this summer, and we put the storage capacity to the test by piling in four large suitcases, backpacks, golf clubs, tennis rackets and groceries. Road tripping with universal hands-free is outstanding, making long car rides relaxing. I'm still looking forward to the launch of point-to-point capabilities later this year.

謝謝你，RJ，各位下午好。如 RJ 所分享，我們在 6 月開始向外部客戶交付 R2。今年夏天稍早，我的家人開著 R2 進行公路旅行；我們把置物空間徹底測試了一番，塞進四個大型行李箱、背包、高爾夫球桿、網球拍以及雜貨。使用通用免手動功能進行公路旅行的體驗非常出色，讓長途乘車變得更放鬆。我仍然很期待今年稍晚推出的點到點能力。

As I discussed previously, delivering a strong 2026 exit rate for R2 production and deliveries is a key focus for our team as we believe it will directly translate into positive automotive gross profit for the business.

如我先前所述，為 R2 的生產與交付在 2026 年底達成強勁的退出速率（exit rate），是我們團隊的重點工作；我們相信這將直接轉化為公司汽車業務的正向毛利。

Turning to the results for the second quarter. Our consolidated revenue was $1.66 billion, a 27% increase over the same quarter last year. Consolidated gross profit was $179 million, and our gross margin was 11%. Gross profit included $138 million of depreciation and amortization and $31 million of stock-based compensation expense. Adjusted EBITDA losses for the quarter were $379 million, driven by our $179 million of gross profit and increased adjusted operating expenses as we prepare to scale R2 and invest in our autonomy road map.

接著說明第二季的業績。我們的合併營收為 16.6 億美元，較去年同期成長 27%。合併毛利為 1.79 億美元，毛利率為 11%。毛利包含 1.38 億美元的折舊與攤銷，以及 3,100 萬美元的以股份為基礎之薪酬費用。本季調整後 EBITDA 虧損為 3.79 億美元，主要反映 1.79 億美元的毛利，以及在我們準備擴大 R2 規模並投資自動駕駛路線圖之際，調整後營運費用增加。

In the second quarter, we produced 12,613 vehicles at our plant in Normal, Illinois, and delivered 12,194 vehicles. Delivery results topped our 9,000 to 11,000 vehicle expectations due to robust growth quarter-over-quarter in EDV and R1, coupled with the introduction of R2 deliveries. Our $1.14 billion of automotive revenue, a 23% increase over the same quarter last year were primarily driven by a 14% increase in vehicle deliveries and $103 million increase in revenues related to automotive regulatory credits, partially offset by a decline in average selling prices from a higher mix of commercial land and R2 deliveries.

在第二季，我們在伊利諾伊州諾默爾（Normal）的工廠生產了 12,613 輛車，並交付了 12,194 輛車。由於 EDV 與 R1 的季度環比強勁成長，加上開始導入 R2 交付，交付成果超出我們原先預期的 9,000 至 11,000 輛。我們的汽車營收為 11.4 億美元，較去年同期成長 23%，主要由車輛交付量增加 14% 以及與汽車法規信用額度相關的營收增加 1.03 億美元所帶動，但部分被因商用車與 R2 交付占比提高而導致的平均售價下滑所抵銷。

Automotive gross profit loss was $36 million compared to a gross profit loss of $335 million for the same quarter last year, a $299 million improvement, primarily due to increases in delivery and production volumes and increase in revenues related to automotive regulatory credits and an IEEPA tariff refund receivable, partially offset by the ramp of R2 production. In the quarter, we recognized approximately $100 million in incremental cost of revenue due to the ramp of R2 production as compared to the production at more normalized levels.

汽車毛利虧損為 3,600 萬美元，相較去年同期毛利虧損 3.35 億美元，改善 2.99 億美元，主要由交付與產量提升、與汽車法規信用額度相關的營收增加，以及 IEEPA 關稅退稅應收款所帶動，但部分被 R2 產能爬坡所抵銷。本季相較於較為正常化的產量水準，我們因 R2 產能爬坡而額外認列約 1 億美元的營收成本增量。

While current macro and geopolitical factors are creating added complexity, cost and uncertainty, our team continues to work hard to manage supply chain risks and increasing commodity and memory costs.

儘管目前的總體經濟與地緣政治因素帶來額外的複雜性、成本與不確定性，我們的團隊仍持續努力管理供應鏈風險，以及大宗商品與記憶體成本上升的壓力。

Our Software & Services segment reported another strong quarter. During the second quarter, the segment generated $515 million of revenue, a 37% year-over-year increase, primarily due to an increase in vehicle electrical architecture and software development services by the joint venture, repairs and maintenance and Autonomy Plus, offset by lower remarketing sales. $8 million or 60% of software and services revenue was attributable to our joint venture with Volkswagen Group. Software & Services gross profit was $215 million, a 42% margin due to higher vehicle electrical architecture and software development services by the joint venture.

我們的軟體與服務（Software & Services）部門再度繳出強勁的一季。第二季該部門創造 5.15 億美元營收，年增 37%，主要來自合資企業提供的車輛電氣架構與軟體開發服務增加、維修與保養，以及 Autonomy Plus 的成長，但部分被較低的再行銷（remarketing）銷售所抵銷。軟體與服務營收中有 800 萬美元（或 60%）可歸因於我們與福斯集團（Volkswagen Group）的合資企業。軟體與服務毛利為 2.15 億美元，毛利率 42%，主要受惠於合資企業提供的車輛電氣架構與軟體開發服務增加。

Looking at our balance sheet, we ended the quarter with approximately $5.3 billion of cash, cash equivalents and short-term investments. In July, we sold 86.25 million Class A shares in a follow-on equity offering to raise approximately $1.3 billion for general corporate purposes, including the funding of equity commitments and reserves for our $4.5 billion Department of Energy loan associated with the construction of our Georgia plant.

從資產負債表來看，我們在季末持有約 53 億美元的現金、約當現金與短期投資。7 月，我們在後續增發（follow-on）股權發行中出售 8,625 萬股 A 類普通股，募集約 13 億美元，用於一般公司用途，包括履行股權承諾，以及為與喬治亞州工廠建設相關的 45 億美元美國能源部貸款提供資金與準備金。

Additionally, later this year, we expect to receive $1 billion in nonrecourse debt from Volkswagen Group and an additional $250 million equity investment from Uber, both subject to the completion of certain conditions. Our total available liquidity and targeted future capital is over $14 billion to support Rivian's investments in growth initiatives. This includes current liquidity, the Department of Energy loan and additional targeted equity investments from Uber and Volkswagen Group, which are all subject to certain conditions.

此外，我們預期在今年稍晚可自福斯集團取得 10 億美元的無追索權債務（nonrecourse debt），並自 Uber 取得額外 2.5 億美元的股權投資，兩者皆須在若干條件完成後方可實現。我們目前可用的總流動性與目標未來資金合計超過 140 億美元，用以支持 Rivian 在成長計畫上的投資。其中包括現有流動性、美國能源部貸款，以及 Uber 與福斯集團額外的目標股權投資，以上皆須符合若干條件。

Finally, for our 2026 guidance, in early July, we announced we expect to deliver 65,000 to 70,000 total vehicles across R1, R2 and our commercial van, a 3,000 unit increase as compared to our prior guidance. This implies approximately 42,400 to 47,400 deliveries in the second half of the year. We expect deliveries will be weighted towards the fourth quarter as we ramp R2.

最後，關於我們對 2026 年的指引，我們在 7 月初宣布，預期 R1、R2 與商用廂型車合計交付 65,000 至 70,000 輛，較先前指引上調 3,000 輛。這意味著下半年約交付 42,400 至 47,400 輛。隨著我們推進 R2 產能爬坡，預期交付將更偏重於第四季。

While we continue to believe our gross profit will increase year-over-year, we expect the complexity of a new vehicle launch will negatively impact our automotive gross profit in the third quarter as it did in the second quarter before becoming a benefit for our overall operations in the fourth quarter as we ramp production and deliveries. As a reminder, we believe this is a transition year for the automotive segments path towards long-term profitability as we scale R2.

雖然我們仍相信毛利將年增，但我們預期新車上市的複雜性將在第三季對汽車毛利造成負面影響，正如第二季所見；然而，隨著第四季產量與交付爬坡，這將轉而成為整體營運的助益。提醒一下，我們認為在擴大 R2 規模的同時，2026 年是汽車部門邁向長期獲利能力道路上的過渡年。

For 2026, we expect an adjusted EBITDA loss of between $2 billion to $1.8 billion, a $50 million improvement at the midpoint due to better-than-expected revenue related to regulatory credits in the second quarter and increased delivery volumes, which were partially offset by raw material, memory and logistic cost increases.

對於 2026 年，我們預期調整後 EBITDA 虧損介於 20 億至 18 億美元之間，中位數較先前改善 5,000 萬美元，主要因第二季與法規信用額度相關的營收優於預期，以及交付量增加；但部分被原物料、記憶體與物流成本上升所抵銷。

While economic and geopolitical conditions, including international conflicts pose risks, we remain steadfast in our plans to invest behind key growth drivers. We continue to progress our autonomy road map and the expansion of our sales and service footprint as we scale with R2. We believe these strategic investments will deliver long-term value to our shareholders and customers.

儘管經濟與地緣政治情勢（包括國際衝突）帶來風險，我們仍堅定依計畫投入關鍵成長動能。在隨 R2 擴張規模的同時，我們持續推進自動駕駛（autonomy）路線圖，以及擴大銷售與服務據點。我們相信這些策略性投資將為股東與客戶帶來長期價值。

Finally, for 2026, we are reducing our capital expenditure guidance by $250 million at the midpoint to $1.7 billion to $1.8 billion, benefiting from project efficiencies and timing of spend. Our CapEx spend is primarily related to finalizing construction and tooling for R2 in Normal, the continued build-out of our sales, service and charging infrastructure and construction for our greenfield plant in Georgia.

