優步 (UBER) 2026 Q2 法說會逐字稿

Hello, and welcome to Uber's Q2 2026 earnings conference call. (Operator Instructions)

各位好，歡迎參加 Uber 2026 年第二季財報電話會議。（接線員指示）

I would now like to turn the conference over to Alax Wang, Head of Investor Relations. You may begin.

現在我想把會議交給投資人關係主管 Alax Wang。您可以開始了。

Thank you, Sarah. Thank you for joining us today, and welcome to Uber's second quarter 2026 earnings presentation. On the call today, we have Uber's CEO, Dara Khosrowshahi; and CFO, Balaji Krishnamurthy.

謝謝你，Sarah。感謝各位今天加入我們，歡迎參加 Uber 2026 年第二季財報簡報。今天與會的有 Uber 執行長 Dara Khosrowshahi，以及財務長 Balaji Krishnamurthy。

During today's call, we will present both GAAP and non-GAAP financial measures. Additional disclosures regarding these non-GAAP measures, including a reconciliation of GAAP to non-GAAP measures, are included in the press release, supplemental slides and our filings with the SEC, each of which is posted to investor.uber.com.

在今天的電話會議中，我們將同時呈報 GAAP 與非 GAAP 財務衡量指標。關於這些非 GAAP 指標的額外揭露（包含 GAAP 與非 GAAP 指標之間的調節表），已收錄於新聞稿、補充投影片以及我們向美國證券交易委員會（SEC）提交的文件中；上述資料均已發布於 investor.uber.com。

Certain statements in this presentation and on this call are forward-looking statements. You should not place undue reliance on forward-looking statements. Actual results may differ materially from these forward-looking statements, and we do not undertake any obligation to update any forward-looking statements we make today, except as required by law.

本簡報及本次電話會議中的某些陳述屬於前瞻性陳述。您不應過度依賴前瞻性陳述。實際結果可能與這些前瞻性陳述存在重大差異；除法律要求外，我們不承擔更新今日所作任何前瞻性陳述之義務。

For more information about factors that may cause actual results to differ materially from forward-looking statements, please refer to the press release we issued today as well as risks and uncertainties described in our most recent Form 10-K and in other filings made with the SEC. We published our quarterly earnings press release, prepared remarks and supplemental slides to our Investor Relations website earlier today. We ask you to review those documents if you haven't already. We will open the call to questions following brief opening remarks from Dara.

如需了解可能導致實際結果與前瞻性陳述存在重大差異的因素，請參閱我們今日發布的新聞稿，以及我們最新的 Form 10-K 與其他向 SEC 提交文件中所描述的風險與不確定性。我們已於今日稍早在投資人關係網站發布本季財報新聞稿、預先準備的發言稿與補充投影片。若您尚未閱讀，敬請查閱上述文件。在 Dara 簡短開場致詞後，我們將開放提問。

With that, let me hand it over to Dara.

接下來，我把時間交給 Dara。

Thanks, Alax. So Q2 was another outstanding quarter for Uber with record audience and engagement, driving durable growth, expanding margins and recording cash -- and record cash generation. Gross bookings grew 22% year-on-year to more than $58 billion, above the high end of our guidance and marking our fourth consecutive quarter above 20% growth.

謝謝，Alax。第二季對 Uber 而言又是非常出色的一季，受眾規模與互動度創下新高，帶動穩健成長、擴大利潤率，並創下現金與現金產生的紀錄。總預訂額年增 22% 至超過 580 億美元，高於我們指引區間上緣，並且是我們連續第四季維持 20% 以上的成長。

Just as importantly, that growth translated into significant operating leverage with non-GAAP EPS growing 35% year-over-year and trailing 12-month free cash flow exceeding $10 billion for the first time in our history. Those results give us the ability to continue investing from a position of strength across our priorities.

同樣重要的是，這樣的成長轉化為顯著的營運槓桿：非 GAAP 每股盈餘（EPS）年增 35%，且過去 12 個月自由現金流首次在公司歷史上突破 100 億美元。這些成果讓我們能夠在穩健的基礎上，持續針對各項優先事項進行投資。

Our recently announced agreement to acquire Delivery Hero is a great example. The combination is going to expand our reach to nearly 100 markets and extend the proven strategy that has underpinned our growth for years by roughly doubling the number of markets where we can offer the full power of our platform across mobility and delivery. By bringing our technology and Uber One to millions more people, we believe this deal will create significant long-term shareholder value.

我們近期宣布的收購 Delivery Hero 協議就是很好的例子。此一結合將把我們的觸及範圍擴展至近 100 個市場，並透過約略將可在「移動出行」與「外送」兩大領域提供完整平台能力的市場數量加倍，延伸多年來支撐我們成長、已被驗證的策略。透過把我們的技術與 Uber One 帶給更多數百萬人，我們相信這筆交易將創造可觀的長期股東價值。

We also continue to invest behind one of the largest opportunities in Uber's history, autonomous vehicles. Over the past year, the conversation around AVs has shifted from whether the technology can deliver a compelling service to how broadly, reliably and economically it can scale. That distinction matters because we believe the industry structure is becoming clearer. It mirrors what's happening in AI broadly.

我們也持續投入 Uber 歷史上最大機會之一：自動駕駛車輛。過去一年，市場對 AV 的討論焦點已從「技術是否能提供具吸引力的服務」轉向「能在多大範圍內、以多高可靠度、以及以多具經濟效益的方式擴張規模」。這個差異很重要，因為我們認為產業結構正變得更清晰。這與 AI 領域整體正在發生的情況相呼應。

A few years ago, many expected AI to converge around a single foundation model. Instead, multiple frontier models have emerged alongside a growing open source ecosystem. We believe AVs, which are essentially physical AI, will follow a similar path.

幾年前，許多人預期 AI 會收斂到單一的基礎模型。然而，實際上出現了多個前沿模型，同時也形成日益壯大的開源生態系。我們相信 AV（本質上是實體 AI）也將走上類似的路徑。

Just in the last week, we've seen NVIDIA release its Alpamayo open-weight model and our partners at Wayve received a permit in the UK and Zoox received approval to scale its robotaxi. The momentum across our partner base is remarkable. Unlike foundation models, however, AVs are physical, regulated systems that have to be deployed market by market with the variable patterns of ride-hailing.

就在上週，我們看到 NVIDIA 發布其 Alpamayo 開放權重模型；我們的合作夥伴 Wayve 在英國取得許可，而 Zoox 也獲准擴大其機器人計程車（robotaxi）規模。我們合作夥伴陣容的動能令人矚目。但與基礎模型不同的是，AV 是實體且受監管的系統，必須逐一市場部署，並因應叫車服務多變的需求型態。

That makes the commercialization layer critical, and that's exactly the opportunity we're building towards. And our ambition is straightforward, to become the world's leading commercialization platform for autonomous vehicles. Taken together, we're executing with discipline today while building the capabilities we believe will define Uber's next decade of growth.

因此，商業化層至關重要，而這正是我們正在打造的機會所在。我們的目標很明確：成為全球領先的自動駕駛車輛商業化平台。綜合而言，我們一方面以紀律執行當下，另一方面也在建立我們相信將定義 Uber 下一個十年成長的能力。

With that, operator, let's open up for questions.

接下來，接線員，請開放提問。

Brian Nowak, Morgan Stanley.

Brian Nowak，摩根士丹利。

Maybe, Dara, a double-pronged one on autonomous, one on mobility and delivery. Just you've made a lot of progress with different partnerships, and there's a lot of technological progress happening. But maybe help us understand, if you look ahead 12 months from now, what are sort of the main milestones or progress sign points you're focused on, on your mobility strategy and your delivery strategy in AV just to ensure you're going down the right path?

