Opendoor Technologies Inc. (OPEN) 2026 Q2 法說會逐字稿

Hey, everyone. Welcome to Opendoor's second-quarter 20,226 financial open-house earnings live stream. I'm Mchael Judd, Opendoor's Head of Investor Relations. Now, a few quick housekeeping items before we get started. Like all things Opendoor, we're going to do this faster.

各位好。歡迎收看 Opendoor 20,226 年第二季財務開放日（open-house）財報直播。我是 Michael Judd，Opendoor 投資人關係主管。在開始之前，先快速說幾項例行事項。就像 Opendoor 的一切一樣，我們會更快地進行。

Details of our results and additional management commentary are available in our earnings release, which can be found at investor.opendoor.com.

我們的業績細節與管理層補充評論已刊載於財報新聞稿，可至 investor.opendoor.com 查閱。

The following discussion contains forward-looking statements within the meaning of the federal securities laws. All statements other than statements of historical fact are statements that could be deemed forward-looking, including, but not limited to, statements regarding Opendoor's financial condition, anticipated financial performance, business strategy and plans, market opportunity and expansion, and management objectives for future operations. These statements are neither promises nor guarantees, and undue reliance should not be placed on them. Such forward-looking statements involve risks and uncertainties that may cause actual results to differ materially from those discussed here.

以下討論包含符合聯邦證券法定義的前瞻性陳述。除歷史事實陳述外，所有陳述均可能被視為前瞻性陳述，包括但不限於關於 Opendoor 的財務狀況、預期財務表現、業務策略與計畫、市場機會與擴張，以及管理層對未來營運的目標。這些陳述既非承諾亦非保證，不應對其過度依賴。此類前瞻性陳述涉及風險與不確定性，可能導致實際結果與此處討論內容出現重大差異。

Additional information that could cause actual results to differ from forward-looking statements can be found in the risk factors section of Opendoor's most recent annual report on Form 10-K for the year ended December 31, 2025, as updated by our periodic reports and other filings with the SEC.

可能導致實際結果與前瞻性陳述不同的其他資訊，請參閱 Opendoor 最近一期截至 2025 年 12 月 31 日止年度之 Form 10-K 年報「風險因素」章節，以及我們向美國證券交易委員會（SEC）提交的定期報告與其他申報文件中的更新內容。

Any forward-looking statements made on this webcast, including responses to your questions, are based on management's reasonable current expectations and assumptions as of today, and Opendoor assumes no obligation to update or revise them, whether as a result of new information, future events or otherwise, except as required by law.

本次網路直播中所作的任何前瞻性陳述（包括對各位提問的回覆），均基於管理層截至今日合理的現行預期與假設；除法律要求外，Opendoor 不承擔因新資訊、未來事件或其他原因而更新或修訂該等陳述的任何義務。

The following discussion contains references to certain non-GAAP financial measures. The company believes these non-GAAP financial measures are useful to investors as supplemental operational measurements to evaluate the company's financial performance. For a reconciliation of each of these non-GAAP financial measures to the most directly comparable GAAP metric, please see our website at investor.opendoor.com.

以下討論將提及若干非 GAAP 財務衡量指標。公司認為，這些非 GAAP 財務衡量指標可作為補充性的營運衡量，協助投資人評估公司的財務表現，具有參考價值。如需將各項非 GAAP 財務衡量指標調節至最直接可比的 GAAP 指標，請至 investor.opendoor.com 查閱。

And with that, let's get into the open house with Kaz and Christy.

那麼，接下來就讓我們與 Kaz 和 Christy 一起進入 open house。

Good afternoon, everyone. I usually start these calls by showing you a clip of what I told you during the last call. But this time, I’m going to tell you a story about what my wife told me. And I don’t have a video clip because it’d be weird if my wife and I just recorded each other all the time. So you’re just going to have to use your imagination.

各位下午好。我通常會以播放一段上次電話會議我說過的內容作為開場。但這次，我要跟各位講一個我太太跟我說的故事。而且我沒有影片片段，因為如果我和我太太一直互相錄影，那會很怪。所以你們只能靠想像了。

When I was leaving home to fly to San Francisco before my first day at Opendoor, I told my wife that I’d be back home the following Wednesday, maybe Thursday, and she didn’t miss a beat. She said, don’t come back until there’s a plan to breakeven. Look, there’s a lot of ways people describe the thing I’m about to tell you. It just depends on which tribe they’re a part of, right?