最後，針對 2026 年，我們將資本支出（CapEx）指引在中位數下調 2.5 億美元至 17 億至 18 億美元，受惠於專案效率提升與支出時點安排。我們的資本支出主要用於在諾默爾完成 R2 的建設收尾與工裝治具（tooling）、持續擴建銷售、服務與充電基礎設施，以及喬治亞州綠地（greenfield）工廠的建設。

In closing, I'd like to congratulate our team again for the start of external R2 sales in June and the continued strong execution in the second quarter. We continue to believe that R2 and our technology road map will be truly transformative for the growth and profitability of our business.

在結語中，我想再次祝賀我們的團隊於 6 月啟動 R2 的對外銷售，以及第二季持續強勁的執行表現。我們仍相信 R2 與我們的技術路線圖將真正為公司成長與獲利能力帶來變革性的影響。

I'd like to turn the call back over to the operator to open the line for Q&A.

我想把電話交回給接線員，開放問答。

(Operator Instructions)

（接線員指示）

Mark Delaney, Goldman Sachs.

高盛（Goldman Sachs）的 Mark Delaney。

Congratulations on the R2 launch. Very exciting to see. I was hoping to start there on R2, can you share more on how the production ramp is progressing? And any examples or metrics maybe you can share on that front to illustrate how the ramp is going? And then as you think about R2 over the next year, think the pace of deliveries is going to be gated more by a supply or demand?

恭喜 R2 上市。看到這個消息非常令人振奮。我想先從 R2 談起，能否分享更多關於產能爬坡進展的情況？以及是否有任何例子或指標可以用來說明爬坡進度如何？另外，當你們看待未來一年 R2 的表現時，你們認為交付節奏更可能受供給端還是需求端所限制？

Well, thanks, Mark, for the question. We're also very excited about R2. Well, I guess, first on the production ramp. Of course, ramping production of a vehicle, especially where it's a first-time launch of a new vehicle is complex. It's an orchestration of ramping hundreds of suppliers, the suppliers all need to ramp at the same rate.

Mark，謝謝你的提問。我們也對 R2 感到非常興奮。我想先談產能爬坡。當然，車輛量產爬坡，尤其是首次推出全新車款時，過程相當複雜。這需要協同數百家供應商一起爬坡，而且所有供應商都必須以相同的速度提升產能。

Of course, our overall production output is throttled by our gated by the slowest moving supplier. And so the coordination to make sure everything is ramping consistently and in an organized fashion across all the different suppliers is really key focus for us. And it's the reason we start on a single shift and get everything moving efficiently and effectively before bringing on a second shift.

當然，我們的整體產出會受到、或說被最慢的供應商所限制。因此，確保所有不同供應商都能一致且有組織地同步爬坡的協調工作，是我們真正的重點。這也是我們先從單班制開始，讓一切能高效率、有效率地運轉之後，再導入第二班的原因。

So saying that, this is a complex effort. There's lots of moving parts, and we went into it fully recognizing the complexity and having been through this with R1 and iterations in R1 with a lot of learnings and knowledge of a business that we certainly didn't have to this level with the launch of R1. And so where it sense today is we're -- we are ramping.

因此可以說，這是一項複雜的工作。其中有許多變動因素；我們在投入時已充分認知其複雜性，並且先前在 R1 以及 R1 的多次迭代中累積了大量學習與知識，而這些在 R1 首次上市時我們並沒有達到同樣的程度。所以就目前而言，我們——我們正在爬坡。

We're continuing to make progress week over week. We've -- as you saw, we've updated our guidance to reflect the progress we're making. And the biggest category of focus for us has been supply base and making sure that all of our suppliers are ready to continue ramping to much higher levels as we look at the back half of the year. And in particular, as we start to plan for running -- fully utilizing the plant and running across two shifts.

我們持續每週都有進展。正如你所看到的，我們已更新指引以反映目前取得的進展。而我們最大的重點領域一直是供應體系，確保所有供應商都已準備好在下半年持續爬坡到更高的水準。尤其是在我們開始規劃——充分利用工廠產能並以兩班制運轉之際。

Now on the demand side, there's an enormous excitement around the product. And we've been really encouraged by the conversion of reservations to orders on our launch addition. We, of course, had our own internal projections as to what that would be, recognizing that we're launching with the launch addition, which is a $58,000 version of the vehicle, and most expensive version of R2.

現在從需求端來看，市場對這款產品有著極大的熱情。而且我們也因為在上市版（launch edition）上，預訂轉為訂單的轉換率而深受鼓舞。當然，我們內部對此有自己的預估，並且也認知到我們是以上市版作為首發，這是一款售價 58,000 美元的車型，也是 R2 中最昂貴的版本。

And that conversion rate from reservation to orders has been meaningfully higher than what we expected, which is a great signal. Now of course, with that said, we are looking forward to introducing the other trims our mid-spec and our base or standard spec as we look at the early part of 2027. But the excitement even for the launch addition and the higher price finance addition has been very, very high.

而從預訂到訂單的轉換率顯著高於我們的預期，這是一個非常好的訊號。當然，話雖如此，我們也期待在 2027 年初推出其他車型配置，也就是我們的中階規格以及入門或標準規格。但即便是對上市版以及更高價格的融資版（finance edition），市場熱度也一直非常、非常高。

Appreciate all those details. My other question was trying to understand what the R2 ramp might mean for the financials and putting that into context with some of the other comments around the input cost environment. So as you look out through the end of the year and start ramping R2, does Rivian still expect to reach a positive gross margin exiting 2026?

感謝分享這些細節。我另一個問題是想了解 R2 的爬坡（ramp）對財務可能意味著什麼，並把它放在你們對投入成本環境的其他評論脈絡中來看。因此，當你們展望到今年年底並開始拉升 R2 產量時，Rivian 是否仍預期在 2026 年底時達到正的毛利率？

Thanks, Mark. As I mentioned in my prepared remarks, we continue to expect that R2 will achieve a positive gross profit as part of our exit rate for 2026. And that's enabled, as RJ mentioned, by the ongoing efficiencies of being able to produce R2 at much higher volumes and the fixed cost leverage that, that will enable within the business.

謝謝你，Mark。如同我在事先準備的發言中提到的，我們仍然預期 R2 將在 2026 年的退出速率（exit rate）中實現正的毛利（gross profit）。而這一點，如 RJ 所提到的，是由於我們能以更高的產量生產 R2 所帶來的持續效率提升，以及這將使業務得以運用固定成本槓桿。

George Gianarikas, Canaccord Genuity.

George Gianarikas，Canaccord Genuity。

I wonder to focus a little bit on the EV. Can you provide an update maybe on commercial adoption beyond Amazon? And as you expand pilot programs for non-Amazon fleets, what's the current primary governor on deployment pace? Is it TCO parity, upfront capital outlay, charging density, anything you could share there?

我想把焦點稍微放在電動車（EV）上。你們能否提供一下 Amazon 以外的商用採用進展更新？以及當你們擴大非 Amazon 車隊的試點計畫時，目前影響部署速度的主要限制因素是什麼？是總持有成本（TCO）達到平價、前期資本支出、充電密度，或其他你們可以分享的因素嗎？

Thanks, George. I think the first point to just call out is the ramp-up that we're seeing within Amazon. And so you're seeing this already in the numbers for the first two quarters of this year that Amazon is purchasing more vans. We're ramping up meaningfully from where we were in previous years. And that's absolutely a reflection of the TCO advantages that our EDV platform represents and enables.

謝謝你，George。我想首先要指出的是，我們在 Amazon 內部看到的爬坡。因此你已經可以在今年前兩季的數字中看到，Amazon 正在購買更多廂型車。我們相較於前幾年正在顯著加速提升。而這完全反映了我們 EDV 平台所代表並帶來的 TCO 優勢。

And we think that those cost benefits that are being so clearly realized by Amazon and so clearly evidenced by the ramping up of volume, we think that's going to translate to other commercial operators ultimately seeing those cost benefits and recognizing both to remain competitive, but also to address their own cost structures that they're going to be pursuing opportunities with EDV.

我們認為，Amazon 如此清楚地實現這些成本效益，並且透過出貨量的提升清楚地展現出來，最終也會轉化為其他商用營運商看到這些成本效益；他們為了保持競爭力，同時也為了改善自身成本結構，將會尋求採用 EDV 的機會。

Now that has taken longer than we had would have expected or we would have hoped for in terms of other fleets adopting a strategy towards electrification and a strategy of adopting not just electric vehicle, but a highly connected intelligent vehicle. But the signal that's being sent through Amazon's significant ramping of this is really outstanding in terms of a billboard for the capability and a billboard for what the EDV is possible delivering from a business point of view.

就其他車隊在電動化策略、以及採用不只是電動車而是高度連網的智慧車輛之策略而言，這件事花的時間比我們原先預期或希望的更久。但 Amazon 顯著的爬坡所釋放出的訊號非常出色，從能力展示的角度來看就像一面大型看板，也像是一面展示 EDV 從商業角度能帶來什麼的看板。

And maybe more of a long-term question. As you continue to benchmark other global EV platforms, particularly from Chinese OEMs, what structural or operational lessons can you apply to the R3 and other future vehicles as you spec those out and try to build them?

再問一個可能更偏長期的問題。當你們持續對標其他全球電動車平台，特別是來自中國 OEM 的平台時，在你們為 R3 與其他未來車款做規格定義並嘗試打造它們時，有哪些結構性或營運面的經驗教訓可以應用？

Of course, the auto industry is a business where every manufacturer is very aware of what our other manufacturers are doing. So Rivian are purchased by probably every car comp of the world and taken apart. And similarly, the Chinese vehicles are purchased by many different -- both OEMs, but also benchmarking companies.