Dara，我想問一個「雙主題」的問題：一個關於自動駕駛，另一個關於移動出行與外送。你們在不同合作夥伴關係上取得了很多進展，技術也在快速演進。但能否協助我們理解：若往後看 12 個月，你們在 AV 的移動出行策略與外送策略上，最關注的主要里程碑或進展訊號是什麼，以確保你們正走在正確的路徑上？

Yes, absolutely, Brian. So in terms of what the milestones we're looking at, the most important milestones are really launches and the number of cities that we are live in, both with vehicle operators and then without vehicle operators. So to remind you, we're live in seven cities, and we're on track to be live in 15 cities by year-end. We've got a Nuro Lucid launch coming in the Bay. We've got Zoox coming in Vegas. We've got Wayve in London and Tokyo, Baidu also in London and then Pony and Verne in Zagreb and potentially more. So there's a bunch of launches coming up in the balance of the year.

好的，當然可以，Brian。就我們所關注的里程碑而言，最重要的其實是「上線」以及我們已經在多少城市正式營運——包含有車上安全員／操作員，以及無車上操作員的兩種模式。提醒大家，我們目前已在 7 個城市上線，並按計畫在年底前達到 15 個城市上線。我們即將在灣區推出 Nuro Lucid 的上線。Zoox 也將在拉斯維加斯上線。Wayve 將在倫敦與東京上線，Baidu 也會在倫敦上線，另外 Pony 和 Verne 會在札格瑞布（Zagreb）上線，且可能還有更多。因此，今年剩餘時間會有一系列上線計畫。

And then really, we're looking for more in 2027. So we'll continue to launch markets with the partners that I talked about, but we'll be adding additional partners into the ecosystem as well. We expect to see Rivian in the market, and these are -- this is kind of a full stack kind of a build, which is software and hardware with a very affordable bill of materials. We expect to be in perhaps San Francisco and Miami in 2028 for Rivian. And then in (inaudible), we expect to be in L.A. and San Francisco in 2027 and then 28 different cities globally by '28 as well.

接著，真正更大的進展會在 2027 年出現。我們會持續與我剛提到的合作夥伴在更多市場上線，同時也會把更多合作夥伴納入生態系。我們預期會看到 Rivian 進入市場；這是一種「全棧」建置，涵蓋軟體與硬體，且物料成本（BOM）非常具競爭力。就 Rivian 而言，我們預期可能在 2028 年於舊金山與邁阿密上線。另外在（聽不清楚）方面，我們預期 2027 年在洛杉磯與舊金山上線，並且到 2028 年也會在全球 28 個不同城市上線。

So really, what we're looking for is launches in markets, accelerating our data collect, which is really driving the newer kind of end-to-end models as well and then starting to really commercialize this model. The numbers are small at this point. But for us, what we're looking at is, first of all, the quality of the service. And then second, the utilization of these vehicles.

因此，我們真正關注的是：在各市場的上線、加速資料蒐集（這會推動更新的端到端模型），以及開始真正把這個模式商業化。目前數量仍然不大。但對我們而言，首先關注的是服務品質。其次是這些車輛的使用率。

And what we've seen is that launching with us as a partner with kind of the built-in kind of demand that we've got, we can drive very significant utilization per vehicle, often mid- to high 20s, low 30s in terms of trips per vehicle per day, which is quite substantial in terms of the needed monetization as it relates to these vehicles. So it's getting partners in market, quality of service and then obviously, the economics of the service that we're looking at.

我們看到，與我們合作上線，憑藉我們內建的需求量，能夠把單車使用率推到非常可觀的水準，通常每日每車行程數可達 20 多趟的中高段，甚至 30 出頭；就這些車輛所需的變現而言，這是相當可觀的表現。因此，我們關注的是讓合作夥伴進入市場、服務品質，當然還有我們所觀察的服務經濟性。

On the delivery side, we're also partnered with a number of partners, whether it's Serve or Cartken for delivery kind of sidewalk robots. What I am increasingly optimistic about are the potential of drones. So we've got, for example, partnerships with Flytrex. There are a number of partnerships coming up in the drone area. And the promise of drones, while it will take time to get the economics, down to kind of the kind of economics that can be sustainable.

在配送端，我們也與多家夥伴合作，不論是 Serve 或 Cartken 這類用於配送的人行道機器人。我愈來愈樂觀的是無人機的潛力。因此我們例如與 Flytrex 建立了合作夥伴關係。在無人機領域還有多項合作正在推進。而無人機的承諾在於——雖然要讓經濟性下降到可持續的水準仍需要時間。

With drone delivery -- with AVs, it's a really cool experience. It's safe, et cetera, but AVs, to some extent, are slower than human drivers at this point. We'll see what happens there. Drone delivery can cut the timing of delivery significantly.

無人機配送——以及自動駕駛車（AV）——都是非常酷的體驗。它很安全等等，但在某種程度上，目前 AV 的速度仍比人類駕駛慢。我們會看看後續發展。無人機配送可以大幅縮短配送時間。

So already kind of ordering dinner and getting it delivered to your home in 30 minutes is a magical experience. We think in 15 minutes, it's going to be -- 10 to 15 minutes is going to be an even more magical experience. So while we're on the sidewalks now as it relates to our delivery business, we're looking forward to getting in the air with some of the partners out there.

所以現在點晚餐並在 30 分鐘內送到家，已經是很神奇的體驗。我們認為若能在 15 分鐘——也就是 10 到 15 分鐘內送達——會是更神奇的體驗。因此，雖然就我們的配送業務而言目前仍在「人行道」上運作，但我們也期待能與部分合作夥伴一起「飛上天空」。

Eric Sheridan, Goldman Sachs.

Eric Sheridan，高盛。

Two, if I could. On the US mobility acceleration, can you unpack a little bit some of the drivers of that acceleration? And how much you think might be building permanent signal as opposed to elements of the World Cup that might have played a role in that? And then I was struck by the comment that the first-time user momentum was as high as it was.

我有兩個問題。關於美國出行（mobility）加速成長，你能否進一步拆解一下推動這個加速的因素？以及你認為其中有多少是可持續的長期訊號，而不是像世界盃這類可能發揮作用的短期因素？另外，我也注意到你提到首次使用者的動能高得驚人。

What do you think you're putting in place there that's resulting in that type of first-time user momentum? And again, how should we be thinking about that sustaining going forward?

你認為你們做了哪些事情，才帶來這種首次使用者動能？同樣地，我們應該如何看待這種動能在未來的延續性？

Thanks, Eric. I'll take the question on US mobility and Dara will take the second one. So -- the World Cup definitely was a benefit, but it was as expected to a large extent. And what I would say is the momentum that we're seeing in the US is consistent with the theme that we have been talking about since the beginning of the year. And it is far broader than any one-off event that we're talking about here.

謝謝你，Eric。我來回答美國出行的問題，第二個問題由 Dara 來回答。世界盃確實帶來了助益，但在很大程度上也符合我們的預期。我想說的是，我們在美國看到的動能，與我們自年初以來一直在談的主題一致。而且它的範圍遠比我們在這裡討論的任何一次性事件更廣。

So just as a reminder, what we have said is we expect the US to accelerate through this year. And we highlighted in Q2, both trips and gross bookings accelerated. We continue to hold our expectation for the remainder of the year. And what's driving that is three distinct items.

提醒一下，我們之前說過，我們預期美國會在今年加速成長。我們在第二季也強調，無論是行程（trips）還是總預訂額（gross bookings）都在加速。我們仍維持對今年剩餘期間的預期。而推動因素有三個明確項目。

The first one is insurance, which we have talked about, is becoming a tailwind this year. We are reinvesting the savings from insurance back into the market. And particularly in California, where there's a significant amount of reinvestment, we are seeing a very clear inflection in growth. And in L.A. and SF, our trip growth, for example, meaningfully outpaced the rest of the country.

