DoorDash, Inc. (DASH) 2026 Q2 法說會逐字稿

Hello, everyone. Thank you for joining us, and welcome to the DoorDash Q2 2026 earnings call. (Operator Instructions) I will now hand the call over to Weston Twigg. Please go ahead.

大家好。感謝各位加入，歡迎參加 DoorDash 2026 年第二季財報電話會議。(接線員指示) 我現在把電話交給 Weston Twigg。請開始。

Thanks, Connor. Good afternoon, everyone, and thanks for joining us for our Q2 2026 earnings call. I'm pleased to be joined today by Co-Founder, Chair and CEO Tony Xu; and CFO, Ravi Inukonda. We'll be making forward-looking statements during today's call, including without limitation, our expectations for our business, financial position, operating performance, profitability, our guidance, strategies, capital allocation approach, and broader economic environment.

謝謝你，Connor。各位下午好，感謝各位參加我們 2026 年第二季財報電話會議。我很高興今天與共同創辦人、董事長兼執行長 Tony Xu，以及財務長 Ravi Inukonda 一同出席。在今天的電話會議中，我們將作出前瞻性陳述，包括但不限於我們對業務、財務狀況、營運表現、獲利能力、指引、策略、資本配置方式以及更廣泛經濟環境的預期。

Forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from those described. Many of these uncertainties are described in our SEC filings, including our most recent Form 10-K and 10-Q. You should not rely on forward-looking statements as predictions of future events or performance. We disclaim any obligation to update any forward-looking statements except as required by law.

前瞻性陳述受風險與不確定性影響，可能導致實際結果與所述內容有重大差異。其中許多不確定性已在我們向 SEC 提交的文件中說明，包括我們最新的 Form 10-K 與 10-Q。您不應將前瞻性陳述視為對未來事件或表現的預測。除法律要求外，我們不承擔更新任何前瞻性陳述的義務。

During this call, we will discuss certain non-GAAP financial measures. Information regarding our non-GAAP financial measures, including a reconciliation of such non-GAAP measures to the most directly comparable GAAP financial measures may be found in our earnings release, which is available on our Investor Relations website at ir.doordash.com.

在本次電話會議中，我們將討論若干非 GAAP 財務衡量指標。有關我們非 GAAP 財務衡量指標的資訊（包括該等非 GAAP 指標與最直接可比的 GAAP 財務衡量指標之調節表），可於我們的財報新聞稿中查閱；該新聞稿可在我們投資人關係網站 ir.doordash.com 取得。

These non-GAAP measures should be considered in addition to our GAAP results and are not intended to be a substitute for our GAAP results. Finally, this call is being audio webcasted on our Investor Relations website. An audio replay of the call will be available on our website shortly after the call ends.

這些非 GAAP 指標應與我們的 GAAP 結果一併考量，且不旨在取代我們的 GAAP 結果。最後，本次電話會議將於我們的投資人關係網站進行音訊網路直播。電話會議結束後不久，網站上將提供音訊重播。

Operator, I'll pass it back to you, and we can take our first question.

接線員，我把電話交回給你，我們可以開始第一個問題。

(Operator Instructions)

(接線員指示)

Michael Morton, MoffettNathanson.

Michael Morton，MoffettNathanson。

Good evening, guys. Thank you for the questions. I wanted to ask about the grocery business as you've talked about improving the unit economics. From our understanding, there are some grocers on the platforms who are -- your platform specifically, who are paying effectively zero or very low take rates, they came on, looking to see as you could drive demand and how well they could work with DoorDash.

各位晚上好。謝謝讓我提問。我想問一下雜貨業務，因為你們談到正在改善單位經濟效益。依我們的理解，平台上有一些雜貨商——特別是在你們的平台上——實際上支付的抽成率幾乎為零或非常低；他們加入是想看看你們是否能帶動需求，以及他們與 DoorDash 的合作效果如何。

I was wondering if that's the case and then what the opportunity is to reprice these relationships going forward, maybe pushing some of the affordability burden that's lending on DoorDash right now back to the grocers. Thank you.

我想知道是否確實如此，以及未來重新定價這些合作關係的機會為何——也許把目前落在 DoorDash 身上的部分可負擔性負擔，轉回由雜貨商承擔。謝謝。

Yeah. Michael, it's Tony. I can start and feel free to chime in, Ravi. What I would say is like we see extremely strong performance in our grocery business. It's the fast-growing part of our marketplace business and we have very healthy relationships with all of the partners on the platform.

是的。Michael，我是 Tony。我先開始，Ravi 也歡迎補充。我想說的是，我們的雜貨業務表現非常強勁。它是我們市集業務中成長最快的部分，而且我們與平台上所有合作夥伴都維持非常健康的關係。

I mean in terms of the economic relationships, I'm not going to really comment about anyone in particular. I mean what I will say is that when you are the fastest grower in the market for them, and you are their source of growth. Put a different way, we might be 100% of the growth that they see in terms of their actual business.

至於經濟合作條件，我不會針對任何特定對象評論。但我想說的是，當你是他們在市場上成長最快的合作方，而且你是他們的成長來源時，換句話說，就他們實際業務而言，我們可能貢獻了他們所看到成長的 100%。

You certainly have opportunities to grow your business with them as well as improve your relationships with them. I mean I think if you look at our business as a whole, I think one of the things you see from this quarter, and frankly, I think many of the time periods leading up to now, is that there are many sources of improving economics.

你當然也有機會與他們一起擴大你的業務，同時改善與他們的合作關係。我認為如果你看我們整體業務，你會從本季、坦白說也包括此前許多期間看到，我們有很多改善經濟效益的來源。

We've improving unit economics across all of our categories. We have improving unit economics in our restaurant business, too. We have improving unit economics in our different geographies in which we operate. We have increasing adoption of our DashPass program as well as accelerating growth in our ads business.

我們在所有品類的單位經濟效益都在改善。我們的餐廳業務單位經濟效益也在改善。我們在不同營運地區的單位經濟效益也在改善。DashPass 計畫的採用率持續提升，同時廣告業務的成長也在加速。

And I think when you add all of that in, we have a business in which there are many levers in which we can control the kind of financial profile in order to make great investments. Ultimately, we're here always seeking the next best investment.

把這些因素加總起來，我們的業務有許多槓桿可用來掌控財務輪廓，以便進行優質投資。最終，我們始終在尋找下一個最佳投資機會。

It doesn't mean that we always make those investments. But when we see the opportunity, we're always leaning in, that includes all of the work that we're doing in grocery, which we think there's a long runway as well as all the other opportunities in front of us.

這不代表我們總是會做出那些投資。但當我們看到機會時，我們總是會加大投入；這也包括我們在雜貨領域所做的所有工作——我們認為那裡仍有很長的成長跑道——以及我們面前的所有其他機會。

And Mike, just to add, right? like look, I mean, if you take a step back and think about our overall grocery business as well as new verticals, we talked about the fact that we became order volume share leaders in Q4. We've continued to extend that lead.

Mike，我再補充一下，對吧？你看，如果退一步思考我們整體雜貨業務以及新垂直品類，我們曾提到在第四季我們成為訂單量市占的領先者。而且我們持續擴大這個領先優勢。

Two, when you look at the underlying growth in MAUs, which is the number of users that use categories outside of restaurants, that number is growing or the frequency is growing. We talked about the fact in the letter that basket sizes are growing.

第二，當你看 MAU（每月活躍用戶）的底層成長——也就是使用餐廳以外品類的用戶數——這個數字在成長，或是使用頻率在提升。我們在股東信中也提到，客單價（basket size）正在成長。

If you look at our historic cohorts, consumers are using us for more use cases, which is driving overall basket sizes higher. Last call, I think I mentioned the fact that we expect our overall new vertical business to be gross profit positive. We're on track for that in the second half of the year. Look, I mean if we think about the business as a whole, we think about retention, order frequency as well as underlying improvement in unit economics and they're all headed in the right direction for us.

如果你看我們歷史用戶分群（cohorts），消費者在更多使用情境中使用我們，這推動整體客單價提高。上一次電話會議我提到，我們預期整體新垂直品類業務將達到毛利為正。我們正按計畫在今年下半年達成。總之，如果我們從整體業務來看，無論是留存、下單頻率，還是單位經濟效益的持續改善，對我們而言都朝著正確方向發展。

Mark Mahaney, Evercore ISI.

Mark Mahaney，Evercore ISI。

Thanks. I'll ask a question about Deliveroo. You've had now three quarters in a row of kind of accelerating growth, I think, in orders, yes, and in GOV and then I think in revenue. So just peel that back a little bit there -- how many different opportunities you've had, what you've been able to pull, what you've been able to change in order to deliver that better performance? And it's a little hard to tell, but is it also showing up on the bottom line? Are you able to -- have you been finding ways to improve the profitability of Deliveroo as well. Thank you very much.

謝謝。我想問一個關於 Deliveroo 的問題。你們現在已經連續三個季度呈現某種加速成長，我想是在訂單量、GOV，然後我想也包括營收。能否稍微拆解一下——你們有哪些不同的機會、採取了哪些槓桿、做了哪些改變，才帶來更好的表現？另外有點難判斷，但這是否也反映在獲利端？你們是否也找到方法改善 Deliveroo 的獲利能力？非常感謝。

Hey, It's Tony. I can start. What I would say on Deliveroo it really is a story that probably started way back in 2021, when we first made -- our first large acquisition overseas, which was Wolt -- and we've learned a ton, obviously about building our own US business, and we've learned a lot in terms of how Wolt has operated in different geographies across Europe as well as how to integrate the lessons that we've learned as well as the lessons that maybe don't apply into each one of these local geographies.

嗨，我是 Tony。我先開始。關於 Deliveroo，我想說這其實是一個可能早在 2021 年就開始的故事，當時我們首次在海外進行大型收購，也就是 Wolt——我們當然從打造美國本土業務學到很多，也從 Wolt 在歐洲不同地區的營運方式學到很多，以及如何把我們學到的經驗、以及那些可能不適用的經驗，整合並落地到各個在地市場。

So what you're seeing in Deliveroo -- I agree with you, Mark, is just accelerating performance, frankly, across the board. And that's super exciting, because I think, a, it's a validation that our integration work is really working and that the lessons that we've learned in building these marketplace businesses around the world do translate into some of these very meaningful geographies and foundational places where we're just seeing, growth in all of our big international markets.

所以你在 Deliveroo 看到的情況——我同意你，Mark——坦白說，就是整體表現正在加速。這非常令人振奮，因為我認為，第一，這驗證了我們的整合作業確實奏效；而且我們在全球打造這些平台型（marketplace）業務所學到的經驗，確實能夠轉化到一些非常重要的地理區域與基礎市場上，在那裡我們正看到所有主要國際市場的成長。

And this doesn't even include the majority of benefits that we expect to see once we actually finish all of our work on building a single tech stack. So as that work kind of comes more fully online, towards the beginning of next year, we expect to see even more benefits as time goes on.