當然，汽車產業是一個每家製造商都非常清楚其他製造商在做什麼的產業。所以 Rivian 的車大概被全球每一家車廠買去拆解研究。同樣地，中國車也被許多不同的——不僅是 OEM，也包括對標研究公司——買去做拆解對標。

And so as a result, there's not really -- there's not like a magic difference in terms of how a Chinese vehicle is built relative to a vehicle built in the West in terms of the manufacturing approaches, the joining technologies, the use of high-pressure die castings, consolidation of stampings. Those methods are being deployed across all, call it, best-in-class vehicles from a manufacturability point of view and a vehicle architecture point of view.

因此，從製造方法、接合技術、高壓壓鑄的使用、沖壓件整合等角度來看，中國車相對於西方製造的車，並不存在什麼「神奇差異」。從可製造性與車輛架構的角度來看，這些方法正被部署在所有可稱為同級最佳（best-in-class）的車款上。

We see that, of course, evidenced in how we're building the R2, use of high-pressure die casting is an extreme focus on part consolidation and part elimination. I think the big difference that exists between US cost structure and Chinese cost structure is the input cost. So you have a much, much lower labor cost in China.

我們當然也在 R2 的打造方式上看到這一點：使用高壓壓鑄，並且極度聚焦於零件整合與零件消除。我認為美國成本結構與中國成本結構之間最大的差異在於投入成本（input cost）。也就是中國的勞動成本要低得多、低得多。

Much, much, much lower capital cost structure where, in many cases, the capital cost is zero, meaning it's being provided by the local government. And that compounds across the supply chain and all the way up through the OEM, and it results in a materially lower cost than what it would take to produce in the Western world.

資本成本結構也低得多、低得多、低得多；在許多情況下，資本成本甚至為零，意思是由地方政府提供。而這種差異會在供應鏈中層層疊加，一路傳導到 OEM，最終導致其成本實質上遠低於在西方世界生產所需的成本。

And so that naturally opens the question around trade policy. And importantly, it opens very complex questions around supply chain strategy. And we're navigating a lot of those complexities around sourcing our components and our -- not just our Tier 1, but even down to the raw materials, where there are coming from recognizing the frameworks that have been established around trade.

因此這自然會引出貿易政策的問題。更重要的是，這也引出非常複雜的供應鏈策略問題。我們正在因應許多這些複雜性，包含我們的零組件採購——不只是 Tier 1，甚至一路到原物料——它們來自哪裡，同時也要考量既有的貿易框架。

And in a world of completely open trade, it would look -- our supply chains would look very different because we'd be optimizing purely around cost, which would be, of course, optimizing around countries that have the lowest input cost structure, be it labor cost, lowest energy cost, lowest land cost, lowest cost of capital.

在一個完全自由貿易的世界裡，我們的供應鏈會看起來非常不同，因為我們會純粹以成本最佳化為目標；而那當然會圍繞投入成本結構最低的國家來最佳化，無論是勞動成本最低、能源成本最低、土地成本最低，或資本成本最低。

And of course, that's not the world we work in. We work in a world where there's very intentional industrial policies that exist that changed the way we might be able to look at this. And in many cases, there's implications around certain components where we believe it's very helpful to make sure those are sourced from the United States.

但當然，我們所處的並不是那樣的世界。我們運作在一個存在非常明確的產業政策的世界裡，而這會改變我們可能看待這些問題的方式。而在許多情況下，某些零組件會牽涉到一些影響，我們認為確保它們在美國採購是非常有幫助的。

Shreyas Patil, Wolfe Research.

Shreyas Patil，Wolfe Research。

I was curious if you could maybe unpack the $50 million increase in adjusted EBITDA for the year. How much of that is tied to the better regulatory credit sales? And then how should we think about input cost inflation?

我想了解你們今年調整後 EBITDA 指引上調 5,000 萬美元的原因，能否請你們拆解一下？其中有多少與更好的法規信用額度（regulatory credit）銷售有關？另外，我們應該如何看待投入成本通膨？

Sure, Shreyas. As you look at the evolution of our EBITDA guidance, so we were able to improve the bottom end of the range by $100 million that moves the midpoint $50 million, as you stated. We had higher-than-anticipated regulatory credits in the first half of this year. And then as we look ahead, we have the added benefit of the incremental delivery volumes with the increase of 3,000 units in our outlook. But offsetting that, we have headwinds associated with increases in raw material costs, increases in memory costs as well as increases in logistics costs that had us arrive at this specific EBITDA range and outcome.

當然可以，Shreyas。就我們 EBITDA 指引的演變來看，我們把區間下緣改善了 1 億美元，這也如你所說，使中位數上移 5,000 萬美元。我們在今年上半年取得了高於預期的法規信用額度。接著展望未來，隨著我們展望中交付量上調 3,000 輛，我們也有額外的增量交付量帶來的好處。但與此相抵，我們面臨原物料成本上升、記憶體成本上升，以及物流成本上升等逆風，因此我們得出這個特定的 EBITDA 區間與結果。

And so as we look at the second half performance relative to the first half, what you can anticipate seeing is on the gross profit side, as I mentioned in my prepared remarks, we'll continue to see some of the inefficiencies and complexities associated with the R2 ramp impact Q3 automotive gross profit, and then we'll start to see some of the tailwinds of the scale and volumes kick in, in the fourth quarter.

因此，當我們比較下半年相對於上半年的表現時，你可以預期在毛利（gross profit）方面，如同我在事先準備的發言中提到的，我們仍會看到與 R2 爬坡相關的一些效率不彰與複雜性，對第三季汽車毛利造成影響；接著在第四季，我們將開始看到規模與產量帶來的一些順風開始顯現。

And then beyond that, we'll see just the ongoing investments that we're continuing to make across R&D and specifically the growth in our autonomy spend and then the continued fixed cost leverage that we'll see as we're growing our sales and service footprint, but certainly growing sales much more rapidly as well.

此外，接下來我們會看到我們持續在研發上的投資，特別是自動駕駛支出的成長；同時，隨著我們擴大銷售與服務據點，我們也會持續看到固定成本槓桿的提升，而且銷售的成長速度也會更快。

Okay. And then maybe just on the Autonomy Plus. So you're going to be launching the point-to-point solution later this year. I'm curious where things stand today with Autonomy Plus. Just if there's a way you can give us a sense of take rates or kind of adoption trends. And then how do you see point-to-point driving both demand, but then also those take rates? So ultimately, the subscription revenue opportunity.

好的。接著也許談一下 Autonomy Plus。你們將在今年稍晚推出點到點（point-to-point）解決方案。我想了解 Autonomy Plus 目前的進展。如果可以的話，能否讓我們大致了解一下選配率（take rate）或採用趨勢。另外，你們如何看待點到點功能同時推動需求，以及推升這些選配率？也就是最終的訂閱收入機會。

We've been really encouraged by the take rates on Autonomy Plus to date. And I think, in particular, what's really encouraging is to see the status that we're at there in terms of customer engagement, customer usage of the future and then, of course, related to that, the take rate around paying for Autonomy Plus. But we have very clear line of sight and visibility into what's coming.

截至目前，我們對 Autonomy Plus 的選配率感到非常振奮。我認為特別令人鼓舞的是，我們在客戶互動、客戶對未來功能的使用情況，以及當然與此相關的 Autonomy Plus 付費選配率方面，所處的狀態。不過，我們對接下來將發生的事情有非常清晰的視野與能見度。

And so as encouraged as we are with the take rates today, we're quite confident that, that will expand meaningfully as we add in additional capabilities and features. And there's going to be a whole -- think of it as like a staircase of features that ultimately leads us to very high levels of autonomy, but a really important step that's happening in the very near future is the capability of point-to-point.

因此，儘管我們對目前的選配率感到鼓舞，我們也相當有信心，隨著我們加入更多能力與功能，選配率將會顯著擴大。而且會有一整套——可以把它想成一段段階梯式的功能——最終帶我們走向非常高的自動化水準；但在非常近期即將發生的一個關鍵步驟，就是點到點能力。

And just to restate what that entails, that means you can get into the car, type the address into the vehicle and the vehicle completely drive itself to that address. And initially, that will be a feature in which it's like Level 2++ where your hands are off the wheel, but your eyes are still primarily on the road, meaning you can look away for short extended moments of time, but you're not in a position to officially be reading a book or be responding to e-mails.

再重申一下這代表什麼：你可以坐進車裡，把地址輸入到車輛中，車輛就能完全自行駕駛到該地址。一開始，這會是一項類似 Level 2++ 的功能：雙手可離開方向盤，但眼睛仍需主要看著路面，也就是你可以在短暫、稍長的片刻把視線移開，但你還不能正式地讀書或回覆電子郵件。

Now this step in 2027 is going further where we go, not only hands off, but also eyes off, and that's really important because then not only is the vehicle fully driving itself and navigating to the address, but importantly, you're able to truly get your time back without the vehicle dinging it, you say, to keep your eyes on the road, but actually be able to read a book, be on an iPad. And so that's really into the like true as per the definition, Level 3 capability.

接著在 2027 年，這一步會更進一步：不僅雙手離開，連視線也可以離開，這非常重要，因為那時不只是車輛完全自行駕駛並導航到目的地，更重要的是，你真的能把時間拿回來，不會被車輛提醒你要把眼睛放在路上，而是可以真正讀書、用 iPad。因此這就更接近真正符合定義的 Level 3 能力。

And then, of course, into 2028, that's when we bring on Level 4. And that's where the vehicles capable of operating as a robotaxi, meaning there's no one even needed to be in the driver seat for any operation of the vehicle, and you can put occupants in the back. You can use that capability for robotaxi applications. But you can also have that, which is one of the things we've been very excited about, have that as a really nice feature on an owned vehicle.