第一個是保險。我們之前談過，今年保險正成為順風因素。我們把保險節省下來的成本再投資回市場。特別是在加州，我們進行了相當程度的再投資，並看到非常明顯的成長拐點。例如在洛杉磯與舊金山，我們的行程成長顯著超過全美其他地區。

Then secondly, our product innovation velocity that we've talked about on both premium products and on affordable products, whether it's Reserve, U4B and Black on the premium side or Wait & Save and other products on the other side of the spectrum, we are seeing really good traction there. U4B, in particular, we saw 40% year-on-year growth. Uber Health is growing even faster. So that's the second theme.

第二，我們在產品創新速度方面——不論是高端產品或平價產品——都如我們所談到的，正在取得進展；高端端包括 Reserve、U4B 和 Black，另一端則有 Wait & Save 等產品，我們看到非常好的牽引力。尤其是 U4B，我們看到年增 40%。Uber Health 的成長更快。這是第二個主題。

And then finally, sparse markets, which we've been talking about as a long-term opportunity, that continues to be a very strong driver for growth as well. And just for context, in the US less than 10% of eligible consumers in our sparse markets have used Uber in the past 12 months, whereas in dense markets, that number is over 50%. So we're making really good progress there in improving supply, reliability and investing in marketing to drive awareness and trial. So I think all in, it's broad and sustainable.

最後是稀疏市場（sparse markets）。我們一直把它視為長期機會，而它也持續成為非常強勁的成長驅動因素。補充背景：在美國，我們稀疏市場中符合條件的消費者，過去 12 個月使用過 Uber 的比例不到 10%；相較之下，在密集市場（dense markets）這個數字超過 50%。因此我們在提升供給、可靠性，並投資行銷以提升認知與試用方面，進展非常好。總體而言，我認為這是廣泛且可持續的。

And then in terms of first-time users, we're obviously very happy about the kind of record first-time users at least over the past couple of years coming into the platform. And the trends are actually -- or the drivers are similar. The ones that I would call out generally are our lower-cost products continue to grow globally, but also in the US as well.

至於首次使用者，我們當然非常高興看到至少在過去幾年中，平台迎來了創紀錄的首次使用者數。而其趨勢——或驅動因素——其實相似。我想特別指出的是，我們的低成本產品在全球持續成長，在美國也同樣如此。

Globally, our low-cost products are kind of two-wheelers and three-wheelers. In the US it's Wait & Save, which allows consumers to kind of trade off time against price, and Wait & Save is growing very, very quickly. So low cost, it kind of introduces a whole new sector of consumers onto the platform, and then we can sell them up to the mainline as well. That's one.

在全球，我們的低成本產品主要是二輪與三輪車。在美國則是 Wait & Save，讓消費者在時間與價格之間做取捨，而 Wait & Save 成長非常、非常快。因此，低成本產品會把一整個新的消費者族群帶進平台，之後我們也能把他們升級導向主流產品。這是一點。

Second is exactly what Balaji talked about, which is sparse markets. We -- the mobility business and the delivery business is growing significantly faster in sparse markets than in dense markets. We're kind of building out inventory in those markets. Margins are actually quite healthy as well. So you get increased growth as well as strong margins as well. And we are a very, very long way from high penetration in those sparse markets, as Balaji pointed out.

第二點正如 Balaji 談到的，就是稀疏市場。我們——無論是出行業務還是配送業務——在稀疏市場的成長都顯著快於密集市場。我們正在這些市場擴充供給與品類（inventory）。利潤率其實也相當健康。因此你同時得到更高的成長與更強的利潤率。而且如 Balaji 所指出，我們距離在這些稀疏市場達到高滲透率仍非常、非常遙遠。

Third for us is cross-platform. And most of our cross-platform activity is -- has been historically our Rides to Eats, and I'll remind folks that only 20% of our consumers at this point use both Rides and Eats, and that's growing 1.5 times faster than single product users as well.

第三點是跨平台。我們的大部分跨平台活動——歷史上主要是從 Rides 導向 Eats；我也提醒大家，目前只有 20% 的消費者同時使用 Rides 與 Eats，而這部分用戶的成長速度也比單一產品用戶快 1.5 倍。

So that's a very sustainable advantage that we have as it relates to cross-platform. And as our Eats business is getting bigger, not only is Rides moving consumers to Eats, but then our Eats business is moving consumers on to Rides as well, and that's true both globally and in the US as well.

因此，就跨平台而言，這是我們非常可持續的優勢。而且隨著 Eats 業務變得更大，不僅是 Rides 把消費者帶到 Eats，我們的 Eats 也會把消費者帶回 Rides；這在全球與美國都同樣成立。

And then last but not least, our new products that we're introducing into the marketplace. So for example, Women Preferred is bringing a lot of women, both drivers and riders into the marketplace. Uber Teens is another new product that's kind of bringing in a whole new demographic and audience into the marketplace.

最後但同樣重要的是，我們正在市場上推出的新產品。例如 Women Preferred 正在把許多女性——不論是駕駛或乘客——帶進市場。Uber Teens 是另一個新產品，正在把全新的族群與受眾帶進市場。

And then, for example, higher-margin product like U4B gives us high margins, but it's also introducing kind of a new enterprise type audience as well. So you put all that together, whether it's low-cost, cross-platform, newer products or sparse markets, that's really what's contributing to both the growth of the business as well as the growth of the audience.

再例如，像 U4B 這類高毛利產品能帶來高利潤率，同時也引入一種新的企業型受眾。把這些因素合在一起——不論是低成本、跨平台、新產品或稀疏市場——這些正是同時推動業務成長與受眾成長的關鍵。

Doug Anmuth, JPMorgan.

Doug Anmuth，摩根大通。

I have two. You highlighted some softness in Brazil mobility trips. Could you just talk about the competitive environment there? And then also if there are any other markets where you need to invest in Moto and low-cost mobility products?

我有兩個問題。你提到巴西出行行程有些疲軟。你能否談談那裡的競爭環境？另外，是否還有其他市場需要你們在 Moto 與低成本出行產品上加大投資？

And then secondly, there's been a lot in the press recently regarding Uber's relationship with Waymo. Is there anything you can add or clarify there and specifically how you're thinking about those Austin and Atlanta partnerships?

其次，最近媒體上有很多關於 Uber 與 Waymo 關係的報導。你能否補充或澄清一下，特別是你們如何看待奧斯汀與亞特蘭大這些合作夥伴關係？

Yes. Absolutely, Doug. So in Brazil, we are seeing a competitive environment. It's not -- it's always been competitive as it relates to mobility. We compete against DiDi, for example, locally. And Latin America generally has been quite competitive. What we're seeing that's new and different in Brazil is there is an enormous amount of competition as it relates to the food business. DiDi's introduced DiDi Food. Meituan has gone in there going against iFood, which is the incumbent in Brazil as well.

是的。當然可以，Doug。所以在巴西，我們看到的是一個競爭環境。就出行而言，一直以來都很競爭。例如，我們在當地與滴滴（DiDi）競爭。而拉丁美洲整體而言也一直相當競爭。我們在巴西看到新的、不同之處在於：在外送餐飲業務方面出現了極其激烈的競爭。滴滴推出了 DiDi Food。美團也進入當地，與同樣是巴西既有龍頭的 iFood 競爭。

That -- all of them are going after two-wheeler delivery supply. And that same two-wheeler delivery supply will switch off between delivering food and also moving people as well. So the cost of securing that supply has gone up pretty significantly. And we're moving incentives that we kind of put on the consumer side. We're moving it to the delivery side to counteract that, if you will.

他們——全部都在爭奪兩輪車外送的供給。而同一批兩輪車外送供給也會在送餐與載客之間切換。因此，取得這些供給的成本已相當顯著地上升。我們也把原本某種程度上放在消費者端的補貼/激勵。轉移到外送端，以抵消這種情況。

The good news as it relates to Brazil is that we continue to hold our share in Brazil but you just see kind of share move from the mobility side of the business to the delivery side of the business because it's a period of time where there's a lot of investment going there. I think the good news, if you want to call that, is our two-wheeler business is one of our newer businesses. The margins in that business are quite low. So it's certainly not hitting the bottom line, but it is affecting trip volumes and it's actually competition outside of our space that is affecting those trip volumes.