而這甚至還不包含：一旦我們真正完成打造單一技術堆疊（single tech stack）的所有工作後，我們預期將看到的大部分效益。因此，隨著這項工作在明年年初更完整地上線，我們預期隨著時間推進，將看到更多效益。

And Mark, just to put a final point right? Like when you look at the actual performance of Deliveroo itself, to your point, volume growth or MAU growth or subscription growth, actually, when you look at it on a year-over-year basis, it's been the highest that we've seen in the last couple of years. At the same point, to your second part of your question, we've increased the unit economics as well. But the way in which we're operating the business is very similar, right?

另外 Mark，最後再補充一點，好嗎？就像你提到的，當你看 Deliveroo 本身的實際表現，不論是交易量成長、MAU 成長或訂閱成長，實際上以年對年來看，這是我們過去幾年看到的最高水準。同時，針對你問題的第二部分，我們也提升了單位經濟效益（unit economics）。但我們經營這項業務的方式其實非常相似，對吧？

We're finding great opportunities to drive investment back in selection. Quality subscription is a big area of focus for us. So we're going to continue to invest back in the business. And my expectation is we'll continue to drive higher top line as well as meet the profitability targets that we set out in the last letter.

我們正在找到很好的機會，把投資再投入到選擇（selection）上。高品質訂閱（Quality subscription）是我們重點聚焦的一大領域。所以我們會持續把投資回灌到業務中。而我的預期是，我們會持續推動更高的營收（top line），同時也達成我們在上一封股東信中所設定的獲利目標。

Nikhil Devnani, Bernstein.

Nikhil Devnani，Bernstein。

Hi. Thanks for taking my questions. I'll stick with the theme of international. And I guess broadly, there's, I guess, a common perception that international might be lower quality growth, because maybe you're not number one everywhere or maybe the competitive set now is better funded and more consolidated today, I would love your perspective on that overall sentiment and really how you structurally see the longer-term earnings power or quality of growth out of these international markets relative to the domestic business. And how important is it to have a pure market share number versus more minimum viable scale in these jurisdictions that allow you to then operate well? Thank you.

嗨。謝謝讓我提問。我會延續國際業務這個主題。我想大致上，外界有一種普遍看法，認為國際市場可能是較低品質的成長，因為也許你們不是在每個地方都是第一名，或是如今競爭對手資金更充足、整合度也更高。我很想聽聽你對這種整體情緒的看法，以及你們在結構上如何看待：相較於本土業務，這些國際市場長期的獲利能力或成長品質。另外，在這些司法管轄區（jurisdictions）中，擁有一個純粹的市占率數字有多重要？相較之下，達到能夠良好營運的最低可行規模（minimum viable scale）是否更重要？謝謝。

Yeah. Nikhil, it's Tony. I can start. I think there are a couple of different questions that you're asking. The first is really what do we see happening internationally. Our aspiration is to be the global leader in local commerce. And we think we are best positioned to do that given that we bring the deepest and the broadest portfolio of products in order to serve those audiences. So like you said, it's a game that is played locally. There are no global network effects in these kinds of businesses.

是的。Nikhil，我是 Tony。我可以先開始。我認為你問了幾個不同的問題。第一個是我們看到國際市場正在發生什麼。我們的願景是成為在地商務（local commerce）的全球領導者。我們認為我們最有條件做到這點，因為我們擁有最深、最廣的產品組合，能夠服務這些受眾。所以就像你說的，這是一個在地作戰的遊戲。在這類業務中不存在全球性的網路效應。

And one of the points you raised is the right one where sometimes there isn't an obvious tie between the market position and kind of your economic profile. And that's because it is a minimum viable scale business. That said, though, when I actually look at our current execution, the vast majority of our international business is concentrated in our top 10 markets outside of the US. And in those markets, we are the leader or we are a very strong number two and we're gaining share in all of the markets.

你提到的一點是正確的：有時候市場地位與你的經濟輪廓（economic profile）之間，並沒有明顯的對應關係。這是因為這是一門「最低可行規模」的生意。不過，當我看我們目前的執行狀況時，我們絕大多數的國際業務都集中在美國以外的前 10 大市場。而在那些市場中，我們不是領導者，就是非常強勢的第二名，而且我們在所有市場都在提升市占。

Some of these markets include places like the UK, Italy, Germany, the Nordics, Israel, Canada, I mean, I can keep going, but we like kind of what we see. And it kind of really is a follow-on to the previous question, where Ravi was talking about how we're making improvements fundamentally to the actual core propositions to all the audiences. We're making -- we're offering wider selection, better prices, better quality of delivery in terms of reliability, accuracy and speed, and we're improving our customer service.

其中一些市場包括英國、義大利、德國、北歐、以色列、加拿大——我還可以繼續列，但我們很喜歡我們所看到的狀況。而這其實也延續了上一個問題，Ravi 談到我們正在從根本上改善面向所有受眾的核心價值主張。我們正在——提供更廣的選擇、更好的價格、更高品質的配送（在可靠性、準確性與速度方面），並且改善我們的客服。

And so whenever I see that, and I also see the opportunity for the runway to bring our portfolio of B2B products, which have done really well in the US but even have more opportunity outside of the US, just given the more nascent development of digital technologies in the restaurant and retail categories overseas, I mean, I just think the potential is very, very big.

因此每當我看到這些，同時也看到我們有很長的跑道可以把 B2B 產品組合帶到國際市場——這些產品在美國表現非常好，但在美國以外甚至有更大的機會，因為海外餐飲與零售類別的數位技術發展仍較早期——我就覺得潛力非常、非常大。

And Nikhil, I mean, the results are pretty clear, right? We've talked about the fact that on Deliveroo side, I mean, the growth is accelerating. In fact, when you look at the underlying cohorts, the growth is some of the highest that we've seen over the last couple of years.

而且 Nikhil，我是說，結果其實很清楚，對吧？我們談過 Deliveroo 這邊的成長正在加速。事實上，當你看底層的分群（cohorts），其成長是我們過去幾年看到的最高水準之一。

Even outside of that, when I look at the portfolio, excluding ROO, we are growing, MAUs are growing, the order frequency is growing. Wolt+ in fact, our subscription program in Wolt had one of the best quarters, which is a record quarter in terms of overall paid subscriber growth.

即便不看那部分，當我看整體產品組合、排除 ROO 之後，我們仍在成長，MAU 在成長，下單頻率也在成長。Wolt+——也就是我們在 Wolt 的訂閱方案——在付費訂閱者成長方面，剛剛經歷了表現最好的季度之一，也是創紀錄的一季。

At the same point, it's not just purely about growth for us, right? Like we are improving the unit economics, not just across ROO, but across Wolt as well. When you look at whether it's gross profit or contribution, both of them have continued to improve on a year-over-year basis.

同時，對我們而言也不只是單純追求成長，對吧？我們正在改善單位經濟效益，不僅是 ROO，Wolt 也一樣。不論你看的是毛利（gross profit）或貢獻（contribution），兩者都持續以年對年基礎改善。

Deepak Mathivanan, Cantor Fitzgerald.

Deepak Mathivanan，Cantor Fitzgerald。

Great. Thank you. Tony, last month, Andy Fang talked with Boris at Claude about how Dash is aiming to translate AI spend into outcomes somewhat closer to the business metrics now, while also letting employees to experiment aggressively with AI tools. Can you talk about where you're seeing this attribution clearly now? And how we should broadly think about AI spend at Dash over the next 12 to 18 months?

很好。謝謝。Tony，上個月 Andy Fang 在 Claude 與 Boris 的對談中提到，Dash 的目標是把 AI 支出更直接地轉化為更貼近業務指標的成果，同時也讓員工能夠更積極地用 AI 工具做實驗。你能談談你現在在哪些地方已經清楚看到這種歸因（attribution）嗎？以及我們應該如何整體看待 Dash 在未來 12 到 18 個月的 AI 支出？

And then maybe, one for you, Ravi. US restaurant GOV acceleration. Can you expand on the drivers of the growth? I know weather was disruptive last quarter, but you also had factors like World Cup. Curious if you can talk a little bit more about the factors of acceleration in Q2. Thank you so much.

然後也許再問你一題，Ravi。美國餐廳 GOV 的加速成長。你能進一步說明成長的驅動因素嗎？我知道上季天氣造成干擾，但你們也有像世界盃這類因素。想請你多談談 Q2 加速的因素。非常感謝。

Yeah, sure. On the AI question, Deepak, I would say a couple of things. The first thing is we want to make sure that our -- any technology that we meet, whether it's AI or frankly, anything else, that it's actually rooted in delivering a better customer experience. Because if it's not, I'm not exactly sure what problem we're actually trying to solve. And I don't think it makes sense to just play with the technology for the technology's sake.

好的，當然。關於 AI 的問題，Deepak，我會說幾點。第一，我們希望確保我們採用的任何技術——不論是 AI 或坦白說任何其他技術——都確實以提供更好的客戶體驗為根基。因為如果不是這樣，我不太確定我們到底想解決什麼問題。而且我不認為為了技術本身而玩技術是有意義的。

And so when you look at some of the things that we've seen success, for example, one of the more recent products we launched was called DoorDash Ask, which is an ordering agent that helps customers discover restaurants that are similar to ones that they've ordered in the past, but that are new to them, that helps them build a grocery cart in under two minutes.

因此，當你回頭看我們看到成功的一些事情時，例如，我們最近推出的產品之一叫做 DoorDash Ask，這是一個下單代理，能幫助顧客發掘與他們過去點過、但對他們而言是新的餐廳相似的餐廳，也能幫助他們在兩分鐘內建立一個雜貨購物車。

So really solving actual pain points that we see in using an increasingly larger and more diversified marketplace. That's one example. On the merchant side, we've seen automation in building catalogs for retailers or menus for restaurants, which includes all of the photos, the metadata around all of the different SKUs and items so that we can actually onboard a merchant faster in order to get same-store sales growth.

所以這真的是在一個規模越來越大、也越來越多元化的市集中，解決我們看到的實際痛點。這是一個例子。在商戶端，我們看到在為零售商建立商品目錄或為餐廳建立菜單方面的自動化，這包含所有照片、所有不同 SKU 與品項的中繼資料，讓我們能更快地讓商戶上線，以帶動同店銷售成長。

For Dashers, we're seeing improvements in routing as well as how we can help Dashers find the best areas to Dash. Those are some examples of how we actually have applied AI in a way that is actually meaningful in terms of driving customer outcomes, which ultimately deliver business results.