然後當然到了 2028 年，我們會導入 Level 4。那時車輛就具備作為機器人計程車（robotaxi）運行的能力，也就是車輛運作不需要任何人坐在駕駛座，你可以讓乘客坐在後座。你可以把這項能力用於 robotaxi 應用。但你也可以把它——這也是我們一直非常興奮的一點——作為自有車輛上的一項非常棒的功能。

And so we think of this as personal Level 4, where your own personal vehicle can drop you at the airport, it can pick your kids up from school. It can go to the grocery store and grab things for you. And that road map is something we're not only very excited about, but we think will fundamentally drive increasing levels of adoption on Autonomy Plus and its future variances we've just described.

因此我們把它視為「個人 Level 4」，也就是你自己的車可以送你到機場、可以去學校接小孩。它可以去雜貨店幫你買東西。這條路線圖不僅讓我們非常興奮，我們也認為它將從根本上推動 Autonomy Plus 以及我們剛描述的未來各種版本的採用率持續提升。

Now with all that said, we also think we're at an inflection point from a consumer behavior and consumer mindset, where historically, autonomous features were not the primary purchase criteria for most customers. Of course, there's technology early adopters for which this has been top of mind for them for some time. But for the vast majority of customers, this wasn't the driving purchase criteria.

在上述前提下，我們也認為消費者行為與心態正處於一個拐點：過去，自動駕駛功能並不是大多數客戶的主要購買標準。當然，有些科技早期採用者很早就把這件事放在心上。但對絕大多數客戶而言，這並不是驅動購買的關鍵標準。

We think this starts to emerge to be much more of a primary purchase criteria. And it's one of the reasons we've made the decision to invest so heavily here. We think ultimately, market share is going to be driven by the capability of the vehicle to drive itself. And of course, linked to that the vehicle being software and AI defined. And those are two very like critical pillars of our technology road map and technology strategy.

我們認為這將開始逐漸成為更主要的購買標準。這也是我們決定在這裡投入如此重資本的原因之一。我們最終認為，市占率將由車輛自我駕駛的能力所驅動。當然，與此相連的是車輛由軟體與 AI 所定義。而這兩點是我們技術路線圖與技術策略中非常關鍵的兩大支柱。

Rajat Gupta, JPMorgan.

Rajat Gupta，摩根大通（JPMorgan）。

I had a question on the R2 customer profile. Any early reads on what that customer is proving to be existing electric vehicle owners, first-time electric wafer buyers, best lacrosse shoppers? Just curious if you have a read on the customer profile of the early R2 orders, and I have a quick follow-up.

我有一個關於 R2 客戶輪廓的問題。對於這些客戶是否為既有電動車車主、首次購買電動車的買家、或是跨級距（best-in-class）比價的消費者，有沒有任何初步觀察？想請教你們對早期 R2 訂單的客戶輪廓是否已有判讀；我還有一個簡短的追問。

Thanks, Rajat. The goal in developing and designing R2 is to have a very diverse set of customer profiles, meaning customers that were coming out of a wide variety of form factors coming out of a wide range of brands and from a broad spectrum of demographics. And that was the intent, that was the target. We now have the benefit of being able to see the data to see who is actually buying R2s. And the great news here is that it is exactly that.

謝謝你，Rajat。在開發與設計 R2 時，我們的目標是涵蓋非常多元的客戶輪廓，也就是客戶來自各式各樣的車身型式、各種品牌，以及廣泛的人口統計族群。這就是初衷，也是目標。現在我們有機會看到數據，了解實際上是哪些人在買 R2。而好消息是，結果正是如此。

We have a very diverse customer set coming out of a broad range of vehicles, which I guess perhaps isn't -- shouldn't be too big or a surprise, the midsize SUV segment, the 2x5 passenger segment is the largest segment by far in the US market. It's generally the largest segment globally as well.

我們的客戶群非常多元，來自各種不同的車款；我想這或許也不——不應該太令人意外，因為中型 SUV 級距、2x5 乘坐級距是美國市場迄今最大的級距。一般而言，它在全球也通常是最大的級距。

But the way that we've been able to really thoughtfully execute this vehicle in terms of packaging, storage capacity, performance, range, efficiency driving dynamics, both on-road and off-road has created a very wide net for attracting this very diverse set of buyers.

但我們在車輛的整體規劃上——包括空間配置、儲物容量、性能、續航里程、效率、操控動態（無論公路或越野）——都做了非常周到的執行，這讓我們能以非常廣的覆蓋面吸引到這一群高度多元的買家。

And maybe one of the most exciting metrics here is we have a significant number of first-time EV owners, and that's really powerful. So for someone to have their first time owning an EV be Rivian. And for that bond with the brand, so to speak, to be created through that first-time ownership is powerful. But it also illustrates the strength of what we've created in terms of the product and the product market fit that this is mobilizing demand for EVs that's been latent, that's been sitting on the sideline because the product that sort of would motivate someone to move from an ICE to EV hasn't existed until R2.

而且這裡可能最令人興奮的指標之一是：我們有相當多的首次 EV 車主，這非常有力。也就是說，有人第一次擁有電動車就是 Rivian。而且可以說，與品牌之間的連結是透過這次首次擁有而建立起來的，這很有力量。同時這也顯示我們在產品與產品市場契合度（product-market fit）上所打造的強度：R2 正在把原本潛在、一直觀望的 EV 需求動員起來，因為在 R2 之前，能夠真正促使人們從燃油車（ICE）轉向 EV 的產品並不存在。

And so this is a great metric. Now of course, there's -- there are EV buyers that are in the mix. You have folks coming in a Tesla or folks going out of Rivian R1s, but the most important metric here is the very significant number of folks that have not owned EVs before.

因此這是一個很棒的指標。當然，也有——也確實有 EV 買家在其中。你會看到有人從 Tesla 轉進來，或有人從 Rivian R1 轉換，但這裡最重要的指標是：有非常顯著的一群人以前從未擁有過 EV。

That's great color. And just a follow-up for Claire, maybe. I noticed in the second quarter gross profit commentary, you mentioned on IEEPA tariff refund benefit. I'm curious how much that was because it didn't see it in the full year outlook. And then just relatedly, it looks like the guidance implies a steeper EBITDA loss in the second half versus first half despite the positive gross margin exit rate commentary. Just trying to bridge those two aspects.

這些資訊很有幫助。另外我想追問 Claire 一下。我注意到你在第二季毛利的說明中提到 IEEPA 關稅退稅（tariff refund）帶來的利益。我想了解那大概是多少，因為我在全年展望裡沒有看到。另外相關地，看起來指引暗示下半年 EBITDA 虧損會比上半年更深，儘管你們提到毛利率退出水準（exit rate）為正。我只是想把這兩個面向串起來理解。

Sure. On the IEEPA tariff refund piece, as we mentioned in our earnings call last quarter, we size the IEEPA tariffs in the tens of millions of dollars. The receivable that we booked this quarter is roughly half of the total. And so the receivable is related to the refund of the amount of anticipated proceeds that are not subject to the existing DOJ appeal.

當然。關於 IEEPA 關稅退稅的部分，正如我們在上一季財報電話會議中提到的，我們估算 IEEPA 關稅金額為數千萬美元等級。我們本季入帳的應收款大約是總額的一半。因此，這筆應收款與預期可收回款項中、不受現有 DOJ（美國司法部）上訴影響的那一部分退稅相關。

And then as you think about the second question on the EBITDA loss being steeper in the second half of the year. One of the big differences there is we did have $164 million of regulatory credits that benefited our gross profit outlook in the first half.

接著談到第二個問題，也就是為何下半年 EBITDA 損失會更陡峭。其中一個很大的差異是，上半年我們有 1.64 億美元的監管信用額度（regulatory credits），對我們的毛利展望帶來助益。

And so as we look at the second half of the year, we'll see, again, some of the complexity associated with the R2 ramp impacting automotive gross profit in the third quarter before showing progress in the fourth quarter. But the bigger driver is really the absence of regulatory credits in second half versus first half and then the ongoing investments in R&D and SG&A.

因此，當我們看下半年時，我們會再次看到與 R2 產能爬坡相關的一些複雜性，在第三季對汽車毛利造成影響，並在第四季展現改善。但更大的驅動因素其實是：下半年相較上半年缺少監管信用額度，以及我們在研發（R&D）與銷售、一般及行政費用（SG&A）上的持續投資。

Itay Michaeli, TD Cowen.

Itay Michaeli，TD Cowen。

Great. Just first, a couple of questions on the R2. RJ, 57,000 demo drives. It's a big jump sequentially Hoping you could talk a little bit about what you're seeing in terms of conversion of those drives into orders? How many people are actually experiencing autonomy plus? And then secondly, just curious whether you do expect to deliver a material amount of the premium trends this year or R2 sales just predominantly be the performance trend this year?

很好。首先，關於 R2 有幾個問題。RJ，5.7 萬次試駕。這個數字較前一季大幅跳升，希望你能談談你們看到這些試駕轉換成訂單的轉換情況如何？實際上有多少人正在體驗 Autonomy Plus？第二個問題是，想請教你們是否預期今年會交付相當數量的 Premium 車型，還是今年 R2 的銷售主要會以 Performance 車型為主？

Yes. I mean we're really excited, as I said in my opening remarks, the 57,000 drives this past quarter is a record for Rivian. And so just the delivery of that, just enabling that number of drivers you said roughly doubling where we've been in the past. The team has had to really come together and we do a lot of prep knowing that the amount of excitement on R2 is going to drive a significant increase in both exposure to the brand, touchpoints of the brand, but of course, that's manifested through these test drives.