就巴西而言，好消息是我們仍持續維持在巴西的市占，但你會看到市占在業務內部從出行端移向外送端，因為這段期間有大量投資流向那裡。我想所謂的好消息（如果你要這麼稱呼）是，我們的兩輪車業務是較新的業務之一。該業務的利潤率相當低。因此它當然不會衝擊最終盈餘，但它確實影響行程量，而且實際上是我們業務範圍之外的競爭在影響這些行程量。

As it relates to Waymo, listen, Waymo is a very, very important partner of ours, and we continue to operate in Austin and Atlanta. We believe we'll continue to operate next year in those marketplaces. It's a terrific product and the on-the-ground partnership continues to be very strong. At the same time, we want to make sure that we're not dependent on one partner. And we're absolutely seeing a plethora of newer players in the AV ecosystem, just like you see in the foundation model space. And while we continue to provide a great service with Waymo in Austin, Atlanta, we'll continue to build our services with our other players as well.

至於 Waymo，聽著，Waymo 是我們非常、非常重要的合作夥伴，我們也持續在奧斯汀與亞特蘭大營運。我們相信明年也會繼續在這些市場營運。這是一個很棒的產品，而在地端的合作關係也持續非常穩固。同時，我們也希望確保不會依賴單一合作夥伴。我們確實看到自動駕駛（AV）生態系中出現大量新玩家，就像你在基礎模型領域看到的一樣。因此，當我們持續在奧斯汀、亞特蘭大與 Waymo 提供優質服務的同時，也會持續與其他玩家一起建立我們的服務。

We'll be in 15 markets by the end of the year. Next year, it will be many, many more markets than that. And just as a reminder of our scale, AVs are doing kind of hundreds of thousands of trips per week. We're at 300 million kind of trips per week as well. So it's even less than 0.5% of our overall trip volume. You compare that again to the foundation model space, people estimate that 20% of search now has gone to AI, 40% of users are using AI search of one way or the other.

我們在今年年底前會進入 15 個市場。明年將會是遠遠多於這個數量的市場。再提醒一下我們的規模：自動駕駛車（AV）每週大概執行數十萬趟行程。而我們每週大約是 3 億趟行程。所以這甚至不到我們整體行程量的 0.5%。再把這和基礎模型領域相比，有人估計現在 20% 的搜尋已轉向 AI，40% 的使用者以某種方式在使用 AI 搜尋。

So the penetration of kind of physical AV is going to be slower. It's going to some extent, be more deliberate. It's way below where it is at AI. Regulations are a consideration there. And we have time to develop the partnerships with our other set of partners to create a competitive playing field with attractive commercials.

因此，實體自動駕駛（AV）的滲透會更慢。在某種程度上會更審慎、更循序漸進。它遠低於 AI 的滲透程度。法規是其中一項考量。而我們也有時間與另一組合作夥伴發展合作關係，打造具競爭性的賽局環境，並形成具吸引力的商業條款。

John Colantuoni, Jefferies.

John Colantuoni，Jefferies。

Can you talk to the evolving regulatory environment in autonomous vehicles and how you see policy playing into the pace of adoption and geographic expansion? And second, regarding overhead expenses, talk about what has allowed you to temper headcount additions relative to your initial plan and the framework you're using to make decisions about reinvesting these tailwinds versus dropping them to the bottom line?

你能談談自動駕駛車的法規環境如何演變，以及你如何看待政策在採用速度與地理擴張上的影響嗎？第二，關於管理費用（overhead expenses），請談談是什麼讓你們能夠相對於最初計畫，放緩人力（headcount）增加的速度；以及你們用什麼框架來決定要把這些順風（tailwinds）再投資，或是直接落到最終盈餘（bottom line）？

Absolutely. John, as it relates to regulation, listen, we are a highly regulated business. We routinely talk to lawmakers, whether they're governors, mayors, council members who have real concerns about both the effect of AI and AVs with their constituencies. And some of these concerns are real. There are concerns about job loss, there are concerns about safety, there are concerns about congestion.

當然。John，就法規而言，聽著，我們是一個高度受監管的業務。我們經常與立法者溝通，不論是州長、市長、議員等，他們都對 AI 與自動駕駛車（AV）對其選民的影響有真實的疑慮。而其中一些疑慮確實存在。有人擔心失業，有人擔心安全，也有人擔心壅塞。

And while AVs have been incredible in the markets in which we've introduced them, there also have been -- they've had their fair share issues, whether it's safe driving through school zones or next to school buses or how they interact with emergency response vehicles or how they react to large power failures where the traffic lights aren't working.

而且，雖然自動駕駛車在我們導入的市場表現非常出色，但也確實——它們也遇到過不少問題：例如在學校區域或校車旁的安全行駛、它們如何與緊急應變車輛互動、或在大規模停電導致號誌失靈時它們如何反應。

These are real issues that have to be discussed. And like, for example, D.C. is talking about this stuff, and we can't have AVs blocking the streets of D.C. when there's a Presidential or Vice-Presidential motorcade going through. Like these are real issues. And we've got to have the proper dialogue with all the constituencies to make sure that any new law reflects the needs of all these stakeholders.

這些都是真實且必須討論的議題。例如，華盛頓特區（D.C.）就在討論這些事情；當總統或副總統車隊通行時，我們不能讓自動駕駛車把 D.C. 的街道堵住。這些都是實際問題。我們必須與所有利害關係人進行適當對話，確保任何新法都能反映這些利害關係人的需求。

And we see sometimes the results of like trying to go too fast and like some of these AI companies with data centers, they were kind of pushing through, you could argue too quickly with NDAs, et cetera. And there's been a huge public blowback against it. And what we think is you need to have the dialogue, you need to have smart regulation and dialogue with our shareholders so you can actually enable innovation going forward, and we can kind of drive AV regulation in a way that's sustainable that doesn't kind of have the same kind of blowback that you're seeing in AI.

而我們有時也看到那種試圖推進得太快的結果；像一些 AI 公司在資料中心方面，他們某種程度上在推動過程中——你可以說——透過 NDA 等方式推得太快。結果引發了巨大的公眾反彈。我們的看法是，你需要對話、需要聰明的監管，也需要與股東對話，這樣才能真正促進未來的創新；同時我們也能以可持續的方式推動 AV 監管，不致出現你在 AI 領域看到的那種反彈。

So sometimes you got to slow down to drive sustainable regulation. We're very much pro AV, but we want any regulation to kind of address the needs of stakeholders and any model that we have is a model that truly lasts.

所以有時候你必須放慢腳步，才能推動可持續的監管。我們非常支持 AV，但我們希望任何監管都能回應利害關係人的需求，而且我們所採用的任何模式都是真正能長久延續的模式。

Great. I'll take the question on the overhead and headcount. So what I'd start by saying is we have now shown a track record of being disciplined on headcount addition over multiple years. But as we have been going through this year, there are a couple of themes that we have been striking at the company. The first one is the investments we're making in AI, they should result in productivity gains, and that should allow us to be moderating some of the headcount additions.

好的。我來回答關於管理費用與人力編制的問題。我首先要說的是，我們已經在多年期間展現出在人力增加方面保持紀律的紀錄。但在今年的推進過程中，公司有幾個我們一直在著力的主題。第一個是我們在 AI 上的投資，應該會帶來生產力提升，這應該能讓我們放緩部分人力增加。

And then the second piece is really focus on organizational effectiveness and how we operate. So I'll take the first one and give you a little bit more color first. On AI, we are very early, but what we are seeing is that we are able to cost efficiently deliver some productivity lifts with developers. We are seeing, at the moment, near 100% adoption with our engineers on AI-based coding tools.