對於 Dashers（外送夥伴），我們也看到路線規劃的改善，以及我們如何幫助 Dashers 找到最適合跑單的區域。這些都是我們如何以真正有意義的方式應用 AI 的例子，從而推動顧客端成果，而這些成果最終會帶來商業結果。

The other thing that we've done is kind of with all things at DoorDash, we care as much about how we do it in order to be efficient as we do about allowing some degree of inefficiency towards invention. And we built a lot of tools, the tools like DashBench and other systems internally that allow us to model the appropriate tools and give those tools to be used for the right level of token spend or intelligence required.

我們做的另一件事是，有點像 DoorDash 一貫的做法：我們同樣重視如何以高效率的方式把事情做好，也同樣重視在發明創新上允許一定程度的低效率。我們打造了許多工具，例如 DashBench 以及其他內部系統，讓我們能夠建模合適的工具，並把這些工具用在適當的 token 花費或所需智慧水準上。

And so I think those are some of the things in which we found ourselves -- that's kind of our approach in terms of how we've applied it towards AI or frankly, any technology. It's why we're excited to keep going and lean in, in a way that I think is disciplined on the one hand, but on the other hand, allows us to bring real customer benefits.

所以我認為，這些就是我們一路走來所形成的做法——這大概就是我們在 AI（坦白說也包括任何技術）上的應用方式。這也是為什麼我們很期待持續推進並加大投入：一方面我認為是有紀律的，另一方面也讓我們能帶來真正的顧客效益。

Deepak, on your second question around restaurant growth, I mean, restaurant growth was quite strong in the quarter, if you look at it. In fact, growth accelerated from Q1 to Q2. A lot of the growth is coming from just increase in DashPass subscribers. A couple of points, right? We wrote in the letter as well. We added more number of DashPass subscribers in the last year compared to the two prior years.

Deepak，關於你第二個有關餐廳成長的問題，我的意思是，如果你看這一季，餐廳成長相當強勁。事實上，成長從 Q1 到 Q2 是加速的。很大一部分成長來自 DashPass 訂閱者的增加。有幾點，對吧？我們也在信中寫到了。我們在過去一年新增的 DashPass 訂閱者數量，超過了之前兩年的總和。

Number two, when I look at the paid subscriber growth in DashPass, it was one of the highest that we've seen in the last couple of years. A lot of that is the underlying product continuing to get better. A lot of that is the increased investment that we made in selection as well as quality. In fact, if you look at mature cohorts, they're continuing to engage higher than what we've seen before.

第二，當我看 DashPass 的付費訂閱者成長時，這是我們過去幾年看到的最高水準之一。其中很大一部分是因為底層產品持續變得更好。其中也有很大一部分來自我們在選擇（品類/供給）以及品質上的加大投資。事實上，如果你看成熟 cohort（成熟用戶群），他們的參與度持續高於我們過去看到的水準。

New consumers continue to be quite strong as well. Ultimately, all of this is driving the growth that you're seeing in restaurants. And what I will also say is if you think about Q2 of last year, it was unusually strong for us.

新消費者也持續相當強勁。最終，這一切都在推動你所看到的餐廳端成長。另外我也想說的是，如果你回想去年 Q2，對我們而言那是一個異常強勁的季度。

So comping against what was a strong Q2 of last year and still putting up the strong numbers in Q2, that's a true testament to, a, the demand that we're seeing in the business as well as the underlying improvements in product.

所以在與去年強勁的 Q2 做比較的情況下，我們在今年 Q2 仍然交出強勁的數字，這真正證明了：a，我們在業務中看到的需求；以及底層產品的改善。

Dominic Ball, Rothschild & Co Redburn.

Dominic Ball，Rothschild & Co Redburn。

Hi, guys. Thank you for the question. Interesting commentary about kind of investing more in merchant services and software. I think it's somewhat well known that DoorDash has been testing its POS product in a few markets in the US, with both SMBs and enterprise restaurants.

嗨，各位。謝謝讓我提問。你們提到在商戶服務與軟體方面加大投資的評論很有意思。我想大家多少都知道，DoorDash 一直在美國少數市場測試其 POS 產品，對象包含中小企業與大型連鎖餐廳。

So would just love to know about how these test trials are going, what products and features are kind of resonating? What is kind of driving some restaurants maybe choosing to use DoorDash here? And then how do we think about a potential more broader commercial launch going forward? Thanks, guys.

所以很想了解這些測試進展如何，哪些產品與功能最能引起共鳴？是什麼在驅動一些餐廳可能選擇使用 DoorDash？另外，我們該如何看待未來更廣泛的商業化推出？謝謝各位。

Our vision is to be the best partner to every local business. That probably is pretty clear from the mission since day one. And the way we do this is that we want to give every business the same tools that we built for ourselves so that they can grow their digital business. So if you think about what that looks like, I mean, we kind of play in this ecosystem where we have at least three offerings today, right? We have our marketplace. We have tools to help build the digital businesses of restaurants and retailers. In fact, that business serves over 150,000 businesses and has grown 40% year-over-year in the quarter.

我們的願景是成為每一家在地商家的最佳合作夥伴。從第一天起，這一點從使命上應該就很清楚。而我們做到這一點的方式，是希望把我們為自己打造的同樣工具提供給每一家商戶，讓他們能成長其數位業務。所以如果你想像那會是什麼樣子，我的意思是，我們目前在這個生態系中大概有至少三項產品/服務，對吧？我們有我們的市集（marketplace）。我們也有工具來協助餐廳與零售商建立其數位業務。事實上，這項業務服務超過 150,000 家商戶，且本季年增 40%。

And then more recently, we've introduced products that actually drive customers inside the store. Some of these products include going out as well as reservations. And then on the business side, it includes SevenRooms, which is a company that we acquired about a year ago.

然後更近期，我們也推出能實際把顧客帶進店內的產品。其中一些產品包括外出用餐以及訂位。而在商戶端，則包含 SevenRooms——這是我們大約一年前收購的一家公司。

And when I think about what this ecosystem allows us to do, I mean, it allows us to help customers build their relationships with the local businesses and ideally forge regulars for each one of these local businesses.

當我思考這個生態系讓我們能做什麼時，我的意思是，它讓我們能幫助顧客與在地商家建立關係，並理想上為每一家在地商家培養常客。

And the reason why we can do this is because we have the biggest scale as well as the deepest customer datasets to allow us to actually drive this engagement. For example, a customer may start by ordering delivery from the DoorDash app and then perhaps, they sign up for a loyalty program on a merchant's first-party channel, something that we've built for them.

而我們之所以能做到，是因為我們擁有最大的規模，以及最深厚的顧客資料集，讓我們能實際推動這種互動。例如，一位顧客可能先從 DoorDash App 開始下外送，接著也許他們會在商戶的第一方渠道（first-party channel）上加入忠誠度計畫——那是我們為商戶打造的。

And when it makes sense, both DoorDash and the merchant can incentivize the customer to go inside the store or go inside the restaurant and actually have a meal. And so when you think about the interaction effects here, what we really are able to do is be the best growth partner for all these businesses, and for consumers, give them the most choice in terms of how they actually want to interact with all these businesses. I think the proof points we kind of highlighted in our quarterly update are really, really strong, and we think that this ecosystem is one that has a very long runway.

而在合適的時機，DoorDash 與商戶雙方都可以激勵顧客走進店內或走進餐廳，實際用餐。因此，當你思考這裡的交互效應時，我們真正能做到的是：成為所有這些商家的最佳成長夥伴；同時對消費者而言，提供他們最多的選擇，讓他們能以自己想要的方式與這些商家互動。我認為我們在季度更新中強調的那些佐證非常、非常強，我們也認為這個生態系仍有非常長的成長跑道。

Jason Helfstein, Oppenheimer.

Jason Helfstein，Oppenheimer。

Thanks. So just two questions. One, can you -- you talked a little bit about Dot deliveries in the release. I guess anything you want to share on how you're thinking about unit economics kind of like today versus where we think it goes long term and how you think that impacts demand around elasticity? And then is it possible to share the AOV for the most recent period in the chart on page 3, where you're comparing the restaurant versus the grocery and retail? Thanks.

謝謝。我有兩個問題。第一，你們在新聞稿中稍微談到 Dot 配送。我想知道你們是否願意分享一下，你們如何看待單位經濟（unit economics）在目前相較於長期可能走向何處，以及你們認為這會如何影響需求的價格彈性？第二，是否可以分享一下第 3 頁圖表中最近一期的 AOV（平均訂單金額），你們在那裡比較了餐廳與雜貨及零售？謝謝。

Sure. Maybe I can start with the question on Dot and then Ravi, feel free to chime in on the second question. I would say a few things about DoorDash Dot. The overall vision for autonomous delivery, at least at DoorDash, is that we want to offer AVs throughout the network so that we can deliver the best service to customers. The best service could be the fastest delivery, the most affordable delivery, delivery from the widest selection, including from very faraway places.

好的。或許我可以先從 Dot 的問題開始，然後 Ravi，你也可以就第二個問題隨時補充。我想就 DoorDash Dot 談幾點。至少在 DoorDash，我們對自動化配送的整體願景是，希望在整個網路中提供自動駕駛車輛（AV），以便為顧客提供最佳服務。所謂最佳服務，可能是最快的配送、最實惠的配送、以及來自最廣泛選擇的配送，包括從非常遙遠的地方送達。

And I would say that the real secret sauce or the magic that we've learned in building DoorDash Dot is that it's really the complexity of marrying the operations with the technology that actually allows you to even have a chance at delivering scaled autonomous delivery.

我會說，我們在打造 DoorDash Dot 時學到的真正關鍵祕訣或「魔法」，其實在於把營運與技術結合的複雜度；正是這點，才讓你有機會實現可規模化的自動化配送。

To put in a different way, DoorDash Dot, or any AV, for that matter alone will not make autonomous deliveries actually scale, certainly not at any level of meaningful penetration for customers. I mean, at the end of the day, you have to solve very challenging operational problems in the physical world. You have to solve the loading problem at the merchant, you have to estimate merchant prep times in the case of restaurants or inventory levels in the case of retailers and grocery stores. You got to solve for difficult drop-off issues like perhaps a doorman inside of a high-rise building or a complicated gate entry in an apartment unit.