是的。我的意思是，我們真的非常興奮，正如我在開場致詞中提到的，上季 5.7 萬次試駕是 Rivian 的紀錄。因此，能夠交付、也就是促成你所說的這麼多駕駛者參與，約等於把我們過去的水準翻倍。團隊必須真正凝聚起來；我們也做了大量準備，因為我們知道 R2 的熱度會顯著提升品牌曝光與品牌接觸點，而這當然就透過這些試駕活動具體呈現出來。

And so with that, we're not providing specifics around conversion, but that's really an important metric we look at internally. And that actually links pretty heavily to the availability of the other variants. And so to start, as you said, we have our performance brand the most expensive variant of R2 that we've launched with. We'll be introducing the premium and then our standard trims as we look at early 2027.

因此，關於轉換率我們不提供具體數字，但那確實是我們內部非常重視的一項關鍵指標。而且它其實與其他車型版本的供應可得性高度相關。如你所說，我們一開始推出的是 Performance 版本，也就是我們推出的 R2 中最昂貴的版本。我們將在 2027 年初推出 Premium，接著是 Standard 等車型等級。

But we are witnessing a higher level of conversion than we expected, meaningfully higher than we expected into the launch edition package, which was great. It's super encouraging. I think reflects just the amount of excitement around the product, but saying that we fully recognize and are excited to also be launching the other trims and know that those will be coming, as I said, at the early part of 2027.

但我們看到的轉換率高於我們的預期，且顯著高於預期，特別是轉換到 Launch Edition 套件，這很棒。這非常令人鼓舞。我認為這反映了市場對產品的熱度；同時我們也完全理解並且很期待推出其他車型等級，我們也知道它們會如我所說在 2027 年初推出。

That's great to hear. And then Claire, maybe just a quick follow-up. R&D, I think, was flat up sequentially. I know you're making a lot of investments in autonomy. Just curious how we could think about R&D kind of trending in the second half of the year.

很高興聽到。接著 Claire，可能再快速追問一下。我認為研發費用（R&D）較前一季大致持平或小幅上升。我知道你們在自動駕駛方面投入很多。想請教我們該如何看待今年下半年研發費用的走勢？

Sure. As we look at the second half of the year, we do anticipate seeing a step up and that's predominantly related to sourcing GPU for our autonomy training that will continue to ratchet up over the course of the second half of the year. In the second quarter, there, we saw some level of production as we're now in production. So there was a decrease in our engineering design and development spend, but we did underneath the R&D spend still see significant increase in terms of our overall autonomy spend.

當然。展望下半年，我們確實預期會看到研發費用上升，主要與為自動駕駛訓練採購 GPU 有關；這部分在下半年會持續逐步增加。在第二季，我們看到一定程度的量產，因為我們現在已進入量產。因此，我們的工程設計與開發支出有所下降，但在研發費用之下，我們仍看到整體自動駕駛相關支出顯著增加。

Joseph Spak, UBS.

Joseph Spak，UBS。

Just maybe the comment about the $100 million of incremental cost of revenues of ramp of R2 compared to normalized levels. Just maybe some context, like what are you classifying as normalized? Is that one shift, two shifts? And are you also assuming a more normalized mix in that analysis?

想請教你提到 R2 產能爬坡相較於常態水準，會帶來 1 億美元的額外銷貨成本（cost of revenues）。可否提供一些背景，例如你們所定義的「常態化」是什麼？是一班制、兩班制？另外，你們在這個分析中是否也假設產品組合更接近常態？

Joe, the way to think about it is the $100 million reflects a combination of ramp-related inefficiencies such as expedited freight costs out of the gates or short-term premiums with suppliers that we anticipate coming off as we increase our production volumes over the course of the next couple of quarters. And then it's also cost that we expensed in period that would have naturally been absorbed into inventory, for example, if we were at more of a Q4 run rate of production volume. So that's a little bit of the way to think about how we've sized the $100 million of estimated ramp costs.

Joe，可以這樣理解：這 1 億美元反映了多項與爬坡相關的低效率，例如一開始的加急運輸費用，或是與供應商之間的短期溢價；隨著未來幾季我們提高產量，這些成本預期會逐步消退。此外，也包含我們在當期費用化、但在更高產量下本來會自然吸收到存貨中的成本；例如，如果我們的產量水準更接近第四季的運行速率（run rate）。以上就是我們估算這 1 億美元爬坡成本的大致思路。

Okay. So yes, that's what I -- I mean, a good portion of that is just like you don't have the scale and leverage over the assets. So if you're saying second shift comes on at some point, later in the year and you're exiting gross profit positive, then that's sort of the -- that could help us sort of triangulate sort of like the units do you think you need to sort of be close to breakeven. But that's a fair way to think about it.

了解。所以是的，這就是我——我的意思是，其中很大一部分就是因為你們尚未具備規模，資產的槓桿效益也還沒出來。所以如果你們說第二班制會在今年稍晚某個時間點啟動，且你們在期末（退出時）達到毛利轉正，那麼這就——這可以幫助我們大致推算你們需要多少產量單位才能接近損益兩平。這樣理解是合理的。

That's generally a fair way to think about it. And so as we talked about it being really at sort of Q4 normalized levels, that's not the full extent of R2 fully ramped up. But is, in the near term, a more normalized production environment that we can compare our standard cost too.

大致上這樣理解是合理的。因此，當我們談到第四季的常態化水準時，那並不是 R2 完全爬坡到位的全部程度。但在短期內，這是一個更常態化的生產環境，我們可以用來與我們的標準成本（standard cost）做比較。

Okay. And then RJ, I appreciate the Autonomy Plus commentary earlier. I was wondering if you could also give us a latest update on just RAP1 and Gen 3 hardware and any progress? Is that still on track for SOP later this year?

了解。另外 RJ，我很感謝你先前對 Autonomy Plus 的說明。我想請你也更新一下 RAP1 與 Gen 3 硬體的最新進度？今年稍晚進入 SOP（Start of Production，量產啟動）是否仍按計畫進行？

Yes. Our in-house silicon RAP1 is on track. As you know, that underpins what we call our Gen 3 autonomous hardware platform. And so both the RAP1, along with the rest of the upgrades to the system are on track for later this year.

是的。我們自研的矽晶片 RAP1 進度如期。如你所知，它是我們所稱的 Gen 3 自動駕駛硬體平台的核心。因此，RAP1 以及系統其餘升級項目，都按計畫在今年稍晚推出。

Dan Levy, Barclays.

Dan Levy，Barclays。

Okay. Great. I just wanted to start with a question for you, Claire, on just unpacking the COGS per unit on Joe's question there. To only see COGS per unit up a few thousand dollars quarter-over-quarter despite that incremental $100 million says that there's probably something incrementally that's going on quarter-over-quarter on the cost side, whether it's within EV or one. So maybe if you can just help unpack if there's other underlying cost improvements and why we're not seeing maybe -- what should have been maybe a step-up in raw material pressure in that cost per unit?

好的。很好。我想先問 Claire 一個問題，延續 Joe 的問題，想更深入拆解一下每台車的銷貨成本（COGS per unit）。在有額外 1 億美元增量的情況下，每台車 COGS 季對季只增加幾千美元，這表示成本端季對季可能還有一些額外變化，不論是在電動車（EV）或其他項目。所以想請你協助拆解：是否還有其他底層的成本改善？以及為什麼我們沒有看到——原本可能預期在每台成本上會出現的原物料壓力上升？

Sure. As you look at the quarter, we had $96,700 a unit of COGS. If you were to back out the $100 million on a per unit basis, that's about $8,200 a unit that would have gotten us to $88,400. That's about a $5,000 per unit improvement on a quarter sequential basis relative to where we were One of the biggest drivers of that improvement is through the volume that we saw in the quarter as well.

當然。就本季而言，我們每台的銷貨成本（COGS）是 96,700 美元。如果把 1 億美元換算成每台並扣除，大約是每台 8,200 美元，則會降到 88,400 美元。相較前一季，這大約是每台改善 5,000 美元。這項改善的最大驅動因素之一，也來自於本季我們看到的出貨量（volume）。

So that's going to be -- continue to be a key theme as we'll see our COGS per unit continue to come down over the corresponding quarters of this year. And then beyond that, we did also have at the IEEPA receivable that was also benefiting this quarter.

所以這將會——並且會持續——成為一個關鍵主題，因為我們將看到今年各個對應季度的單位 COGS 持續下降。此外，我們本季也有 IEEPA 應收款項，這也對本季帶來助益。

That would have been revenue? Or was that booked in COGS?

那會被認列為營收嗎？還是記在 COGS 裡？

Booked in COGS.

記在 COGS。

Okay. As a follow-up, RJ, I want to ask something that -- I know you've addressed in the past, but I think it's important just in the early days of the R2 ramp. If we look back to R1, the days of R1 and early ramp, there were -- you probably bid off a bit more than you could to, which is something I think you've acknowledged. There's a lot going on, a lot of different configurations. And we look now, I know you talked about focusing on just the launch of tradition.

好的。作為追問，RJ，我想問一件事——我知道你過去談過，但我認為在 R2 產能爬坡的早期階段特別重要。如果回頭看 R1、R1 的早期爬坡階段，當時——你們可能接下的事情有點超出可負荷範圍，這點我想你也承認過。當時有很多事情在同時進行，很多不同的配置。而現在來看，我知道你提到會專注在傳統版的上市。

But broadly within Rivian, there are a lot of different initiatives between RAP and I'm sure there's multiple other future iterations planned, your Autonomy road map, what's going on with network architecture and then also just Mind also just a number of different initiatives here. So maybe you can just talk us through how this time is a bit different versus R1, how you're approaching R2 execution given that you still have a lot of things on your plate and how you're looking at resource allocation in this environment?