第二個重點則是聚焦於組織效能，以及我們的營運方式。我先從第一點多補充一些細節。在 AI 方面，我們還在非常早期，但我們看到的是：我們能以具成本效率的方式，為開發人員帶來一些生產力提升。目前我們看到工程師對 AI 程式碼工具的採用率接近 100%。

And for the measurement that we are looking at right now, we are seeing a doubling in the code output per engineer. So there is a lot more nuance to this, and we are being quite smart internally on how we're measuring this and not getting ahead of our skis. But at a minimum, it has allowed us to moderate the pace with which we are adding headcount there.

就我們目前正在觀察的衡量指標而言，我們看到每位工程師的程式碼產出翻倍。因此其中有更多細微差異，我們在內部對如何衡量這件事相當精明，也不會操之過急。但至少，它讓我們得以放緩在那邊增加人力編制的速度。

Then there are other areas where you have very discrete investments in AI. Customer support is a clear area where we should be able to up the quality of our customer support interactions as well as reduce the cost of our effort there. And then on the marketing side as well, we are making some investments here. So there's a broad-based effort that should allow us to moderate the pace of hiring.

接著還有其他領域，你會看到非常明確的 AI 投資。客戶支援是一個很清楚的領域，我們應該能提升客戶支援互動的品質，同時也降低我們在那方面投入的成本。另外在行銷端，我們也在這裡做一些投資。因此這是一個廣泛的努力，應該能讓我們放緩招募的步伐。

Then on the organizational effectiveness side, from time to time, we will review how our organizations are structured. And through the quarter, we were able to surgically, in a couple of organizations, cut headcount by about 10% to 20%, and that drives some savings as well, although modest in the grand scheme of things.

再來在組織效能方面，我們會不時檢視組織架構的設計方式。在本季期間，我們能夠在幾個組織中以精準方式將人力編制削減約 10% 到 20%，這也帶來一些節省，雖然就整體而言影響不大。

I think the objective for us here is to really remain disciplined. And then as we -- if and wherever we find some savings that allows us to reinvest some of those savings back into driving our marketplace and delivering value to our customers, whether it's consumers or earners.

我認為我們在這裡的目標是要真正保持紀律。然後當我們——如果以及在任何我們找到一些節省之處時——就能把其中一部分節省再投資回去，用於推動我們的市集並為客戶創造價值，不論是消費者或是賺取收入的一方。

Justin Post, Bank of America.

Justin Post，美國銀行。

Dara, in the prepared remarks, you talked about $10 billion of investment in AVs. How do you think that flows through to the income statement over time? Any thoughts on margin impact of that? And second, maybe you could comment a little bit on how you think about Lucid's ability to make those vehicle commitments given some recent news in the press.

Dara，在事先準備的發言中，你提到對自動駕駛車（AV）投資 100 億美元。你認為這會如何隨時間反映到損益表上？對利潤率的影響有什麼看法嗎？第二個問題，也許你可以談談，鑑於近期媒體的一些消息，你如何看待 Lucid 履行那些車輛承諾的能力。

And then Balaji, if you want to comment at all on take rates. I think bookings were up 22% constant currency, revenues up 19%. Just some high-level thoughts on that.

另外 Balaji，如果你願意的話，也請談談抽成率（take rates）。我想預訂量以固定匯率計上升 22%，營收上升 19%。對此有沒有一些高層次的看法？

I'll let Balaji talk to the investment in AV and margins and then maybe the take rate as well, and I'll talk about Lucid.

我會讓 Balaji 談談 AV 投資與利潤率，然後也許也談談抽成率；我來談 Lucid。

Okay. Great. So Justin, just for context, the $10 billion of AV investments we're talking about here, this is investment over a multiyear period. And most of the investments we're talking about, you have seen us announce behind the partners that we've been announcing over the last year or so. When you dig into this, there are two distinct kinds of investments we're talking about.

好的。很好。所以 Justin，先提供一些背景，我們這裡談的 100 億美元 AV 投資，是跨多年期間的投資。而我們所談的大部分投資，你已經看到我們在過去一年左右，隨著我們宣布的合作夥伴而對外公布。深入來看，這裡我們談的是兩種不同類型的投資。

The first one is investments in our AV software partners. These typically tend to be equity investments with clear milestones that help us get road map visibility and position us front of the line for commercialization with these partners. And the additional benefit for our partners is that our anchor role helps catalyze external fundraising for those partners as well. And so far, for every dollar that we have invested, our partners have been able to raise an additional $2.50 from other investors. So that's the first piece.

第一種是對我們的 AV 軟體合作夥伴的投資。這些通常是股權投資，並設有明確的里程碑，能幫助我們掌握產品路線圖的能見度，並讓我們在與這些合作夥伴商業化時取得優先位置。對合作夥伴而言，額外的好處是：我們作為錨定投資人的角色，也有助於催化他們向外部募資。到目前為止，我們每投資 1 美元，合作夥伴就能從其他投資人那裡再募到額外 2.50 美元。這是第一部分。

The second piece is really focused on using our balance sheet selectively to bootstrap the AV infrastructure on the ground. And this can be in the form of support for fleet ops, for real estate or for the OEMs who need some offtake commitments as we go in and deploy on these partners. We talked about 120,000 vehicle commitments that we are looking to deliver over the next few years, and a lot of that would be sort of the example of what we're talking about here.

第二部分則是聚焦於選擇性地運用我們的資產負債表，來啟動地面端的 AV 基礎設施。這可以是以支援車隊營運、不動產，或是對於在我們與這些合作夥伴進場部署時需要一定承購承諾（offtake commitments）的 OEM 提供支持等形式呈現。我們談到未來幾年要交付的 120,000 輛車承諾，其中很大一部分就是我們在這裡所說的例子。

Now all of this, as Dara has said earlier as well today, this is designed to catalyze the ecosystem. In parallel, we are very, very actively also looking at financializing the ecosystem, and we are working with third-party financial sponsors to look at what sort of vehicles make sense here. So it's not just Uber's balance sheet that's working on this, but we can turbocharge this with that sort of additional support.

如同 Dara 今天稍早也提到的，這一切的設計目的都是為了催化整個生態系。同時，我們也非常、非常積極地在推動生態系的金融化，並與第三方金融贊助方合作，評估在這裡什麼樣的投資載體（vehicles）是合理的。因此不只是 Uber 的資產負債表在運作，我們也能透過這類額外支持來加速推進。

In terms of the P&L versus cash flow impacts, I would say -- we will give you more visibility into that as we go. The closer we get to deployment and scale out, there will be a P&L impact, and we'll size that for investors clearly as we have historically done. Do you want to touch on Lucid and then I can come back to the other questions?

至於損益表（P&L）相對於現金流的影響，我會說——我們會在推進過程中提供更多能見度。越接近部署與規模化擴張，損益表就會受到影響；我們也會像過去一樣，清楚地為投資人界定其規模。你要不要先談談 Lucid，然後我再回來回答其他問題？

Yes, definitely. As it relates to Lucid, they did announce their latest quarter. Silvio has come on as a new CEO. And listen, he's taking some bold steps to go back to the fundamentals refactoring the cost base of the business, really focusing on the quality of the product. These can be tough kind of actions, but they're necessary, and we think positive actions.

好的，當然。就 Lucid 而言，他們確實公布了最新一季的業績。Silvio 已上任成為新任執行長。坦白說，他正在採取一些大刀闊斧的措施，回到基本面，重新調整業務的成本結構，並真正聚焦在產品品質上。這些行動可能很艱難，但它們是必要的，我們也認為是正向的作為。

And I've talked with Silvio, we remain very close with Lucid and kind of the B2B program that they are building and their program with us as it relates to AV are highly strategic, and it's a big order with kind of guaranteed volume that we know we can monetize as well with a car that's -- a beautiful car that can come in at kind of at the 70,000, 80,000 range. So it's something that we're quite encouraged by.