換個方式說，DoorDash Dot，或任何 AV，就其本身而言，並不會讓自動化配送真正實現規模化，當然也不可能在顧客端達到任何有意義的滲透率。我的意思是，歸根結底，你必須解決現實世界中非常具挑戰性的營運問題。你必須解決商家端的裝載問題；在餐廳情境下，你必須估算商家備餐時間；在零售商與雜貨店情境下，你必須估算庫存水位。你還得解決困難的交付問題，例如高樓大廈內的門房，或公寓單位複雜的門禁入口。

These are all of the issues that we've encountered and are candidly, like maybe [1/1,000] of the issues that we've encountered in building DoorDash Dot in our test markets. And so I think to get to the milestone that we have today with Dot where we have meaningful scale in our test market has been a huge accomplishment by the team.

這些都是我們在測試市場打造 DoorDash Dot 時遇到的問題，而且坦白說，可能只是我們遇到問題的[1/1,000]。因此，我認為我們今天在 Dot 上達成的里程碑——在測試市場已具備有意義的規模——對團隊而言是一項巨大的成就。

And it gives us confidence that you can actually truly scale autonomous delivery if you both can master the operations and the technology. And that's why I think we're in the best position to do it because we run the network. And we're also building the technology ourselves, which gives us the lowest level of detailed understanding of how to actually make this happen.

這也讓我們更有信心：如果你同時能掌握營運與技術，就能真正把自動化配送做大規模。這也是為什麼我認為我們處於最佳位置來做到這件事，因為我們營運這個網路。而且我們也自行打造技術，這讓我們對如何真正落地有最細緻、最深入的理解。

And it doesn't mean that we have to build everything. I mean we actually -- whether it's by land or by air, we have a variety of partners that we interact with. And for everyone, we kind of solve all of the challenging operational problems for them, and we do that through our autonomous delivery platform. So for a merchant, you can take your existing DoorDash integration and you get access to any and all AVs. And for customers, you're going to, actually, one day, get the benefit of these technologies.

但這不代表我們必須什麼都自己做。我的意思是，無論是陸運或空運，我們都與各式各樣的合作夥伴互動。而對每一位合作夥伴，我們都會替他們解決所有具挑戰性的營運問題，並透過我們的自動化配送平台來完成。因此，對商家而言，你可以沿用既有的 DoorDash 串接，就能接入任何及所有 AV。而對顧客而言，未來某一天，你將能真正享受到這些技術帶來的好處。

And with respect to the cost profile, we're very excited by what we see. I mean it's exponential progress. But I think that the first milestone for us has really been, can you actually commercialize this in a way that you've seen certain robotaxi providers like Waymo actually do it with rideshare, we've been able to now do it with delivery.

至於成本結構，我們對目前看到的結果非常振奮。我的意思是，進展是指數級的。但我認為我們的第一個里程碑其實是：你能否以商業化的方式落地——就像你看到某些 robotaxi 供應商（例如 Waymo）在共乘上做到的那樣——而我們現在也已經能在配送上做到。

Hey, Jason, it's Ravi. On the second point, right, let me start with what we're seeing in the business. Look, as we continue to operate and expand our new verticals in grocery business, what you're seeing is as the product is getting better, the basket sizes are increasing. This was what we had originally expected, which is as we add more selection, as the quality of the product continues to get better, you would have customers use us for more use cases, which will ultimately drive the basket sizes to be higher.

嗨，Jason，我是 Ravi。關於第二點，對，我先從我們在業務上看到的情況談起。你看，隨著我們持續營運並擴張雜貨等新垂直業務，你會看到：產品變得更好時，客單價（basket sizes）就會提高。這正是我們原先的預期：當我們增加更多選擇、產品品質持續提升時，顧客會在更多使用情境中使用我們，最終推動客單價上升。

We're seeing that in older cohorts. We're seeing that in newer cohorts, which is visible overall in the business as well. But let me actually take a step back and walk you through what our thesis was and why we put the second chart in the letter. Look, our core thesis was twofold. One is we knew that as customers and consumers habituate on the platform, they will spend more with us. They'll spend more with us on the restaurants business.

我們在較早期的用戶群（cohorts）中看到這點。我們在較新的 cohorts 中也看到這點，整體業務表現也反映得很清楚。不過我想先退一步，帶大家回顧我們的論點，以及為什麼我們在股東信中放了第二張圖表。你看，我們的核心論點有兩個層面。第一，我們知道當顧客與消費者在平台上形成使用習慣後，他們會在我們這裡花更多錢。他們會在餐廳業務上花更多錢。

They'll adopt newer categories. They'll spend more with us on the new verticals business. The example that we took in that chart was one of our older cohorts, which is largely representative actually of the other cohorts that we see in the business, where consumer spend on restaurants is increasing, consumer spend on new verticals is increasing. And at the same time, DashPass penetration is increasing. And if you think about it, right, this is almost like a self-reinforcing loop where the product gets better, consumers adopt and habituate to DashPass.

他們也會採用更新的品類。他們會在新垂直業務上花更多錢。我們在那張圖表中採用的例子，是我們較早期的一個 cohort；它其實在很大程度上也代表我們在業務中看到的其他 cohorts：消費者在餐廳的支出增加、在新垂直品類的支出也增加。同時，DashPass 的滲透率也在提升。而如果你仔細想，這幾乎像是一個自我強化的循環：產品變得更好，消費者採用並逐漸習慣使用 DashPass。

As they adopt DashPass, they continue to use the product more, which ultimately leads to more growth as well as more profit dollars in the system. This has largely been the focus for us, and that's largely what you're seeing in the underlying cohort. And we're very pleased with the performance of that in the underlying business.

當他們採用 DashPass 後，就會更頻繁地使用產品，最終帶來更多成長，也讓系統中的利潤美元（profit dollars）增加。這一直是我們的重點，而這也正是你在底層 cohort 數據中看到的情況。我們對其在底層業務中的表現非常滿意。

Shweta Khajuria, Wolfe Research.

Shweta Khajuria，Wolfe Research。

Thank you for taking my questions. Let me try two, please. First is on the DashMart Fulfillment Services since your launch and since working with a handful of partners, I understand, what have you learned so far that you could potentially quantify or some sort of tangible learnings that you could share, whether it is on the magnitude of customer experience improvement or something else? And then what metrics do you look at to be able to make that decision to scale DashMart Fulfillment Services?

謝謝讓我提問。我想問兩個問題。第一個是關於 DashMart 履約服務（Fulfillment Services）：自從你們推出並且我理解已與少數合作夥伴合作以來，你們目前學到了哪些可以量化、或某種具體的心得可以分享？例如顧客體驗改善的幅度或其他面向？另外，你們會看哪些指標來做出是否擴大 DashMart 履約服務的決策？

And then the second question is just overall EBITDA growth, where now we are in the back half of this year and in the light of investments this year, but more importantly, as we think about balancing growth and top line growth and EBITDA growth, Ravi, how are you thinking about that as we think about the demand trends that you see right now and balancing growth with profitability? Thanks a lot.

第二個問題是關於整體 EBITDA 成長：現在我們已進入今年下半年，並考量今年的投資；但更重要的是，當我們思考如何在成長、營收成長與 EBITDA 成長之間取得平衡時，Ravi，基於你目前看到的需求趨勢，你是如何思考在成長與獲利能力之間做平衡的？非常感謝。

Yeah. I can start by answering about DashMart Fulfillment Services. To state the obvious, we want every local business to be successful in their local communities. And that's true in every category from restaurants to grocery to retail.

好的。我可以先回答關於 DashMart 履約服務的問題。顯而易見的是，我們希望每一家在地商家都能在其在地社區中成功。這在所有品類都成立，從餐廳到雜貨到零售皆然。

And one of the things that we learned five years ago when we launched our grocery business, even though it's going so well, and we're helping lots of grocers compete, is that there's this structural challenge with grocery delivery where grocers don't know their inventory. In some ways, it's almost impossible for a whole host of reasons, one of which is because consumers come in and they move things around. And so it's a very difficult proposition then to offer customers a use case where you're asking them to pay a premium for delivery, but they don't get exactly what they ordered.

我們在五年前推出雜貨業務時學到的一件事是：即使進展非常順利、我們也在協助許多雜貨商提升競爭力，但雜貨配送存在一個結構性挑戰——雜貨商並不了解自己的庫存。在某些方面，基於各種原因，這幾乎是不可能的，其中一個原因是消費者會進店把商品移來移去。因此，要向顧客提供一個使用情境：要求他們為配送支付溢價，但卻無法拿到完全符合其下單內容的商品——這就變成一個非常困難的命題。

So our solution to this is DashMart Fulfillment Services, where we are managing warehouses where we control the inventory and sell exactly what's in stock so that we can actually offer near perfect accuracy and give customers the selection that they want from any place inside the city and also very, very quickly.

因此，我們對此的解決方案是 DashMart 履約服務（DashMart Fulfillment Services）。我們在管理倉庫，由我們掌控庫存，並且只銷售確實有現貨的商品，這樣我們就能提供近乎完美的準確度，讓顧客在城市內任何地方都能選到他們想要的品項，而且速度也非常、非常快。

And so in terms of what we've seen so far since our announcement last fall and the launch that we've had with several partners, is we're seeing lots of incremental demand because these warehouses are running near 24/7.

因此，就我們自去年秋天宣布以來、以及與多家合作夥伴推出後目前所看到的情況而言，我們看到大量的增量需求，因為這些倉庫幾乎是 24/7 在運作。

If you compare that to traditional store operating hours, that's a dramatic increase in TAM as well as really just solving the needs of customers because customers sometimes don't get freed up until maybe after store closing hours before they can think about the next day or planning their groceries or their shopping needs. So we're seeing lots of incremental demand.

如果你把這和傳統門市的營業時間相比，這不僅讓 TAM 大幅提升，也真正解決了顧客的需求，因為顧客有時候要到店家打烊後才有空，才會開始思考隔天的安排，或規劃他們的採買、雜貨或購物需求。因此我們看到大量的增量需求。

And we're also seeing 10 times better error rates because we're running the inventory. And as a result, we're selling customers exactly -- the customers are getting exactly what they ordered, and we're selling them exactly what's actually in stock. And so all the signs right now are really positive for us to scale DashMart Fulfillment Services. But look, this is complicated. I mean this is building physical infrastructure.

而且我們也看到錯誤率改善了 10 倍，因為庫存是由我們在管理。因此，我們賣給顧客的正是——顧客拿到的正是他們下單的商品，而我們賣給他們的也正是實際有現貨的商品。所以目前所有跡象都非常正面，支持我們擴大 DashMart 履約服務的規模。但坦白說，這很複雜。畢竟這是在建置實體基礎設施。

You're obviously adding technology to it by managing the inventory and obviously running the entire fulfillment. So -- but you can imagine a world in which these warehouses can power all of the needs inside of the city. And you can -- what you can really do when you do something like that is you can really unlock the amount of selection available to customers because today, I would say DoorDash is delivering probably 1/10 in most cities, the available selection in terms of retail and grocery and the like.