在我加入 Opendoor 的第一天前、準備離家飛往舊金山時，我跟我太太說我下週三、也許週四就會回家，她完全不假思索。她說：在有損益兩平的計畫之前，不要回來。你看，大家描述我接下來要講的這件事，有很多種說法。這其實取決於他們屬於哪個「部落」，對吧？

Paul Graham has a famous essay about it. Finance people call it a glide path to profitability, mostly because I think finance people are legally required to say things like glide path. But basically, the question is this, if nothing changes and you keep doing what you’re doing, what happens? Let’s see. There’s no magic here. It’s just math.

Paul Graham 有一篇很有名的文章談到這件事。財務圈的人稱它為「通往獲利的滑行路徑」（glide path to profitability），主要是因為我覺得財務人員在法律上好像被要求一定要講「滑行路徑」這種詞。但基本上，問題是：如果什麼都不變、你持續做你正在做的事，會發生什麼？我們來看看。這裡沒有魔法。只有數學。

This next section, it’s going to take me a few minutes, but it’s incredibly important. I want to give you all the same framework we use internally so you can see things the way we’re seeing them right now. Opendoor’s core business math is simple. It’s how many homes we transact on times our contribution margin minus our OpEx and our financing costs. Let’s go through each of these four numbers.

接下來這一段會花我幾分鐘，但它非常重要。我想把我們內部使用的同一套框架提供給各位，讓你們能用我們此刻的視角來看事情。Opendoor 核心業務的數學其實很簡單。就是：我們成交的房屋數量 × 我們的貢獻毛利（contribution margin）－ 我們的營運費用（OpEx）與融資成本。我們逐一拆解這四個數字。

First, volume. Right now, we’re signing more than 500 contracts every single week. Last week, we signed around 700. That’s our highest contract week in years. That’s over 5x higher year over year and 5x higher since I joined the company. And just think about when we’re doing this. We’re doing this in the weakest housing market in a generation and in the worst season of the year for us. The spring and summer seasons are basically the only times of the year where the traditional real estate system still kind of actually works, right?

第一，量（volume）。目前我們每週簽下超過 500 份合約。上週我們簽了大約 700 份。那是我們多年來合約數最高的一週。這比去年同期高出超過 5 倍，也比我加入公司時高出 5 倍。而且想想我們是在什麼情況下做到的。我們是在一代人以來最疲弱的房市、以及對我們而言一年中最差的季節做到的。春季和夏季基本上是一年中傳統房地產體系仍然還算能運作的唯一時段，對吧？

Over 500 sellers are still saying yes to Opendoor every single week. If you’ve been following along on accountable.opendoor.com, you’ve seen this, right? When we put up our ranges last quarter, those numbers weren’t sandbags. Those were numbers we thought we’d see to put us on the path we’d need to be at the end of this year. The fact that we have been above the high end of the projections every single week for the last quarter is the reason that I’m so confident about what I’m about to say.

每週仍有超過 500 位賣方對 Opendoor 說「好」。如果你一直在關注 accountable.opendoor.com，你已經看到了，對吧？我們上季公布的區間數字並不是刻意保守（sandbag）。那是我們認為會看到的數字，好讓我們走在今年年底所需的路徑上。而我們在過去一季的每一週都高於預測區間上緣，正是我對接下來要說的話如此有信心的原因。

Look, our ANI breakeven framework assumes 6,000 quarterly transactions at $375,000 each for around $9 billion in revenue. We’re pacing well above that on a contract basis in Q2. And look, some of these won’t close. That’s normal. That hasn’t changed. That isn’t the point. The point is we’ve achieved this in a market and at the time of the year that is the worst for us.

你看，我們的 ANI 損益兩平框架假設：單季 6,000 筆交易、每筆 375,000 美元，約可帶來 90 億美元營收。以第二季的合約簽訂節奏來看，我們明顯高於這個水準。當然，其中有些不會完成交割。這很正常。這點沒有改變。重點不在這裡。重點是：我們是在對我們最不利的市場環境、以及一年中最不利的時間點做到的。

I think it’s super reasonable to expect that we’re going to end up north of the $9 billion mark in revenue. Now, it’s important to take a second and describe a pattern that we’re seeing as we ramp up for this revenue. Look, companies have kind of two ways to artificially increase growth in absence of actual improvement in the company, right? Those two levers are marketing and pricing. I’m going to get to marketing in a second, but I want to start by talking about pricing.