但更廣泛地說，在 Rivian 內部，從 RAP 到我相信還有多個其他未來迭代規劃、你們的自動駕駛路線圖、網路架構在做的事情，然後還有 Mind 以及其他一系列不同的計畫。所以也許你可以帶我們梳理一下，這次相較於 R1 有哪些不同？在你們仍然有很多事情要做的情況下，你們如何推進 R2 的執行？以及在這個環境下你們如何看待資源配置？

I mean the -- it does seem in some ways like yesterday, we were launching R1, and then it also seems like it was a very, very long time ago. But with that said, the company that's launching R2 today is a very different business than what launched R1. I say that with regards to the maturity of our processes, the maturity of the team, the maturity of the development sequencing.

我的意思是——在某些方面，感覺就像昨天我們才在推出 R1，但同時也覺得那已經是非常、非常久以前的事了。不過話說回來，如今要推出 R2 的公司，已經是一個與當初推出 R1 時非常不同的企業。我這麼說是指我們流程的成熟度、團隊的成熟度，以及開發節奏安排的成熟度。

And when we look at the launch and what we're going through today in terms of ramping production, ramping suppliers, what we're experiencing today is actually an artifact of what we did in the two or three years leading up to it. And so a big difference between R1 and R2 is the structured approach we took to what we call design validation builds.

當我們看今天的上市，以及我們在產量爬坡、供應商爬坡方面正在經歷的事情時，今天所遇到的狀況其實是我們在前兩到三年所做工作的結果。因此，R1 與 R2 的一個重大差異，是我們對所謂「設計驗證試製」（design validation builds）採取了更有結構的方法。

So these are vehicles that were in production spec, but built on our pilot line. We went through multiple iterations of that. And then subsequent to that, our manufacturing validation builds where it was, of course, the production spec, but built in the manufacturing plant. And the MVP builds, those manufacturing valuation builds started at the very beginning of this year, at the beginning of 2026.

這些車輛是量產規格，但在我們的試產線上打造。我們做了多次迭代。之後則進入製造驗證試製（manufacturing validation builds），同樣是量產規格，但在製造工廠內生產。而 MVP 試製——這些製造驗證試製——是在今年年初、也就是 2026 年初就開始了。

And so that gave us time to work through issues around ramp, work through issues around design elements, of course, supplier-related issues as we're doing those validation builds. And so the number of unexpected surprises in now the production ramp-up of R2 is far lower incredibly -- it's an incredibly different situation than what we had in R1.

這讓我們有時間去處理爬坡相關問題、處理設計要素相關問題，當然也包括在做這些驗證試製時的供應商相關問題。因此，現在 R2 量產爬坡階段出現的意外狀況大幅降低——這與 R1 時期相比是截然不同的情境。

Now that's not to say there aren't surprises. It's -- as I said at the start, building and launching in new cars, just an inherently very, very complex activity. And it's complex if you're a company that's been building cars for 100 years, it's complex if you're Rivian. And so we've done everything we can to prepare. But as we now are going through the ramp, we're still being very thoughtful around how rapidly we ramp up our supply chain, making sure that all of our suppliers can ramp consistently, planning that very thoughtfully before we bring on the second shift.

當然，這並不是說完全沒有意外。正如我一開始所說，打造並推出新車，本質上就是一項非常、非常複雜的活動。不論是一家造車 100 年的公司，或是 Rivian，這件事都很複雜。所以我們已盡一切所能做好準備。但在我們現在進入爬坡過程時，我們仍非常審慎地看待供應鏈的爬坡速度，確保所有供應商都能一致地提升產能，並在導入第二班之前就非常周全地規劃。

And a big enabler to help make that ramp process more seamless and more straightforward was the simplification of the product relative certainly to what we did before. As you said in R1, we didn't just launch our one. We launched R1T, R1S, and a commercial van, all within the same three-month window. And then we had a very complex set of build variations essentially thousands of different possible build combinations for the R1T and the R1S.

而讓這個爬坡流程更順暢、更直接的一個重要推手，是相較於過去我們對產品做了簡化。如你所說，在 R1 時，我們不只是推出單一車型。我們在同一個三個月的窗口內推出了 R1T、R1S，以及一款商用廂型車。接著我們還有一套非常複雜的生產變體——對 R1T 與 R1S 來說，基本上有數千種可能的組合。

And in sharp contrast to that in R2, we have a very limited set of build combinations that we're building today, which is all embedded in the launch addition. Even within that, we've limited the number of color combinations. And so that was not an accident. That was highly intentional to facilitate a smoother and faster ramp and that the decisions that we took around what combination of features we would have in this launch addition was also something we thought about a lot.

而與此形成鮮明對比的是，在 R2 上，我們目前生產的組合非常有限，全部都包含在首發版（launch edition）之中。即便在首發版內，我們也限制了顏色搭配的數量。所以這不是偶然。這是高度刻意的安排，目的是促成更平順、更快速的爬坡；而我們在這個首發版中決定要提供哪些功能組合，也是一件我們深思熟慮很久的事。

And as I said before, we had a whole bunch of our own expectations around what take rate would be to convert a reservation into an order for the launch addition. And we've been pleasantly surprised that, that conversion from reservation to order has been notably higher than what we expected, which is great news. And that doesn't, of course, mean we're not going to launch those other variants. So for any customers that are waiting for a standard trim or a premium trim, don't worry, we're on it. But those will be coming in the early part of 2027.

而且如我先前所說，我們對首發版從預訂轉為下單的轉換率（take rate）原本有一整套自己的預期。我們很高興地看到，從預訂到下單的轉換率明顯高於我們的預期，這是很好的消息。當然，這並不代表我們不會推出其他版本。所以對於正在等待標準版或高階版的客戶，別擔心，我們正在推進。但那些版本會在 2027 年初推出。

Andrew Percoco, Morgan Stanley.

Andrew Percoco，摩根士丹利。

Can you guys hear me?

你們聽得到我嗎？

Yes.

聽得到。

Awesome. Well, I guess before I get to my question, I do want to say I was out of the 57,000 demo drives that you guys talked about. So as a potential future customer, I just have to give you guys some kudos here on what you've done with the R2.

太好了。我想在進入問題之前，我確實想說，我是你們提到的 57,000 次試駕之一。所以作為一位潛在的未來客戶，我必須在這裡稱讚一下你們在 R2 上所做的成果。

And just recognize that upfront. But I guess just to get to my question, we talked a lot about autonomy already, but I'm just kind of curious how you're thinking about the pricing side and potential uplift to ARPU as you go to point-to-point and then eyes off in 2027 and how you're kind of thinking about enacting those pricing changes over time.

先在一開始表達這點。不過回到我的問題，我們已經談了很多自動駕駛，但我有點好奇你們如何思考定價面，以及當你們在 2027 年推進到點到點（point-to-point）再到「眼睛可離開」（eyes off）時，對 ARPU 可能帶來的提升；以及你們如何看待隨時間推進這些定價調整。

We certainly believe there's opportunity to grow pricing for our autonomous capabilities as we add those capabilities, but we believe very deeply that, that needs to be earned through the expansion of the feature set and expansion of the of the capability.

我們確實相信，隨著我們加入更多自動駕駛能力，在自動駕駛功能的定價上有機會提高；但我們也非常堅信，這必須透過功能集合的擴展與能力的提升來「贏得」這樣的定價。

I'd say the other element here, which we have to be cognizant of is just the market dynamics around this. And it's -- I think it's going to be a very quite fluid around how some of these features get priced. As I said, we're going through an inflection point where customers are going from this being a nice-to-have feature in terms of advanced autonomous capabilities to a must-have feature.

我想這裡的另一個要素，是我們必須留意這方面的市場動態。而且——我認為這些功能的定價方式會相當具有流動性。如我所說，我們正處於一個拐點：客戶正從把先進自動駕駛能力視為「可有可無」的功能，轉變為「必備」功能。

There's a growing awareness around the capability set. And I really believe the transition to Level 3 capabilities where you can go hands-off, eyes off makes it much easier to implement or to use the feature. It's much easier to understand. Early adopters, it's fine to say you have to be hands-on -- or eyes on the road, but you're going to be hands off.

市場對能力組合的認知正在提升。而我真的相信，轉向 Level 3 能力——也就是可以手離方向盤、眼睛離開（hands-off, eyes-off）——會讓這個功能更容易落地或更容易使用。也更容易理解。對早期採用者來說，你可以說你必須手放在方向盤上——或眼睛看著路面——但你會是手離方向盤。

But for the broad full spectrum of customers, it'd be much easier to just say the car will drive itself. You don't have to look at the road. You don't have to touch the wheel. And we think when that happens, it will make it far easier to explain, far easier to see mass scale adoption well beyond just early adopters. And with that, the desire and the expectation for these features to be in technology forward vehicles.

但對於廣泛、全光譜的客戶而言，直接說這台車會自己開，會容易得多。你不必看路。你不必碰方向盤。而我們認為當這件事發生時，將更容易解釋，也更容易看到超越早期採用者、在大眾市場的大規模採用。同時，也會提升人們對這些功能應該出現在科技導向車款中的渴望與期待。

And so ultimately, what the pricing potential is there is something that we're actively spending a lot of time ourselves thinking about. But ultimately, we believe it represents an opportunity to expand pricing beyond where we are today.

因此，最終在定價潛力方面會是什麼樣子，是我們目前投入大量時間在思考的事情。但總體而言，我們相信這代表一個機會，能把定價空間擴展到超越我們今天所在的位置。

Got it. Okay. That makes a lot of sense. And then I was just hoping you can talk a little bit about this new California rebate program. In my understanding, it does require some level of OEM buy-in. So I'm just kind of curious if that's something you're planning on participating in and whether or not you've started to see some incremental demand from that program.