我也和 Silvio 談過；我們與 Lucid 仍保持非常緊密的合作。他們正在打造的 B2B 計畫，以及他們與我們在 AV 相關的合作計畫，都具有高度策略性；而且這是一筆大訂單，帶有某種保證量，我們也知道可以將其變現——同時那是一款很漂亮的車，價格大概可落在 70,000 到 80,000 美元區間。因此這件事讓我們相當受到鼓舞。

The Nuro and Lucid teams are working together very, very closely and the integration of Nuro's AI into kind of Lucid's driver training on the vehicle in terms of the vehicle weight, in terms of tying into the APIs, all of that is going well, and it's something that we're watching. And then, of course, Lucid is backed by the Public Investment Fund. They are a big investor of ours.

Nuro 與 Lucid 的團隊正在非常、非常緊密地合作；Nuro 的 AI 整合到 Lucid 車輛的駕駛訓練中——包括車重方面、以及與 API 的串接等——一切進展順利，我們也在持續關注。當然，Lucid 也有公共投資基金（Public Investment Fund）的支持。他們也是我們的重要投資人。

They have a terrific Board representative on our Board as well. And the Public Investment Fund is a definition of a long-term fundamental investor that has held Uber for years and years and continues to have confidence in and back Lucid as well. So we think the combination of ourselves, Nuro and then the Public Investment Fund backing Lucid, along with the actions that Silvio is taking are kind of the right formula for them to deliver on the commitments that we have on the books with them.

他們也在我們董事會中有一位非常出色的董事代表。公共投資基金可說是長期基本面投資人的典型，持有 Uber 多年並持續對我們有信心，同時也支持 Lucid。因此我們認為：我們自身、Nuro，以及公共投資基金對 Lucid 的支持，再加上 Silvio 正在採取的行動，這樣的組合是讓他們能履行我們與他們已簽訂承諾的正確配方。

On the revenue take rate, so again, remember that we have talked about a business model change in the UK which is primarily an impact on our mobility business. So if you were to look at our revenue margin for delivery, it's largely stable. So I'll set that aside, and I'll touch on mobility, where you are seeing a nearly 500 basis points decline year-on-year. Of that 500 basis points, about 400 basis points is entirely related to this UK business model change, and it's an optical impact. It moves cost from cost of revenue.

關於營收抽成率（revenue take rate），再提醒一次，我們談過英國的商業模式變更，這主要影響我們的移動出行（mobility）業務。因此如果你看外送（delivery）的營收利潤率，基本上大致穩定。我先把那部分放一邊，來談 mobility：你會看到年對年接近 500 個基點的下滑。在這 500 個基點中，大約 400 個基點完全與英國商業模式變更有關，這是呈現上的影響（optical impact）。它把成本從營收成本（cost of revenue）中移出。

And other than that, it's really deliberate investments that we have been talking about, largely a function of some investments in our lower-cost offerings, the sort of investments we're talking about for Moto in Brazil. And what I would ask investors to focus on is not necessarily the take rate we're reporting on revenues, but the net sort of take rate that is disclosed in our 10-Q filing that will largely show take rate remaining broadly stable.

除此之外，主要就是我們一直在談的、刻意為之的投資，很大程度上是因為我們在一些低成本產品上的投入，例如我們在巴西 Moto 所談到的那類投資。我希望投資人關注的未必是我們在營收上所報告的抽成率，而是我們在 10-Q 申報文件中揭露的淨抽成率（net take rate），那會更能顯示抽成率整體上仍大致維持穩定。

And then from an operating income standpoint, our mobility operating income margin remains very strong at 7.6% as well. So that's all the risk to it in terms of the take rate movement here.

接著從營業利益的角度來看，我們的移動業務營業利益率同樣維持非常強勁，達到 7.6%。因此，就此處抽成率（take rate）變動而言，風險大致就是這些。

Mark Mahaney, Evercore.

Mark Mahaney，Evercore。

I want to ask an AI question and a capital allocation question. There's this line in here about AI making Uber more intuitive for consumers. I absolutely would think that AI is leading to greater personalization across a series of services. And it would show up in kind of better conversion rates, more spend per consumer, et cetera.

我想問一個 AI 的問題，以及一個資本配置的問題。這裡有一句提到 AI 讓 Uber 對消費者而言更直覺。我完全認同 AI 正在推動一系列服務更高度的個人化。而這會反映在更好的轉換率、每位消費者更高的支出等等。

Dara, is there anything you could kind of quantify there about how much better the Uber process from a consumer perspective can be because of AI? And then Balaji, this comment about steadily returning to a more normalized level of activity in terms of the capital allocation share repurchases. Can you just kind of put a little timing on that? Like is that steadily return over a year? Are we talking years or quarters?

Dara，關於 AI 能讓 Uber 的消費者流程變得更好，你能否量化一下能改善多少？然後 Balaji，關於資本配置中的庫藏股回購，你提到會穩步回到更正常的活動水準。你能否給一個大概的時間點？例如這個「穩步回到」是指一年內？我們談的是以年為單位，還是以季為單位？

Yes, absolutely, Mark. So when we think about AI, there are a number of functions that it can serve. One is just to make kind of the consumer experience better, easier. And one example of that is Cart Builder for us, where you can introduce kind of a shopping cart, whether you take a picture of a dish or scribble down a recipe, the AI actually interacts with you to kind of build out a cart.

當然可以，Mark。我們在思考 AI 時，它可以扮演多種功能。其中之一就是讓消費者體驗更好、更容易。我們的一個例子是 Cart Builder（購物車建立器）：你可以引入一個購物車，不論是拍一張菜色照片或隨手寫下食譜，AI 會與你互動，協助把購物車內容建構出來。

And the effect there is consumers love it, the ones who use it, but also the size of those carts is often twice the size of kind of non-AI built carts. So that's one where your average order size, you kind of drive consumer delight and you are driving average order size as well.

其效果是，使用它的消費者非常喜歡；而且這些購物車的金額/規模往往是非 AI 建立購物車的兩倍。所以這是一個同時提升平均訂單金額、帶來消費者驚喜體驗的例子，也確實推動了平均訂單金額。

Second for AI is -- and this isn't kind of, call it, foundation model work, but just much larger models can be much smarter about making predictions as it relates to consumer behavior based on a much broader array of signals that they're taking in. You take a bunch of signals from consumers. Now these models can take signals based on history. They can take real-time signals, they can take behaviors across the platform. And we can then show those consumers much more relevant information.

第二個 AI 的面向是——這不一定是所謂的基礎模型（foundation model）工作——而是更大型的模型能更聰明地做預測，特別是在基於更廣泛的訊號來判斷消費者行為方面。你會從消費者取得一堆訊號。現在這些模型可以根據歷史訊號。也可以採用即時訊號，還能納入跨平台的行為。接著我們就能向這些消費者呈現更相關的資訊。

So it might be ads that are more highly targeted that can convert better, that help our merchants monetize more effectively. It might be just the sort order for organic results on restaurants so that you find that kind of local gem. It might be deal ranking algorithms that get deals in front of you so that you can save money while you're kind of getting what you want as well or it can be AI suggesting that a particular item that you have in your shopping cart is likely to be out of stock. So maybe you'll pick another item as well.

例如可能是更精準投放、轉換更好的廣告，幫助我們的商家更有效地變現。也可能是餐廳自然排序（organic results）的排序方式，讓你找到那家在地寶藏店。也可能是優惠排序演算法，把優惠更好地呈現在你面前，讓你在得到想要的同時也能省錢；或者 AI 建議你購物車中的某個品項很可能缺貨。那你也許就會改選另一個品項。

All of these either drive conversion or improve the fundamental experience of the consumer. I will stress that we are very, very, very early in terms of the development of the path. And Mark, kind of a lot of people expect these technologies that are revolutionary and AI is revolutionary to have some giant hit.