你顯然也在其中加入科技，透過管理庫存並且實際運作整個履約流程。所以——但你可以想像一個世界，這些倉庫能夠支撐城市內所有需求。而當你做到這件事時，你真正能做的是大幅解鎖可提供給顧客的選擇量，因為以今天來看，我會說 DoorDash 在多數城市所能配送的零售、雜貨等可得選擇，大概只有全部可得選擇的 1/10。

And so we can actually build this capability and do it in concert with all of the great retail and grocery selection inside of the city, I think customers get the best product. They get all the selection that they want at perfect quality very quickly.

因此，我們其實可以建立這項能力，並與城市內所有優質的零售與雜貨供給一起協同運作；我認為顧客會得到最好的產品。他們能以完美的品質、非常快速地拿到他們想要的所有選擇。

And so to your second question around -- look, I mean, trying to balance both growth as well as profitability, right? You asked about the demand trends. Look, when I look at the underlying business, the demand trends continue to be quite strong. I mean, as you know, we focus and spend a lot of our time on cohorts. MAUs, when I look across the board, they have hit all-time highs.

接著回到你的第二個問題——你看，我的意思是，如何在成長與獲利之間取得平衡，對吧？你問到需求趨勢。你看，當我觀察核心業務時，需求趨勢仍然相當強勁。如你所知，我們把很多時間花在 cohort（分群）分析上。從整體來看，MAU 已經創下歷史新高。

Subscription, I talked about the fact that both domestically and internationally, those are one of our best quarters in the sense that subscription continues to be at record highs. Restaurant growth accelerated. [Deliveroo] continues to do quite well.

訂閱方面，我提到過，無論在美國本土或國際市場，這都是我們表現最好的季度之一，因為訂閱持續創下新高。餐廳成長加速。[Deliveroo] 也持續表現相當不錯。

Overall, when I look at the underlying improvements in the product, that's driving the improvements that you're seeing both from retention as well as order frequency. At the same point, I mean, look, I mean, the quarter was very strong from an underlying profitability perspective as well.

整體而言，當我看產品底層的改善時，這正在推動你所看到的留存與下單頻率的提升。同時，你看，從底層獲利能力的角度來看，本季度也非常強勁。

A lot of that is because the underlying unit economics continue to improve. For us, the philosophy has always been the same, right? We're consistently trying to improve efficiency. There's many sources across the board, whether it's sales and marketing, fixed costs up and down the P&L. And our goal is to consistently reinvest back in the business.

其中很大一部分原因是單位經濟（unit economics）持續改善。對我們而言，理念一直都一樣，對吧？我們持續嘗試提升效率。這有很多來源，涵蓋各個面向，無論是銷售與行銷，或是損益表（P&L）上下的固定成本。而我們的目標是持續把資源再投資回業務。

Look, we talked about the fact that we're investing back into the business and building product, some of which we've talked about earlier on the call, whether it's autonomy, unification of the global tech stack or investing back in merchant services, all of those are going well. They're all going on budget and on plan according to what I had estimated a couple of quarters ago.

你看，我們談到過我們正在把資源再投資回業務並打造產品，其中一些我們在電話會議前面也提過，無論是自動化（autonomy）、全球技術堆疊（global tech stack）的整合，或是再投資於商戶服務（merchant services），這些進展都很順利。它們都在預算內、也符合計畫，與我在幾個季度前所估計的一致。

These will increase the surface area. Ultimately, the goal for us is to continue to drive both growth as well as profit dollars. As long as we make the product better, what we're seeing in the business is we're able to do both, right? This is how we think about balancing both growth as well as profit dollar production in our business.

這些將擴大我們的覆蓋面（surface area）。最終，我們的目標是持續推動成長以及獲利金額（profit dollars）。只要我們把產品做得更好，我們在業務上看到的是兩者都能做到，對吧？這就是我們如何思考在業務中平衡成長與獲利金額產出。

Josh Beck, Raymond James.

Josh Beck，Raymond James。

Thanks for taking the questions. I had -- maybe a product-oriented question. So on Dot getting to high single digit within a market by the end of the year, could you give us maybe like some characteristics? Is this because it's maybe more of a longer route suburb type of market?

謝謝讓我提問。我有——可能是一個偏產品面的問題。關於 Dot 在年底前在單一市場達到高個位數占比，你能否提供一些特徵？這是因為它可能更像是路線較長、偏郊區型的市場嗎？

And if you were to maybe add in all of the different modalities between maybe sidewalk, pathway, drones, autonomous vehicles, is there some type of ceiling that you have in your mind in terms of maybe the percentage of orders that could be filled through autonomous? And then just secondarily on the AI assistant, obviously, lots of attractive characteristics, discovery, basket, et cetera.

另外，如果把各種不同的運送模式都納入考量，例如人行道、步道、無人機、自動駕駛車輛，你心中是否有某種上限——例如有多少比例的訂單可以透過自動化來完成？然後第二個問題是 AI 助理，顯然有很多吸引人的特性，像是探索（discovery）、購物籃（basket）等等。

Curious if there's been other observations with respect to better frequency of these customers? I'm curious on maybe what you have to say about ad monetization. It seems like maybe time spent could be less, but conversion could be better. So just curious on those two topics. Thank you.

我好奇是否還有其他觀察，例如這些顧客的下單頻率是否更好？我也想問你對廣告變現（ad monetization）有什麼看法。看起來使用時間可能更少，但轉換率可能更好。所以就這兩個主題想請教一下。謝謝。

Alright. I think you had like maybe seven or eight questions in there, Josh, but I'll do my best. Look, on DoorDash Dot, it is a representative DoorDash market. We're testing in Phoenix. And so -- and we're testing it with real scale now, which is -- like it's a real accomplishment when you think about -- these are not demos, these are not prototypes. There are no fixed routes. This is real life, and this is real life for tens of thousands of customers that are receiving real deliveries. And so that's very exciting.

好的。Josh，我想你剛剛大概問了七、八個問題，但我會盡力回答。你看，關於 DoorDash Dot，它是一個具代表性的 DoorDash 市場。我們在鳳凰城（Phoenix）進行測試。因此——而且我們現在是在真正的規模下測試，這——當你想想看，這不是展示（demo），也不是原型（prototype）。沒有固定路線。這是真實世界，而且是為數以萬計的顧客提供真實配送的真實世界。所以這非常令人振奮。

It's also very challenging. I mentioned maybe five or six issues that might be literally 1/1000 the number of issues that you have to solve to actually have a chance of making autonomous deliveries actually happen. And that's true whether you're doing it by land, and it's also true whether you're doing it by air. I mean we've seen this with DoorDash Air as well in addition to all the partners that we test with and bring our scale to.

但這也非常具有挑戰性。我提到的可能五、六個問題，實際上可能只是你必須解決的問題數量的 1/1000，才能真正有機會讓自動化配送成真。無論你是在地面上做，或是在空中做，都是如此。我們在 DoorDash Air 也看到了這點，此外還包括我們測試並導入我們規模的所有合作夥伴。

And so what I would say is the -- what's going to determine the ceiling or the penetration of autonomous delivery is whether or not you can master both the operations and the technology. And I believe we're best positioned to do that because we're actually doing both in-house. And so we are getting to the lowest level of detail, chopping down the very heavy wood of every issue that exists in the real world, that exists in every single restaurant, retailer on the road.

因此，我會說——決定自動化配送上限或滲透率的，將是你是否能同時精通營運與技術。而我相信我們最有條件做到，因為我們兩者都是內部自建（in-house）在做。因此我們能深入到最底層的細節，把現實世界中存在的每一個問題——存在於每一家餐廳、每一家零售商、每一條道路上的問題——逐一攻克。

I mean, Dot travels road, sidewalk and bike lane. It's the only vehicle in the world to do that autonomously. And there's a lot of challenges when you actually try to take on that multimodality. And one of the interesting things, perhaps the most interesting thing of what we're building with autonomy is this autonomous delivery platform.

我的意思是，Dot 會在道路、人行道和自行車道上行駛。它是世界上唯一能夠自主做到這一點的車輛。而當你真的要處理這種多模式（multimodality）時，會有很多挑戰。而我們在自動化領域所打造的其中一個有趣之處——也許是最有趣的——就是這個自主配送平台。

You can think of this as the brains that actually makes it all happen, that deciphers which vehicles go to which orders, that decides whether you have a mixed route, where you have human dashers as part of the legs of the journey and autonomous vehicles on other parts of the journey, that looks into the configuration of the package size and the package design and the package weight.

你可以把它想成真正讓一切發生的大腦：它會解讀哪些車輛要去處理哪些訂單；它會決定你是否採用混合路線——也就是旅程的某些路段由人類外送員（dasher）完成，其他路段由自動駕駛車輛完成；它也會考量包裹尺寸、包裝設計與包裹重量的配置。

And there's a lot of complexity. But all of that gets reduced and kind of goes and disappears because we kind of handle it through our autonomous delivery platform so that merchants get the same integration that they have with DoorDash today.

而這裡面有很多複雜性。但這些複雜性都被簡化了，某種程度上也被隱藏起來，因為我們透過自主配送平台來處理，讓商家仍然能夠使用他們今天與 DoorDash 相同的整合方式。

They don't have to change a single thing about their workflow, and customers just get access to the benefits of autonomous delivery, which will be, in the future, speed, cost, and more selection. And so I think that's going to be really, really, really exciting. And again, like, it's really going to be the execution that determines the ceiling. I think we're best suited for that execution.

他們不需要對工作流程做任何改變，而顧客則能享受到自主配送的好處——未來會是速度、成本，以及更多選擇。所以我認為那會非常、非常、非常令人興奮。而且再說一次，真正決定上限的會是執行力。我認為我們最適合去把這件事執行好。

I think your second question is on the AI assistant. I mean the short answer is, of course, if you're -- whenever you're making -- whenever you're reducing friction in a product, you get more usage. So if we're making it easier to build a grocery cart, you get more grocery carts, and you get bigger grocery carts. And it doesn't change at all the ads profile or anything else.

我想你的第二個問題是關於 AI 助理。我的意思是，簡短的答案是：當然，只要你在產品中降低摩擦——每當你在降低摩擦——使用量就會增加。所以如果我們讓建立雜貨購物車更容易，你就會有更多雜貨購物車，而且購物車也會更大。而這完全不會改變廣告（ads）的輪廓或其他任何東西。

If anything, you actually get just more incremental orders and more incremental opportunities because people order more often. The same thing is true about ordering restaurants. If you now discover that there's something slightly healthier or faster or cheaper or just better or different from what you typically order, then you tend to order more often and possibly for more use cases.