了解。好。這很有道理。接著我想請你談談這個新的加州補助計畫。依我的理解，它確實需要一定程度的 OEM 參與。所以我有點好奇你們是否打算參與，以及你們是否已經看到該計畫帶來一些額外的增量需求。

Sure. We're really happy with the California EV incentive program. And consistent with our mission, we're really excited about the opportunity to attract new first-time EV owners into the Rivian community. So for the program, it's $3,500 for new vehicles and $1,750 for used vehicles with the OEM and the State of California splitting the share of those incentives. There are caps by OEM. And so there will be limitations in terms of the full population set that can take advantage of some of these incentives.

當然。我們對加州電動車激勵計畫非常滿意。而且與我們的使命一致，我們也非常期待能藉此吸引新的首次電動車車主加入 Rivian 社群。就這個計畫而言，新車補助為 3,500 美元、二手車為 1,750 美元，且由 OEM 與加州政府分攤這些補助金額。各 OEM 都有上限。因此，在可享有部分補助的完整人群範圍上，會有一些限制。

Chris Pierce, Needham.

Chris Pierce，Needham。

You've talked a lot about conversion. I guess I'd love to hear the flip side of that. Like what are customers that aren't converting telling you? Are they -- they put in a reservation a long time ago. They're waiting for a different variants. I guess, I'd just like to hear about people maybe falling out.

你們談了很多關於轉換率。我想我也很想聽聽相反的一面。例如，那些沒有轉換的客戶都在跟你們說什麼？他們——很久以前就下了預訂。他們在等不同的車型版本。我想我只是想了解一下，可能有些人會流失的情況。

Well, thanks, Chris. Yes, of course, there's a long list of reasons that one may decide not to convert. But the biggest is they're waiting for a different build combination than what we have in a launch package. So that's by far and away the most common reason for someone to not convert a reservation into an order.

謝謝你，Chris。是的，當然，有一長串原因可能會讓人決定不轉換。但最大的原因是，他們在等一個不同的配置組合，而那不是我們在首發套件（launch package）中提供的。所以這是迄今最常見、也是最主要的原因，導致有人不把預訂轉成正式訂單。

But as I said, recognizing that conversion rates are higher, that's been something we've had to dynamically really watch super carefully because we're providing delivery windows to our launch tradition customers and that requires us to make projections around what the conversion rates will be. Ultimately, when you think about working through the backlog of reservations. And so the way that we've tried to provide guidance on this is representative of thinking about those conversions, but ultimately beginning the delivery of those different trims in 2027.

但如我所說，鑑於轉換率更高，這一直是我們必須動態、非常謹慎地密切觀察的事情，因為我們正在為首發傳統（launch tradition）客戶提供交付窗口，而這需要我們對轉換率做出預測。最終，當你思考如何消化預訂積壓時也是如此。因此，我們嘗試提供的指引，是以這些轉換的思考為代表，但最終會在 2027 年開始交付那些不同的車型等級（trims）。

Okay. And then I think you mentioned you might have another Autonomy in AI day this year. Could this be -- I mean it's been a while since the IPO and long-term financial targets were put out. Could this be something where things have changed for there for worse, for the better now? Like could we see updated long-term targets for people to sort of anchor around?

好的。然後我想你提到今年可能還會再辦一次 Autonomy in AI Day。這——我的意思是，自 IPO 以來也過了一段時間，當時也提出了長期財務目標。這會不會是那種——事情可能已經改變了，不管是變差或變好？例如，我們是否可能看到更新的長期目標，讓大家可以作為某種錨點來參考？

For the Autonomy in the AI Day, we'll be more focused on the technology road map. We're working over the coming quarters to think about the right time to set up a broader-based Investor Day as well that would be more focused on longer-term financial targets. But this one, RJ, I don't know if you want to provide a little bit of context of some of the updates or plans for the day itself.

就 Autonomy in AI Day 而言，我們會更聚焦在技術路線圖。我們正在接下來幾個季度思考，何時是安排一個更廣泛的投資人日（Investor Day）的合適時點，屆時會更著重於較長期的財務目標。但這一次，RJ，我不知道你是否想就當天本身的一些更新或規劃提供一些背景。

Yes. We had our first Autonomy AI Day last year, and we pulled back on a lot of the technology development that had been underway for quite some time with in Rivian. So that was our in-house silicon program, the road map to Level 4 capabilities. And subsequent to that, we've now announced a few important things, one of which is the deal that we've done with Uber for our robotaxi program, which, as I said earlier, is going to be available in for targeting for that to be available in 2028. But there's a lot of other details that we'd like to provide around what that road map looks like.

是的。我們去年舉辦了第一屆 Autonomy AI Day，並對 Rivian 內部已經進行相當長一段時間的許多技術開發做了回顧與揭露。那包括我們的自研晶片（in-house silicon）計畫，以及通往 Level 4 能力的路線圖。在那之後，我們也宣布了幾件重要的事情，其中之一是我們與 Uber 就 robotaxi 計畫達成的合作，如我先前所說，我們的目標是讓它在 2028 年可用。但還有很多其他細節，我們希望能就那個路線圖的樣貌提供更多說明。

We'd like to show the progress that's happened on the vehicle to a what's in customer vehicles today, but actually from a technical point of view on the path to Level 4. And of course, our objective is to provide better visibility and therefore, rethink a higher level of confidence around what those targets -- around achieving those starts.

我們希望展示從技術角度來看，在通往 Level 4 的路徑上，車輛相較於今天客戶車上所具備的能力，已經取得了哪些進展。當然，我們的目標是提供更好的可見度，並因此重新建立更高程度的信心，圍繞那些目標——以及達成那些起點（starts）。

Philippe Houchois, Jefferies.

Philippe Houchois，Jefferies。

Yes. My first question is on the relationship with Uber. We've seen some tension between Uber and Waymo and their cooperation. And of course, there's a bit of uncertainty about the financial stability of Lucid and the fact that Lucid work with neuro. So we started as integrated a solution as what you offer in the parts with Uber. So I'm just wondering how much of a upside to your relationship there is, maybe not in the timing, but in the magnitude of what you could be working with Uber in terms of offering ruble taxes?

是的。我的第一個問題是關於你們與 Uber 的關係。我們看到 Uber 與 Waymo 之間以及他們的合作出現了一些緊張情況。當然，Lucid 的財務穩定性也有一些不確定性，而且 Lucid 與 Nuro 合作。所以你們與 Uber 所提供的是一個相當整合的解決方案。因此我想請教，你們這段關係的上行空間有多大——也許不是在時間點上，而是在規模上：就你們能與 Uber 合作、提供 robotaxi 的量級而言，可能會到什麼程度？

Well, a core element of our approach is a strongly-held belief that vertically integrating the vehicle and the technology stack together allows for meaningful, both cost advantages and technical simplifications in terms of the vehicle architecture and autonomy hardware architecture. And so that means how sensors are integrated into the vehicle, the redundancies built around the controls, the way that we've architected the compute platform in the inference in our case because we brought silicon in-house, we have a very low-cost way to achieve very high levels of inference within the vehicle.

嗯，我們方法的一個核心要素，是我們堅信把車輛與技術堆疊（technology stack）做垂直整合，能在車輛架構與自動駕駛硬體架構上，帶來顯著的成本優勢與技術簡化。因此，這涵蓋了感測器如何整合進車輛、控制系統周邊建立的冗餘設計、以及我們如何架構運算平台與推論（inference）——在我們的案例中，因為我們把晶片自研內製，所以能以非常低的成本，在車內達到非常高水準的推論能力。

And so we believe that's a technical approach we've taken, from a hardware point of view, creates a structural cost structure, I should say, that's highly advantaged for delivering Level 4. Along with that, we believe that the fact that we're building a neural net-based approach, an approach that's an end-to-end trained model with what we call LDM, or large driving model, means that the progress that we make on the consumer platform. So that's a point-to-point hands-off eyes off Level 3, ultimately getting into Level 4.

因此我們相信，從硬體角度來看，我們採取的這種技術路徑，形成了一個在交付 Level 4 上具高度優勢的結構性成本結構（structural cost structure）。此外，我們相信我們正在打造一種以神經網路為基礎的方法——也就是我們所稱的 LDM（large driving model，大型駕駛模型）的端到端訓練模型——意味著我們在消費者平台上取得的進展。也就是點到點、免手（hands-off）免眼（eyes-off）的 Level 3，最終進入 Level 4。

That LDM workflow is identical across all those different steps. So that staircase up to Level 4 is not -- is one work stream. It's not multiple work streams where we have a Level 4 platform and a Level two platform, but rather there's one LDM, one end-to-end train model. And the vehicles will have slightly different sensor topologies to cover corner cases differently between, let's say, a Level 3 and Level 4 vehicle.

那套 LDM 的工作流程在所有不同階段之間是完全相同的。因此，通往 Level 4 的那個階梯並不是——而是一條工作流。它不是多條工作流：不是我們同時有一個 Level 4 平台和一個 Level 2 平台；相反地，是一個 LDM、一個端到端訓練模型。而車輛會有些微不同的感測器拓撲（sensor topologies），以便在 Level 3 與 Level 4 車輛之間，針對邊角案例（corner cases）以不同方式覆蓋。

And along with that slightly different inference capabilities built around RAP1 but from a platform, from a model point of view, it's the same. And so that was a key shift for us. And I think it's a key shift for the industry in terms of how we approach autonomy that historically prior to, let's say, 2022, 2023, the technology topologies used for Level 4 were very different than the topologies used for Level 2.

同時也會有一些圍繞 RAP1 建構的、略有不同的推論能力；但從平台、從模型的角度來看，是相同的。因此，這對我們而言是一個關鍵轉變。而我認為，這也是產業在自動駕駛方法論上的一個關鍵轉變：在過去（例如 2022、2023 年之前），用於 Level 4 的技術拓撲與用於 Level 2 的拓撲是非常不同的。

And we've seen that merge with the approach of using transformer-based encoding and the end-to-end models that is really to the advantage of our architecture to the advantage of the data flywheel that exists through the deployed fleet that we have.