這些都能提升轉換，或改善消費者的核心體驗。我要強調的是，我們在這條發展路徑上仍然非常、非常、非常早期。而 Mark，很多人期待這些具革命性的技術——AI 的確是革命性的——會出現某個巨大的「爆款」。

But the fact is that the way that we operate is we optimize the system quarter after quarter after quarter, year after year after year, and you should expect AI to contribute to average order size, the quality and reliability of our service as well and then putting kind of the right product in front of you at the right time.

但事實是，我們的運作方式是每一季、每一季、每一季地優化系統，年復一年地持續優化；你應該期待 AI 會在平均訂單金額、服務品質與可靠性上做出貢獻，並在正確的時間把正確的產品呈現在你面前。

So like another example is three fourth of our rides on Uber happen via a personalized destination suggestion. three fourth of the time, we're actually guessing where you're going to go based on your history, and we're getting it right and don't require any typing whatsoever. That's another small example of what AI can do. And I wouldn't look for like one giant hit from AI. It's going to be thousands of small hits and improvements to our ecosystem that's going to drive, we think, growth for the foreseeable future.

再舉個例子：我們在 Uber 上有四分之三的行程，是透過個人化目的地建議完成的。四分之三的時間，我們其實是根據你的歷史在猜你要去哪裡，而且猜對了，完全不需要你打字。這也是 AI 能做到的一個小例子。我不會期待 AI 帶來單一的巨大爆款。它會是成千上萬個小幅命中與對我們生態系的改進，我們認為這將在可預見的未來推動成長。

I'll take the question on the buybacks. So just as a reminder, we have generated about just a little bit over $10 billion in free cash flows over the trailing 12 months. And what we've said historically is that we want to be deploying about 50% of our free cash flows towards buybacks. Where we are so far halfway through the year is that we have bought back about $3.5 billion of our stock this year. And we tactically pivoted quite heavily towards M&A in the second quarter.

我來回答回購的問題。提醒一下，我們在過去 12 個月（trailing 12 months）產生了略高於 100 億美元的自由現金流。而我們過去的說法是，希望把大約 50% 的自由現金流用於回購。截至目前、今年過了一半，我們今年已回購約 35 億美元的股票。而在第二季，我們在策略上相當大幅度地轉向併購（M&A）。

And largely, that was a function of our market purchases of Delivery Hero stock. We deployed about $4 billion of capital in the second quarter to ensure that we were in a place to be able to act fast on an opportunity that existed there. Now that we are through that announcement, we are steadily looking to rebuild our share repurchase levels. So to answer your question, Mark, I don't think this is a year away. We are talking about months, not quarters.

主要原因是我們在市場上買入 Delivery Hero 股票。我們在第二季投入約 40 億美元資本，以確保我們能在那個機會出現時快速行動。現在我們已經完成該公告之後，我們正穩步地把股票回購水準拉回來。所以回答你的問題，Mark，我不認為這會是一年之後的事。我們談的是以月為單位，而不是以季為單位。

Nikhil Devnani, Bernstein.

Nikhil Devnani，Bernstein。

I had a couple of separate ones, please. So first on M&A, just given Delivery Hero, you've done bolt-on deals in the past across the US and other markets. What have the learnings been on integration? What worked well? What were some mistakes to avoid as you look to integrate Delivery Hero, which is a bigger and more complex asset?

我有幾個彼此獨立的問題，謝謝。首先關於併購（M&A），考量到 Delivery Hero，你們過去在美國及其他市場做過一些補強型（bolt-on）交易。在整合方面有哪些學到的經驗？哪些做得很好？在你們準備整合 Delivery Hero（這是一個更大、更複雜的資產）時，有哪些錯誤需要避免？

And then separately, in markets where you have AV products that are competing with Uber and not on the Uber network yet, what have you generally observed about mobility cohort engagement and retention behavior?

另外，在你們所處的一些市場中，有自動駕駛（AV）產品在與 Uber 競爭、但尚未接入 Uber 網路；就你們的觀察，移動出行用戶群（mobility cohort）的參與度與留存行為通常呈現什麼樣的情況？

Thanks, Nikhil. I'll take the Delivery Hero question, and then Dara will take the next one. So the -- what we have seen historically when we have made these kind of acquisitions, albeit at a much smaller scale historically, is that you need to have a clear integration hypothesis before you even pursue the deal. Where we are with Delivery Hero in particular, is that we have very high confidence on our integration plan.

謝謝，Nikhil。我來回答 Delivery Hero 的問題，接著 Dara 會回答下一題。我們過去在進行這類收購時——雖然歷史上規模小得多——觀察到的一點是：在你甚至追求這筆交易之前，就必須先有清晰的整合假設（integration hypothesis）。就 Delivery Hero 而言，我們對整合計畫非常有信心。

We are looking to primarily migrate onto an existing global tech platform supported by a proven integration playbook, a realistic time line, and we are holding a pretty high bar on a disciplined underwriting of the synergies that we have communicated to the Street.

我們主要會遷移到既有的全球技術平台上，並由已被驗證的整合作戰手冊（integration playbook）、務實的時間表所支撐；同時，我們也對已向市場溝通的綜效（synergies）採取相當高標準、嚴謹的紀律性評估（underwriting）。

We have already looked at the asset in quite a lot of detail. where we are is we know that we operate a single tech platform for our delivery business, whereas Delivery Hero's business with the exception of Baemin right now is on one stack, but there is a lot of opportunity for that all to be collapsed and migrated onto a modern tech stack like Uber's.

我們已經相當深入地檢視這項資產。

Then secondly, from an execution timetable standpoint, assuming we close in the second half of 2027, we'll spend 2028 on planning and development before executing the primary migrations in 2029, and that's an appropriate time line for a transaction of this size.

其次，從執行時程來看，假設我們在 2027 年下半年完成交割，我們會在 2028 年進行規劃與開發，並在 2029 年執行主要遷移；以這種規模的交易而言，這是合適的時間表。

And then finally, what we are talking about on synergies here, they're rooted in areas where we have a very high degree of confidence, moving to a common tech platform, platform overlap infrastructure, duplicative roles and services and then finally, consolidating shared services such as payments and cloud infrastructure. So that's the plan here, and we remain committed to delivering the synergies that we communicated a couple of weeks ago.

最後，關於我們在此所談的協同效益，它們根植於我們非常有信心的幾個領域：遷移到共同的技術平台、平台重疊的基礎設施、重複的職能與服務，最後是整併共用服務，例如支付與雲端基礎設施。這就是我們的計畫，我們也仍然致力於實現我們在幾週前所溝通的協同效益。

Yes. And as far as operating trends in some of the more mature AV market, they're really strong and kind of the strength in cohorts and new users shows in our trip growth, specifically, for example, our trip growth in San Francisco, L.A. and Phoenix accelerated in Q2 versus Q1. And our category position in these markets is actually higher today than it was a year ago.

是的。至於一些較成熟的自動駕駛（AV）市場的營運趨勢，表現非常強勁；而分群（cohorts）與新用戶的強勢也反映在我們的行程成長上。具體而言，例如，我們在舊金山、洛杉磯與鳳凰城的行程成長在第二季相較第一季有所加速。而且我們在這些市場的品類地位（category position）如今實際上比一年前更高。

And kind of the comparison that I would make is with Google Search. The penetration of search as it relates to AI is way, way, way higher than the penetration of AV has been. Again, AV is a physical good, there's regulation, et cetera, all the reasons that we've talked about previously. But even with a 20% penetration of AI search on Google, Google searches are actually up on a year-on-year basis because AI is increasing the overall market as well.