如果說有什麼影響，那其實是你會得到更多增量訂單與更多增量機會，因為人們下單更頻繁。餐廳點餐也是同樣道理。如果你現在發現有些選項比你平常點的稍微更健康、更快、更便宜，或只是更好、不同，那你往往會更常下單，甚至可能涵蓋更多使用情境。

We eat 20 times to 25 times a week. And DoorDash only touches a fraction of that. And that is really the runway we have. When I think about the number of meal occasions and shopping needs on top of that, it's north of 100 per month. And so we have a lot of shots on goal that we can go capture. And if we just reduce the friction of actually ordering and then master and continue to master the operations of fulfillment, all of the metrics will go in the right direction.

我們每週會吃 20 次到 25 次。而 DoorDash 只觸及其中的一小部分。這就是我們真正的成長跑道。當我想到餐食場景的次數，再加上購物需求，每月超過 100 次。所以我們有很多次「射門機會」可以去把握。而如果我們只是降低實際下單的摩擦，並且掌握、持續掌握履約（fulfillment）的營運，所有指標都會朝正確方向發展。

Youssef Squali, Truist Securities.

Youssef Squali，Truist Securities。

Awesome. Thank you so much. So guys, I just want to double-click on the margin question again. Can you -- Ravi, can you maybe just talk about the drivers for the material beat in adjusted EBITDA in Q2? It came quite a bit outside of the guidance range. More recently, you guys have been guiding to, at some point, hit somewhere in the midpoint. This one is dramatically higher. Were there any investments that got pushed back into Q3 and then Q4? And then are we still tracking to show higher year-on-year adjusted EBITDA margin for 2026 ex Deliveroo? Thank you.

太棒了。非常感謝。各位，我想再深入追問一次關於利潤率（margin）問題。Ravi，你能不能談談第二季調整後 EBITDA 大幅超預期（material beat）的驅動因素？結果明顯超出指引區間。最近你們的指引是，在某個時間點會落在區間中點附近。但這次顯著更高。是否有任何投資被延後到第三季、第四季？另外，若排除 Deliveroo，我們是否仍在軌道上，於 2026 年呈現更高的年對年調整後 EBITDA 利潤率？謝謝。

Hey, Youssef, let me take that, right? Like look, at the highest level, what I would say is the core restaurant business continues to perform on all cylinders. Growth, like I said on Deepak's question, right, like growth accelerated, unit economics continue to improve. New verticals growing much faster than our restaurants business as well as it's on track to being gross profit positive by the end of the year. International continues to do well.

嗨，Youssef，我來回答這題，好嗎？整體來看，我會說核心餐廳業務持續全速運轉、各方面表現都很強。成長方面，就像我在回答 Deepak 的問題時提到的，成長加速；單位經濟（unit economics）也持續改善。新垂直業務的成長速度也遠快於我們的餐廳業務，而且也正朝著在年底前達到毛利（gross profit）轉正的目標前進。國際業務也持續表現良好。

We've talked about some of the strength that you're seeing in Deliveroo as well as the beat on unit economics there. More specifically, what we saw in the quarter was the unit economic improvement came in ahead of our expectations, specifically in a couple of areas. Ads was one of them, subtotal was the other one. But in the second half of the quarter, those unit economics came in slightly ahead of what we had expected.

我們也談過你在 Deliveroo 看到的一些強勁表現，以及那邊單位經濟的超預期。更具體來說，本季我們看到單位經濟的改善超出我們的預期，特別是在幾個面向。廣告（Ads）是一個，客單價小計（subtotal）是另一個。但在本季後半段，這些單位經濟的表現略優於我們原先的預期。

In addition, ROO beat our own internal volume expectations, ROO's contribution profit positive. That led to some of the upside that you're seeing in the business. But look, I mean, more broadly, what I would say is our philosophy in how we operate the business, that is not changing.

此外，ROO 的成交量也超出我們內部的預期，ROO 的貢獻利潤（contribution profit）為正。這帶來了你在業務中看到的一部分上行空間。但更廣泛地說，我想強調的是，我們經營業務的理念並沒有改變。

Look, it's a very tightly managed business. There's a lot of levers that we control up and down the P&L. What we're consistently trying to do is try to find sources of efficiency. Like I said earlier, there's many sources of efficiency that we work on and take those and reinvest that back in the business. We're not trying to optimize the last dollar from one quarter to the next.

你看，這是一個管理非常嚴謹的業務。我們在損益表（P&L）上下有很多可以調控的槓桿。我們一直在做的，是持續尋找效率來源。就像我先前說的，我們會在很多地方提升效率，然後把這些效率再投資回業務。我們並不是要在每一季之間去極致優化最後一塊錢。

Look, our focus has always been on how do you build a large, durable business over time while continuing to increase the overall profit dollar production. To your second point, look, I mean, our focus has always been on landing inside the range. And if you're thinking about the second half from a margin perspective, I would expect us to land inside the range of the guidance that we've given in Q3.

我們的重點一直是：如何隨時間建立一個規模大、且具韌性的業務，同時持續提高整體利潤金額（profit dollar）的產出。至於你第二點，坦白說，我們的重點一直是落在指引區間之內。如果你從利潤率角度看下半年，我預期我們在第三季會落在我們所給出的指引區間內。

There are going to be times like in Q2, where the EBITDA beat comes in later in the quarter. In those times, we just don't have enough time to reinvest back in the business, especially at the levels of efficiency that we desire. Q2 was one of those quarters where we're happy to drop it to the bottom line.

有時候會像第二季一樣，EBITDA 的超預期是在季末才出現。在那種情況下，我們就沒有足夠時間把它再投資回業務，尤其是以我們希望達到的效率水準來執行。第二季就是其中一季，我們很樂意讓它直接落到最終獲利（bottom line）。

Ross Sandler, Barclays.

Ross Sandler，Barclays。

Thanks, guys. Just a quick follow-up on the AV and then I had a question about the charts in the letter. So Tony, I think you have a few hundred robots in Phoenix. How quickly is the plan to kind of bring this to other cities? That's the first question.

謝謝，各位。先就自動駕駛車（AV）快速追問一下，然後我還有一個關於信中圖表的問題。Tony，我想你們在鳳凰城有幾百台機器人。把這個計畫推廣到其他城市的速度規劃是怎樣？這是第一個問題。

And then on those charts, we love the DoorDash charts. But I'm guessing if Uber were to put together their sub penetration relative to gross profit, it would probably look kind of the same for their leading markets. So are there examples of like cities or countries that you are kind of demonstrating the same trend as the US, but you've either come from behind or you've kind of come like head-to-head and overtaken one of your competitors on DashPass or Wolt penetration for subscribers? Thanks a lot.

然後關於那些圖表，我們很喜歡 DoorDash 的圖表。但我猜如果 Uber 也把他們的訂閱滲透率相對於毛利（gross profit）整理成圖，在他們的領先市場看起來可能也差不多。所以，有沒有一些城市或國家的例子：你們正在展現與美國相同的趨勢，但你們要嘛是後來居上，要嘛是正面交鋒並在 DashPass 或 Wolt 的訂閱者滲透率上超越某個競爭對手？非常感謝。

Sure. I can start on the AV question. Yeah, I mean, we're very excited about what's happening in Phoenix. But like as I mentioned, I mean, there's a lot we got to go figure out. And I'm a big believer that you really have to nail something before you scale it, especially in the area of autonomy where you're really solving like the problems of like six separate individual companies, almost, and you kind of have to be great at all of them.

當然。我可以先回答 AV 的問題。是的，我的意思是，我們對鳳凰城正在發生的事情感到非常興奮。但就像我提到的，我的意思是，還有很多事情我們得去釐清。而我一直很相信，在你把事情擴大規模之前，你真的必須先把它做到位，尤其是在自動駕駛這個領域，你其實像是在解決幾乎六家彼此獨立公司的問題一樣，而且你必須在所有這些方面都做得很出色。

And it's this tightly orchestration, again, between the operations and the technology. I can't stress that enough because that is the name of the game. I think just doing one or the other is not going to get it done. And so there's a lot of work to be done. We are in parallel, of course, securing permits because we work with cities to actually unlock a lot of this.

而這一切都取決於營運與技術之間緊密的協同編排。我再怎麼強調都不為過，因為這就是成敗關鍵。我認為只做其中一邊，是無法把事情做成的。所以還有很多工作要做。當然，我們也在同步推進許可的取得，因為我們會與各城市合作，才能真正解鎖其中許多事項。

And we do have plans to expand. We'll share certainly more as time comes, but we thought that would be helpful to offer just one milestone that we've accomplished and are excited about. But look, the road ahead for AV is very exciting, but it's going to take time. And mostly, it's going to take great execution between the operations and the technology, and that's what we're most excited about.

而且我們確實有擴張計畫。我們一定會在適當時機分享更多，但我們認為先提供一個我們已經達成、也感到興奮的里程碑會很有幫助。不過你看，AV 的前路非常令人期待，但這需要時間。而最重要的是，這需要營運與技術之間出色的執行力，這也是我們最興奮的地方。

Ross, on your second point, right, let me take a step back and talk about subscription more broadly. I mean if you think about subscription, it's been a key area of focus for us for the past couple of years. Subscription continues to do well. Whether it's subscribers in the US or international, the growth rate are some of the highest that we've seen in the last couple of years.

Ross，關於你的第二點，對，讓我先退一步，從更廣的角度談訂閱。我的意思是，如果你想想訂閱，它在過去幾年一直是我們的重點領域。訂閱業務持續表現良好。不論是美國或國際市場的訂閱用戶，成長率都是我們過去幾年所見最高的水準之一。

And the whole thesis for us was as DashPass penetration continues to increase, overall gross profit per MAU continues to increase. And I look at the penetration levels, we're still very, very early. We're seeing similar behavior in the international markets compared to what we see in the US, albeit some of the international countries are slightly behind because we launched subscriptions slightly later than what we did in the US

而我們的整體論點是，隨著 DashPass 滲透率持續提升，每位 MAU 的整體毛利也會持續提高。我看目前的滲透水準，我們仍然還非常、非常早期。我們在國際市場看到的行為與美國相似，只是部分國家稍微落後一些，因為我們在那些國家推出訂閱的時間比美國晚一些。

And for specific examples, right, I'm going to talk about some of the examples on Deliveroo. Look, if you're thinking about the UK market, we are gaining share. We're one of the fastest growing in that market. We're accelerating growth. We're accelerating paid subscriber growth, volume growth compared to what we've seen in the last couple of years. Even outside of Deliveroo, when you look at some of the countries that we operate in Wolt, in majority of the countries that we operate, we're continuing to gain share.