而且我們已經看到，這與採用以 Transformer 為基礎的編碼方式以及端到端模型的方法相結合，確實有利於我們的架構，也有利於透過我們已部署車隊所形成的資料飛輪。

But specifically on Uber, you think there's a big opportunity for you? Because I mean, when you announced the partnership, there was -- you were one of several partners that uberwas working with. Is there an opportunity for Rivian to be actually a much bigger partner of Uber in Autonomy?

但特別就 Uber 而言，你認為對你們來說有很大的機會嗎？因為我的意思是，當你們宣布合作夥伴關係時，當時——你們是 Uber 正在合作的幾個夥伴之一。Rivian 是否有機會在自動駕駛（Autonomy）方面，實際上成為 Uber 更大的合作夥伴？

Well, I'd say the working relationship with Uber has been fantastic from the top of the organization through the working levels of the organization. So the engagement from Dara in terms of the importance of this relationship, along with the working level teams, it's been very positive we're working towards our next milestone at the end of this year and deploying in a couple of cities in 2028 and then rapidly growing following that to a large number of cities.

嗯，我會說與 Uber 的合作關係一直非常出色，從組織高層到各工作層級都是如此。因此，Dara 對於這段關係重要性的投入，再加上工作層級團隊的參與，一直都非常正面；我們正朝著今年年底的下一個里程碑邁進，並計畫在 2028 年於幾個城市部署，之後再迅速擴展到大量城市。

And so this is we're quite encouraged by the relationship and the partnership. But ultimately, I do believe that the focus we have and the focus, I think, appropriately where we're placing our energy is making sure that the technology is ready. The vehicle is capable of delivering Level 4 performance and doing it in a thoughtful way in terms of really clean execution at the vehicle level. And really cost-effective ways of delivering that capability.

因此，我們對這段關係與合作夥伴關係感到相當鼓舞。但最終，我確實相信，我們的重點——我認為我們把精力放在這裡是恰當的——是確保技術已經準備就緒。車輛必須能夠提供 Level 4 的表現，並且在車輛層級以周全的方式做到真正乾淨俐落的執行。以及以真正具成本效益的方式交付這項能力。

Right. And then maybe more for Claire about the ramp and -- I think you still haven't told us exactly how many R2s you delivered. It's just maybe your handful. But you had warned us about the risk to gross margin as you ramp up and in fact, I know there was a bit of an outside help in Q2 from IEEPA, from Z income. But the dip in the gross margin has actually not really happened.

了解。接著可能更多是問 Claire，關於爬坡（ramp）以及——我想你們仍然沒有告訴我們到底交付了多少台 R2。可能只是少量。但你們先前提醒過，隨著 R2 產能爬坡，毛利率可能面臨風險；而且事實上，我知道第二季（Q2）在 IEEPA、以及 Z income 方面有一些外部助益。但毛利率的下滑其實並沒有真的發生。

And now you're warning us that they could be in Q3. I'm just wondering, is it because the ramp really hasn't happened that much on R2 in the second quarter, or has it actually happened more in a smoother manner than you feared. And of course, you're cautious about Q3, but we could have a positive surprise in the Q3 gross margin as well.

而現在你們警告說，可能會在第三季（Q3）出現。我只是想知道，這是因為第二季 R2 的爬坡其實還沒有發生太多，還是實際上爬坡進行得比你們擔心的更平順。當然你們對 Q3 謹慎，但 Q3 的毛利率也可能出現正向驚喜。

As we think about the second quarter, you can think about us converting from a preproduction environment to our start of production and then start of our cost structure hitting our cost of goods sold for roughly two-third of the quarter. So it was not a full quarter's impact. And so as we look ahead to the third quarter, we'll have a full quarter's impact of the ramp. We'll also, as RJ noted, had the introduction of a second shift of operation and so some additional labor costs associated with the ramp as well.

當我們看第二季時，你可以把它想成我們從試產環境轉換到開始量產，然後我們的成本結構開始反映到銷貨成本（COGS）中，大約涵蓋了該季約三分之二的時間。所以它並不是整季的完整影響。因此展望第三季，我們將會承受爬坡的整季影響。另外，如 RJ 所提到的，我們也導入了第二班制運作，因此爬坡也會伴隨一些額外的人力成本。

And then in the fourth quarter, that's where you'll start to see the benefits of scaling production, the fixed cost leverage, both across R2 as well as the broader plant. And then with that maybe also invite Javier to jump in as well, just as we think about driving towards manufacturing efficiencies and executing the ramp in the second half.

然後到了第四季，你們就會開始看到擴大量產的效益、固定成本的槓桿效應，這不僅體現在 R2 上，也會體現在整個工廠更廣泛的層面。另外，或許也請 Javier 一起補充，談談我們在下半年推動製造效率、以及執行爬坡的思考。

Yes. Thank you, Claire. Indeed, there's a big focus -- laser-focused on the teams in normal to ramp up this first shift. That's what we started in quarter -- and we continue in quarter three with the preparation of R2. We are not expecting any material contribution to volumes with the second shift in this quarter, but we will see it in quarter 4.

是的。謝謝你，Claire。確實，團隊在 Normal（工廠）對於把第一班制拉升產能有很大的重點——高度聚焦。這是我們在上一季開始做的——而且我們在第三季會持續進行 R2 的準備。我們不預期第二班制在本季會對產量帶來任何實質貢獻，但在第四季會看到。

The capability of the teams has very much increased compared to the launch of R1. And that's a big advantage. I mean the winning teams are the ones that are capable to solve problems at the speed. And that's what we are doing now, focusing on training of the members in the far lines, training the team, the automation and getting the output higher and higher as we ramp. And then, of course, focusing on the supply chain that is always not surprised.

相較於 R1 的上市，團隊的能力已經大幅提升。這是一個很大的優勢。我的意思是，勝出的團隊是那些能以速度解決問題的團隊。而這正是我們現在在做的：專注於產線成員的訓練、團隊訓練、自動化，並在爬坡過程中把產出拉得越來越高。當然，也會專注在供應鏈上，因為供應鏈總是會帶來意外。

I believe many launches in my life and always the supply chain is a part that requires special attention. So we are focusing on a handful of suppliers with boots on the ground and with a great example of collaboration where the suppliers are really willing to get our support and we are willing to give them. So that's the focus in normal today.

我一生中經歷過很多次新產品導入（launch），而供應鏈永遠是需要特別關注的一環。因此我們正把重點放在少數幾家供應商上，派駐人員到現場，並以一個很好的協作案例來運作：供應商真的願意獲得我們的支援，而我們也願意提供支援。所以這就是目前在 Normal 的重點。

This concludes the Q&A section of the call. I would now like to turn the call back over to RJ Scaringe for closing remarks.

本次電話會議的問答環節到此結束。接下來我想把電話交回給 RJ Scaringe，請他做結語。

Thanks, everybody, for joining us on today's call. Hopefully, you're observing a lot of excitement from the team here around R2. That excitement is echoing the excitement that we're seeing from consumers and it's openly been something that we as a business have been building towards for a while. The ramp that we're going to see in the second half of the year, so Q3 into Q4, as Javier just described with Q4 being a really important element where we start to see the benefits of our second shift online. That's a major focus for us.

謝謝各位參加今天的電話會議。希望你們能感受到我們團隊對 R2 的高度興奮。這份興奮也呼應了我們從消費者端看到的熱度，而且坦白說，這也是我們這家公司一段時間以來一直在努力打造、邁向的目標。我們將在今年下半年看到的爬坡，也就是從 Q3 進入 Q4；正如 Javier 剛才所描述的，Q4 是非常重要的一個環節，因為我們會開始看到第二班制上線所帶來的效益。這是我們的重大重點。

And our operational teams, our engineering teams, our customer-facing teams are all very much aligned to be focused on delivering on that ramp. And as you also heard, we're continuing to focus on our technology road map. So that's continued progress on our software-defined architecture. And as we really think of it as our AI defined architecture, so vehicles that have a growing set of features and capabilities that growing the features and capabilities allows not only the vehicle to create a better customer experience, but exists within our ecosystem more efficiently.

而我們的營運團隊、工程團隊、以及面向客戶的團隊，都高度一致地聚焦在把這個爬坡交付出來。同時，如你們也聽到的，我們也持續專注於我們的技術路線圖。也就是在我們軟體定義架構上的持續進展。而當我們把它視為 AI 定義架構時，也就是車輛具備一套持續成長的功能與能力；這些功能與能力的成長，不僅讓車輛能創造更好的客戶體驗，也讓它能更有效率地存在於我們的生態系之中。

And then, of course, the enormous emphasis that we have on our self-driving platforms, our self-driving platform and growing capability over the course of this year with the introduction of point-to-point late this year and then ultimately expanding that in 2027 and 2028 to ultimately get to Level 4. And so this is the exciting time for us in the business.

然後，當然，我們也非常強調自動駕駛平台；在今年內，隨著今年稍晚推出點對點（point-to-point）功能，我們的自動駕駛平台能力會持續提升，並在 2027 與 2028 年進一步擴展，最終達到 Level 4。因此，對我們這家公司而言，這是一個令人振奮的時刻。

We see it as an inflection point for us as a business as well, both in terms of our path to profitability, but also in terms of the scale and scope of the business given the mass market nature of R2. Once again, thanks, everyone, for joining today's call and look forward to future discussions.

我們也把它視為公司的一個拐點（inflection point），不僅在我們邁向獲利的路徑上，也在業務規模與範疇上——考量到 R2 的大眾市場（mass market）定位。再次感謝各位參加今天的電話會議，並期待未來的進一步交流。