我會做的比較是與 Google 搜尋。搜尋在 AI 相關的滲透率遠遠、遠遠、遠遠高於 AV 的滲透率。再說一次，AV 是實體產品，還有法規等因素，都是我們先前談過的原因。但即便 Google 的 AI 搜尋滲透率達到 20%，由於 AI 也在擴大整體市場，Google 的搜尋量仍然在年對年基礎上成長。

And what you saw with Google was they were later to the market, but the power of their distribution, the power of their brand allowed them to be a significant player and a winner in search. And we think the same is going to be true for us. So as kind of we sit here, the trends in those markets are strong. And I think the way that you've seen these markets kind of develop shows that our distribution, our brand can win on a long-term basis.

你在 Google 身上看到的是，他們較晚進入市場，但其通路分發能力與品牌力量，使他們成為搜尋領域的重要玩家與贏家。我們認為同樣的情況也會發生在我們身上。因此，就我們目前所處的位置而言，這些市場的趨勢很強勁。而且我認為，這些市場的發展方式顯示，我們的通路與品牌能在長期勝出。

I think I'll add that the opportunity we have, especially in the US is far, far broader than the markets where AVs operate today. Only 30% of our US gross bookings and 25% of our profits are coming from the top 20 cities and the long tail of thousands of other cities and suburbs, they will remain a primary growth and profit engine for Uber for many years to come. So we are continuing to invest in both humans and in AVs. And you should expect that the US remains a great market for these teams.

我想補充的是，我們的機會，尤其在美國，遠遠比目前 AV 運行的市場更廣泛。我們在美國的總預訂額（gross bookings）只有 30% 以及 25% 的利潤來自前 20 大城市；而由數以千計的其他城市與郊區構成的長尾市場，未來多年仍將是 Uber 的主要成長與獲利引擎。因此，我們會持續同時投資於人力與 AV。你們應該預期，美國仍將是這些團隊的一個很棒的市場。

Sarah, we'll take our last question please.

Sarah，請讓我們進行最後一個問題。

Ross Sandler, Barclays.

Barclays 的 Ross Sandler。

Dara, the new AV lab, just what are those guys working on that complements what your partners are working on? And then BK, the delivery business has these kind of small acquisitions kind of we're lapping one in Turkey and then we're adding these two in the third quarter. It looks like organic growth is picking up. So could you just talk about the organic growth that you're seeing in 2Q and what's implied in 3Q for the delivery?

Dara，新的 AV 實驗室，他們正在做哪些工作，能夠補足你們合作夥伴正在做的事情？另外 BK，外送業務有一些小型併購，我們在土耳其那一筆正在進入比較基期（lapping），而第三季又新增這兩筆。看起來有機成長正在回升。所以你能否談談你們在第二季看到的有機成長，以及第三季外送業務所隱含的趨勢？

Yes, Ross. So what we're seeing as it relates to the development of AV models and physical AV in particular, is that end-to-end models are replacing the heuristics approach, like a bunch of these companies, including Waymo, including Nuro, et cetera, they have been building AV for years and years, but it was based on heuristics. It was based on if-then kind of logical functions.

好的，Ross。就 AV 模型的發展、尤其是實體 AV 而言，我們看到端到端（end-to-end）模型正在取代啟發式（heuristics）方法。像許多公司，包括 Waymo、Nuro 等，已經打造 AV 多年，但主要是基於啟發式方法。也就是基於 if-then 這類邏輯函數。

And that is quickly being replaced by end-to-end models that take in enormous amounts of data and then make decisions as humans do. And what we want to do is position Uber with AV labs to help accelerate the development and the training of these models on an L4 basis efficiently with real kind of data from rideshare-specific scenarios.

而這正迅速被端到端模型所取代；端到端模型會吸收海量資料，然後像人類一樣做出決策。我們希望透過 AV 實驗室，讓 Uber 站在有利位置，協助加速這些模型在 L4 層級的開發與訓練，並以來自共乘特定情境的真實資料來高效率地進行。

So we're building out hundreds of cars that are riding in rideshare-specific scenarios with robotaxi-grade kind of sensors, and we're collecting kind of a super set of data that then we can provide to all of our partners. One of the issues as it relates to having different partners in the ecosystem is each partner has to collect the unique data sets to go out and kind of acquire all of the tail data that you need to train on to make sure that your AV driver is safe.

因此，我們正在打造數百輛車，讓它們在共乘特定情境中行駛，配備機器人計程車等級的感測器，並收集一個「超集合」的資料，之後我們可以提供給所有合作夥伴。在生態系中擁有不同合作夥伴的一個問題是，每個合作夥伴都必須收集獨特的資料集，去取得你需要用來訓練的各種長尾資料，以確保你的 AV 駕駛是安全的。

We can go out, collect one set of data that's rideshare-specific at very high fidelity with advanced sensors, and we can provide that data to all of our partners so that we bring the benefit of scale to AV development. And we are seeing that development accelerate. We want to be a part of that acceleration and AV labs is kind of what we think is a very, very strong partner and accelerator to the development of the entire AV ecosystem.

我們可以出去收集一套共乘特定的資料，以先進感測器在非常高的擬真度（fidelity）下取得，並把這些資料提供給所有合作夥伴，讓我們把規模效益帶進 AV 的開發。而我們也看到這項開發正在加速。我們希望成為這種加速的一部分，而 AV 實驗室正是我們認為能成為整個 AV 生態系非常、非常強的合作夥伴與加速器。

Yes. I'll take the M&A question. So just as a reminder, we closed our acquisition in Turkey, Trendyol Go in the back half of June 2025. So we are rolling off of that in Q3. We are lapping that acquisition fully. At the same time, we just closed the second acquisition in Turkey earlier this month, which was the acquisition of Getir. That will have a positive contribution to growth in Q3 and beyond. And we have a smaller acquisition with Careem reconsolidating as well.

是的。我來回答併購（M&A）問題。提醒一下，我們在土耳其的收購案 Trendyol Go 已於 2025 年 6 月下旬完成交割。因此在第三季，我們會把這部分從比較基期中剔除。我們將完全進入該收購案的比較基期。同時，我們本月稍早剛完成在土耳其的第二筆收購，也就是收購 Getir。這將在第三季及之後對成長帶來正向貢獻。此外，我們也有一筆較小的收購，涉及 Careem 的重新併表（reconsolidating）。

So there's a couple of puts and takes there. On the whole, it is a headwind to delivery reported growth on a net basis, both for gross bookings and trips because Trendyol Go was a lot larger in size than the two acquisitions we're talking about for this quarter.

因此其中有一些此消彼長的因素。整體而言，對外送的「報告口徑」成長（reported growth）在淨額基礎上是逆風，無論是總預訂額或行程數，因為 Trendyol Go 的規模遠大於我們本季所談的另外兩筆收購。

Underneath that, our organic delivery business is accelerating quite nicely. We saw very strong trends in the US We are continuing to see very, very strong trends internationally as well. We gained category position in all of our large markets. And so what we are seeing is an organic acceleration and inorganic handoff from a large acquisition to a small one. But on the whole, what you see should be relatively healthy trends for delivery in the third quarter.

在此之下，我們的有機外送業務正在相當不錯地加速。我們在美國看到非常強勁的趨勢，國際市場也持續非常、非常強勁。我們在所有大型市場都提升了品類地位。因此，我們看到的是有機加速，以及從一筆大型收購到一筆小型收購的無機成長交接。但整體而言，你們在第三季應該會看到外送相對健康的趨勢。

Well, thank you, everyone, for joining the call. Thank you to the entire Uber team for another quarter of really, really strong execution and looking forward to all of the innovation in the industry and everything that we're going to build for you. So we'll talk to you next quarter. Thanks.

好的，感謝各位參與本次電話會議。也感謝整個 Uber 團隊在本季再次展現非常、非常強的執行力；我們也期待產業中的所有創新，以及我們將為各位打造的一切。那我們下季再聊。謝謝。

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

今天的電話會議到此結束。感謝您的參與。您現在可以中斷連線