至於具體例子，我會談一些 Deliveroo 的例子。你看，如果你在想英國市場，我們正在提升市占。我們是該市場成長最快的業者之一。我們正在加速成長。相較於過去幾年，我們正在加速付費訂閱用戶成長與交易量成長。即使不看 Deliveroo，當你看我們在 Wolt 營運的部分國家，在我們營運的大多數國家，我們也持續提升市占。

And the key thing for us is we're not just looking at share gains. We're looking at what the order rate improvements are. We look at either the three months or the six months, we have continued to improve order rates. At the same point, across both Deliveroo as well as Wolt, it's been a way for us to drive efficient growth, where when you look at the unit economics either on a year-over-year basis or over the last couple of years, we've done a pretty good job of improving that.

而對我們來說關鍵是，我們不只是看市占提升。我們也在看下單頻率的改善幅度。不論看三個月或六個月的期間，我們都持續改善下單率。同時，在 Deliveroo 與 Wolt 兩邊，這也成為我們推動高效率成長的方式；當你看單位經濟，不論是年對年或過去幾年的趨勢，我們在改善方面做得相當不錯。

Net-net, I mean, look, we've talked about some of the countries before, right, whether it's UK, Israel or some of the other markets, we're continuing to do really well in terms of overall share gain as well as the underlying improvements in core metrics.

總結來說，你看，我們之前談過一些國家，對吧，不論是英國、以色列或其他市場，我們在整體市占提升以及核心指標的底層改善方面，都持續表現得非常好。

Brian Nowak, Morgan Stanley.

Brian Nowak，摩根士丹利。

Hey, guys. Thank you for taking my questions. I want to ask more about the global tech stack and sort of the unification of the global tech stack. So now that you've got the tech stack sort of built as of the spring, I know you've been testing a lot of modules and new capabilities.

嗨，各位。謝謝你們回答我的問題。我想更深入問一下全球技術堆疊，以及全球技術堆疊的整合。所以既然你們在今年春季左右已經把技術堆疊大致建好，我知道你們一直在測試很多模組和新能力。

Can you give us some examples where you're seeing early signal with actual quantifiable benefits of the new modules that give you confidence you're going to get real return and ROIC on these investments as we go into 2027?

你們能否舉一些例子，說明你們在哪些地方看到早期訊號，並且能以量化方式呈現新模組帶來的效益，讓你們有信心在邁向 2027 年的過程中，這些投資能帶來真正的回報與 ROIC？

Sure. Yeah. Hi, Brian, I mean, I would say we're still building the way you build these single tech stacks, it's not like -- I mean, the better analogy is it's not like a LEGO project where there's like a finishing step in the instruction manual. It's more like you're constantly -- well, first, you have to like replace an engine while you're flying a plane that's growing in speed and climbing in an altitude and then you're constantly making tweaks. That's probably the more appropriate analogy. But we're doing it.

當然。是的。嗨，Brian，我會說我們仍在建置中；你建這種單一技術堆疊的方式並不是——我的意思是，更貼切的比喻是，它不像一個 LEGO 專案，說明書裡有一個「完成」的步驟。它更像是你得在一架速度越來越快、爬升高度越來越高的飛機上更換引擎，然後你還得不斷做微調。那可能是更合適的比喻。但我們正在做。

And where are we seeing benefits? I mean, a lot of places already. I mean you're seeing conversion wins from improvements in search. You're seeing wins in automation in terms of our customer support flows and by bringing things from one place that was more automated to another place that was less automated.

那我們在哪裡看到效益？我的意思是，很多地方已經看到了。你會看到搜尋改善帶來的轉換率提升。你也會看到自動化方面的成果，例如在客服流程上，將某個地方較自動化的做法帶到另一個自動化程度較低的地方。

The theme really is what we're trying to do and why there's a thesis for a return here is because you're taking the best-of-breed feature and literally offering it to all of our 41 markets. And so this is not like taking one stack and then just like copy and pasting everything into all of the brands. It's actually literally taking the best of each and then putting it into a completely new engine, and then running -- and it's like building a new company.

核心主題其實是：我們想做的事，以及為什麼我們相信這裡會有回報，是因為你把同級最佳（best-of-breed）的功能，真正提供給我們 41 個市場的所有用戶。所以這不是把一套堆疊直接複製貼上到所有品牌。而是把各自最好的部分取出來，放進一個全新的引擎裡，然後運行——這就像在打造一家新公司。

And so -- and that's why it's really hard. And candidly, it's not something that you would do if you do not believe more in the future than you did in the past. And that doesn't even take into account the velocity benefits after you complete the project where if you were to ship once from an infrastructure perspective, that it actually gets shipped everywhere.

所以——這也是為什麼它真的很難。坦白說，如果你不比過去更相信未來，你是不會做這件事的。而這甚至還沒把專案完成後的速度效益算進去：從基礎設施的角度來看，如果你只要發布一次，實際上就能在所有地方同步上線。

Ronald Josey, Citigroup.

Ronald Josey，花旗集團。

Thanks for taking the questions. I wanted to go back to the gross profit and DashPass chart that we have in the letter here. And so some of the lines more recently are steepening for both, which is tracking all the trends that we saw. So I'd love to hear more maybe, Tony, on just the plans or strategies to continue adding value to the program and the push and pull what that value does to overall gross profit. And then more recently, I think the company launched newer or greater fees for larger delivery radiuses. Just talk to us about the reasoning for those fees and the benefits? Thank you.

謝謝讓我提問。我想回到股東信裡我們看到的毛利與 DashPass 圖表。因此最近兩條線的斜率都變得更陡，這與我們看到的所有趨勢一致。所以我很想多聽一些，也許 Tony，你們有哪些計畫或策略，持續為這個方案增加價值，以及這些價值的推拉效應如何影響整體毛利。另外更近期，我想公司針對較大的配送半徑推出了新的或更高的費用。請跟我們談談設定這些費用的原因與帶來的效益？謝謝。

Yeah. Hey, Ron, let me take a stab at both of those, right? Like look, subscription continues to do well. Our thesis has always been as consumers habituate and we graduate them to DashPass, their overall value proposition from not just a gross profit, but order frequency as well as retention goes up. And the example that we put in the chart was largely a Jan 2021 cohort, and we wanted to use that because it was sufficiently old enough where you could actually see the trends.

是的。嘿，Ron，我來嘗試回答這兩點，好嗎？你看，訂閱業務持續表現良好。我們的核心論點一直是：隨著消費者逐漸養成習慣並且我們把他們升級到 DashPass，他們的整體價值主張不僅在毛利上提升，訂單頻率以及留存也會提高。而我們放在圖表裡的例子主要是一個 2021 年 1 月的同批用戶（cohort），我們之所以用它，是因為它夠久，你才能真正看出趨勢。

Also, it's largely representative of what we see in the other cohorts in the rest of the portfolio as well. Where what we see is as consumers order more with us, as they retain more, they graduate to DashPass. And as they graduate to DashPass, they continue to spend more with us.

此外，它也在很大程度上代表了我們在投資組合中其他同批用戶所看到的情況。我們看到的是：當消費者在我們平台下單更多、留存更高時，他們就會升級到 DashPass。而當他們升級到 DashPass 後，他們也會持續在我們平台花更多錢。

And we're seeing that not just in the older cohorts, but in some of the newer cohorts as well, which is ultimately leading to some of the growth that you're seeing in DashPass, right? I talked about the fact that in US in DashPass Q2, the growth rate in terms of paid subscribers was one of the highest that we've seen probably in the last two years.

而我們看到的不只是較早期的同批用戶，在一些較新的同批用戶中也同樣如此，這最終帶動了你所看到的 DashPass 成長，對吧？我提到過，在美國市場，DashPass 在第二季（Q2）付費訂閱者的成長率，是我們可能在過去兩年中看到的最高之一。

We added more number of paid subscribers in the last year compared to the two prior years. A lot of that is because the underlying product continues to get better, right? Now if you're on DashPass, you get to access retail, grocery, other categories, which ultimately drives more value to DashPass subscribers.

我們在過去一年新增的付費訂閱者數量，超過了此前兩年的總和。其中很大一部分原因是底層產品持續變得更好，對吧？現在如果你是 DashPass 用戶，你可以使用零售、雜貨以及其他品類，這最終為 DashPass 訂閱者帶來更多價值。

The way we increase the value proposition of DashPass, to your question, is making the underlying product better, right? It's more selection, making the quality of the product better, continue to drive affordability. And when we do that, we see clear improvements in both adoption of DashPass as well as the engagement from a DashPass perspective.

針對你的問題，我們提升 DashPass 價值主張的方式，就是把底層產品做得更好，對吧？也就是更多選擇、提升產品品質、持續推動可負擔性。而當我們做到這些時，我們看到 DashPass 的採用率以及從 DashPass 角度的互動度都有明顯改善。

And your second point, look, I mean, if you think about the new fee service that we talked about last couple of weeks ago, look, it's largely a realignment of consumers, what they pay, compared to the time and effort that Dashers have put in a delivery. If you're thinking about it from like an impact to the P&L perspective, I wouldn't think of it that way. Based on what we've seen in the market so far, the fee is largely similar or slightly less actually for the vast majority of the orders. I wouldn't expect it to be a massive impact, especially in the market that we've launched it so far.

至於你的第二點，你看，如果你想到我們在幾週前談到的新服務費，這在很大程度上是把消費者支付的費用，與 Dashers 在一次配送中投入的時間與努力重新對齊。如果你從損益表（P&L）的影響角度來看，我不會那樣去理解。根據我們目前在市場上看到的情況，對於絕大多數訂單而言，這個費用大致相近，甚至實際上略低一些。我不預期會有很大的影響，尤其是在我們目前已推出的市場。

Justin Post, Bank of America.

Justin Post，美國銀行（Bank of America）。

Thanks for taking my question. Just wondering if you can give us any agentic traffic update, if you're seeing any traffic from there. And then given your huge merchant scale, are there ways where you could really capitalize on that traffic as they roll out booking capabilities and maybe even lower your marketing costs? Thank you.

謝謝讓我提問。想請問你們能否提供任何 agentic 流量的最新情況，是否有看到來自那邊的流量。另外，考量你們龐大的商戶規模，當他們推出預訂（booking）能力時，你們是否有方法能真正把那部分流量變現，甚至降低你們的行銷成本？謝謝。

Hey, Justin, I'll take that one and feel free to add, Ravi. I mean what I'd say, in short, is no, the volume is quite low, I think, from some of the agentic partners that we've been testing with. But it also isn't that surprising, I think, for a couple of reasons.

嘿，Justin，我來回答這題，Ravi 也可以隨時補充。簡單說，沒有，我認為我們與一些正在測試的 agentic 合作夥伴相比，流量規模相當低。但這其實也不太令人意外，我想原因有幾個。

I think, first, I think especially for some of the larger platforms out there, their core focus has been on the enterprise and much less, especially in coding agents and probably less on the agentic side. But the second thing is just structurally speaking, if you look at it from a consumer's perspective, consumers don't really care what you call this thing, whether you call it agentic flows and pre-agentic flows, post-agentic flows. They honestly just care about getting their burrito or their pair of Nike or their stock of weekly groceries. That's what they care about.

第一，我認為尤其對一些較大型的平台而言，他們的核心重點一直放在企業端（enterprise），而不是那麼著重在——特別是 coding agents，可能也比較少放在 agentic 這一側。第二點是從結構上來看，若從消費者角度出發，消費者其實不在乎你怎麼稱呼這個東西，不管你叫它 agentic flows、pre-agentic flows 或 post-agentic flows。他們老實說只在乎能拿到他們的墨西哥捲餅（burrito）、一雙 Nike，或是一週份的雜貨補貨。他們在乎的是這些。

And at the end of the day, what that means is they care about the end-to-end experience, right? And so if you think about it, we're effectively the only place that can offer that, whether that starts by knowing where all the inventory sits, what's in stock, what's not in stock, obviously, managing the logistics at both the merchant as well as a drop-off and then, of course, solving exception handling when things were to go awry if there's the wrong item or the wrong promotion applied to an item.

而歸根結底，這代表他們在乎的是端到端的體驗，對吧？所以如果你仔細想想，我們基本上是唯一能提供那種體驗的地方：從知道所有庫存在哪裡、哪些有貨、哪些缺貨開始，當然還包括同時管理商戶端與送達端的物流，然後，當事情出狀況時（例如拿錯商品或某個商品套用了錯誤的促銷），也要能處理例外狀況。

I think these are all the details you kind of have to get right if you want to do agentic commerce for our category. And that's just not something that I think a lot of people are doing. But we're kind of filling the void, right? It's why DoorDash launched DoorDash Ask, where we effectively are solving that. But we're still very open, very excited to test all sources of incremental traffic, which is what we believe can happen. It's something I said probably five years ago, is that in the business world, this is 2021, I was forecasting that there would be two big wars that would occur.

我認為，如果你想在我們這個品類做 agentic commerce，這些細節都必須做到位。而這不是我認為很多人正在做的事情。但我們某種程度上正在填補這個空缺，對吧？這也是為什麼 DoorDash 推出了 DoorDash Ask，因為我們實際上正在解決這些問題。不過我們仍然非常開放、也非常期待去測試所有增量流量來源，我們相信這是可能發生的。我大概五年前就說過，在商業世界裡——那是 2021 年——我當時預測會發生兩場大的戰爭。

One is kind of the battle for attention, and you see that playing out with chat assistants and other types of more powerful assistants now. And then the other is kind of the battle for atoms. And I do think that the two services will come and partner with one another. And our focus is squarely on making sure that we master the physical world so we can be the most useful to all these digital assistants when they kind of come around to focus on agentic commerce that will be willing partners and grow together.

一個是所謂的注意力之戰，而你現在看到它正透過聊天助理以及其他更強大的助理形式上演。另一個則是所謂的「原子之戰」（battle for atoms）。我確實認為這兩種服務最終會彼此合作、成為夥伴。而我們的重點非常明確：確保我們精通實體世界，這樣當這些數位助理開始把重心轉向 agentic commerce 時，我們就能對它們最有用，並成為願意合作的夥伴，一起成長。

Doug Anmuth, JPMorgan.

Doug Anmuth，摩根大通（JPMorgan）。

Great. Thanks for taking my question. I know you don't manage for take rate, but I just wanted to get a little bit more color just on net revenue margin and the pickup that you saw kind of in 2Q. Just trying to understand some of the drivers there in terms of how much might have been Deliveroo contribution versus advertising and fee changes and how you think about that going forward? Thanks.

很好。謝謝讓我提問。我知道你們不以抽成率（take rate）作為管理目標，但我想更了解一下淨營收利潤率（net revenue margin）以及你們在第二季（2Q）看到的提升。我想理解其中一些驅動因素：有多少可能來自 Deliveroo 的貢獻、相較之下廣告與費用調整的影響各是多少，以及你們未來如何看待這件事？謝謝。

Hey, Doug, let me take that one. Look, I mean, you're right. I mean we're not operating the business towards a take rate or net revenue margin percentage. Our goal has been always to optimize for overall profit dollars. And as you can see, a pretty strong quarter from that perspective as well as the Q3 guide.

嘿，Doug，我來回答這題。你看，你說得沒錯。我們並不是以抽成率或淨營收利潤率百分比來經營這門生意。我們的目標一直是優化整體利潤金額（overall profit dollars）。而如你所見，從這個角度來看，本季表現相當強勁，第三季（Q3）的指引也是如此。

Look, lots of moving parts within the take rate. The Q-on-Q increase, which is what I think you're referring to, is largely from Dasher. Look, Dasher costs are seasonal for us. When you go from Q1 to Q2, Dasher costs are lower in Q2. That's what gave rise to sort of the tick up in take rate that you saw from Q1 to Q2.

你看，抽成率裡面有很多變動因素。你提到的那個季對季（Q-on-Q）上升，主要來自 Dasher。你看，Dasher 成本對我們而言具有季節性。當你從第一季走到第二季，第二季的 Dasher 成本會比較低。這就是你看到從第一季到第二季抽成率略為上升的原因。

And if you're thinking about from a modeling perspective, what I would expect for the rest of the year, I'd expect take rate to slightly be in the similar range in Q3, so flattish from Q2 to Q3, and lower in Q4. Again, as a reminder, I mentioned this before, Q4 Dasher costs are higher for us. That will impact take rates. So you should expect Q4 take rate to be slightly lower than where Q3 is. But net-net, look, I mean, the goal for us is not to manage to a specific line in the P&L, especially the take rate percentage. Our goal is to invest flexibly up and down the P&L.

如果你從建模的角度來看，對於今年剩下時間我會怎麼預期，我會預期第三季的抽成率大致會落在相近區間，所以從第二季到第三季大致持平，而第四季會更低。再次提醒，我之前提過，第四季我們的 Dasher 成本較高。這會影響抽成率。所以你應該預期第四季的抽成率會比第三季略低。但總體而言，你看，我們的目標不是去管理損益表（P&L）上的某一條特定線項，尤其不是抽成率百分比。我們的目標是在損益表上更有彈性地進行投資，上下調整。

Tom Champion, Piper Sandler.

Tom Champion，Piper Sandler。

Good afternoon. Thanks for taking the question. Ravi, I'd just be curious about your big picture view on AI spending and the ROI that you're seeing. And I'm curious if it's impacted your future hiring plans at all. Thank you.

下午好。謝謝讓我提問。Ravi，我想了解你對 AI 支出的整體看法，以及你目前看到的投資報酬率（ROI）。我也想知道這是否對你們未來的招募計畫有任何影響。謝謝。

Sure. I mean I think -- let me start with -- I mean, it's not just purely about driving cost efficiency for us, right? Like we said earlier on the call, look, we are encouraging our teams to use AI across the board. For us, ultimately, the goal is how do you build better products for customers, which will ultimately drive both growth as well as overall profit dollars. Ask was one of the examples in which we're driving benefit to customers.

當然。我的意思是，我想——讓我先從——我的意思是，這對我們來說不只是單純在推動成本效率，對吧？就像我們在電話會議前面提到的，你看，我們鼓勵各團隊全面使用 AI。對我們而言，最終目標是如何為客戶打造更好的產品，這將最終同時帶動成長以及整體利潤金額。Ask 就是其中一個例子，我們透過它為客戶帶來效益。

We've done similar things on both merchant side where we're helping merchants onboard faster. On the Dasher side, we build conversational bots where Dashers, if they're stuck, they can actually chat with the agent to help themselves get unstuck.

我們在商家端也做了類似的事情，協助商家更快完成上線／導入。在 Dasher 端，我們建立了對話式機器人，Dashers 如果遇到卡關，實際上可以和客服人員聊天，幫助自己解決卡住的問題。

At the same point, look, we're seeing productivity gains across the board. Internally, we're using it in sales, accounting, marketing, finance as well. The goal for us has always been it's not just purely about encouraging the usage, how do you actually drive efficiency as well.

同時，你看，我們也看到整體各方面的生產力提升。在內部，我們也把它用在銷售、會計、行銷、財務等領域。我們的目標一直是：不只是鼓勵使用而已，也要思考如何真正帶動效率提升。

We've taken a number of steps. We've built models where internally, the tasks are routed to the right model, depending on what the actual cost, quality and efficiency is. We've put caps in place. We've also incorporated some of the AI budgets into teams' existing budgets.

我們採取了多項措施。我們建立了模型，讓內部的任務依據實際成本、品質與效率，被分派到最合適的模型。我們也設置了上限。此外，我們也把部分 AI 預算納入各團隊既有的預算之中。

Look, we are seeing gains from the usage of AI. The sharper question for us is how do you take the efficiency gains and reinvest that back in the business. Look, we are investing in building some large areas, right? We've talked about autonomy. We've talked about AI. We've talked about the unification of the tech stack.

你看，我們確實從 AI 的使用中看到成效。對我們而言更尖銳的問題是：如何把效率提升帶來的收益再投資回業務。你看，我們正在投資打造幾個大型領域，對吧？我們談過自動化／自主化（autonomy）。我們談過 AI。我們也談過技術堆疊（tech stack）的整合統一。

These are all areas where we think it's going to be strong long-term ROI for us. The goal for us is how do you take the efficiency gains, whether it's AI or any other part of the P&L. But the philosophy is the same, right? How do you reinvest that back in the business ultimately to build scale and durability over a longer period of time, which leads to higher overall free cash flow production. That's largely how we're thinking about the efficiency gains, but we are happy with what we're seeing in the business today.

這些都是我們認為長期將帶來強勁投資報酬率（ROI）的領域。對我們而言，目標是如何把效率提升的成果——不論來自 AI 或損益表（P&L）的任何其他部分——加以運用。但理念是一樣的，對吧？如何把這些再投資回業務，最終在更長的時間尺度上建立規模與韌性，進而帶來更高的整體自由現金流產出。這大致就是我們看待效率提升的方式；而且我們對目前在業務中看到的成果感到滿意。

There are no further questions at this time. This concludes today's call. Thank you for attending. You may now disconnect.

目前沒有其他問題。今天的電話會議到此結束。感謝各位參與。您現在可以中斷連線。