我認為非常合理可以預期，我們最終的營收會高於 90 億美元這個門檻。現在，先花一點時間描述一下：當我們為了達成這個營收規模而加速時，我們觀察到的一個模式。你看，公司在缺乏實質改善的情況下，通常有兩種方式可以「人為」拉高成長，對吧？這兩個槓桿是行銷與定價。我等一下會談行銷，但我想先從定價開始。

If you’ve followed Opendoor for a while, when you see high acquisition volumes, you should be skeptical. You should be asking, are we buying growth through more risk and lower spreads? Or using regular language that everyday people use, are we paying more than fair prices for homes? Opendoor did this during the COVID era, right? To get volume, it took lots of risks because it believed that price was the only lever it had at its discretion to increase conversion. And if it couldn’t reduce its overall cost structure, it would lose money.

如果你關注 Opendoor 一段時間了，當你看到很高的收購量時，你應該保持懷疑。你應該問：我們是不是透過承擔更高風險、以及更低的價差（spreads）來買來成長？或者用一般人更常用的說法：我們是不是用高於合理價格去買房？Opendoor 在 COVID 時期確實這麼做過，對吧？為了衝量，它承擔了很多風險，因為它相信在自己可控的範圍內，價格是它唯一能用來提高轉換率的槓桿。而如果它無法降低整體成本結構，就會虧錢。

What I’m about to show you is probably the slide that has made me most excited about what we’ve gotten done so far this year. This shows our true seller conversion at different spread levels. So true sellers are the people who request an offer from Opendoor and then either sell to Opendoor or list on the open market. So it’s the conversion on people who actually want to sell their home.

我接下來要給各位看的，可能是今年到目前為止，最讓我對我們所完成的成果感到興奮的一張投影片。這張圖顯示我們在不同價差（spread）水準下的真實賣方轉換率。所謂真實賣方，是指向 Opendoor 申請報價，之後要嘛賣給 Opendoor、要嘛在公開市場掛牌出售的人。因此，這是針對真正想賣房的人所計算的轉換率。

What this chart shows is that we are converting dramatically more sellers at the same spread levels than we’ve had in the past. We’re on track to hit more than the volume we need, and we’re not doing it by paying above fair prices for homes. That should mean something for our contribution margin. So let’s talk about that.

這張圖表顯示，在相同的價差水準下，我們轉換的賣方數量比過去大幅增加。我們正朝著達成所需量能以上的目標前進，而且不是靠用高於合理價格去買房來做到的。這對我們的貢獻毛利（contribution margin）應該會有意義。那我們就來談談這件事。

Our contribution margins have been improving every single quarter this year and are now in the target range that we told you we would be in. There are about 100 or so homes left from the Opendoor 1.0 era that are going to be a drag on our contribution margin, but they’re going to be sold mostly this quarter and we’re going to be done with them, and I’m never going to talk about them again.

今年以來，我們的貢獻毛利每一季都在改善，現在已經落在我們先前告訴各位的目標區間內。大約還有 100 間左右屬於 Opendoor 1.0 時代的房屋，會拖累我們的貢獻毛利，但它們大多會在本季售出，之後我們就會把它們處理完，而且我也再也不會提它們了。

But on the new cohorts, our margin is performing and cohort curves are doing what we want them to do. Now, look, it’s for sure true that Q3 is a seasonally worse contribution margin quarter for us than Q2. You should expect quarter-over-contribution margin to go down from Q2 to Q3.

但在新的批次（cohorts）上，我們的毛利表現良好，批次曲線也正如我們所期望的那樣運作。現在，先說清楚，對我們而言，第三季（Q3）在季節性上確實比第二季（Q2）的貢獻毛利更差。你應該預期貢獻毛利會從 Q2 下降到 Q3。

This year, we also have the impact of the Doma acquisition, which is a temporary drag on contribution margin while we integrate Doma into Opendoor. But we expect to break with Opendoor’s historical trend and have a Q4 that is higher in contribution margin than Q3. Every year in Opendoor’s public company history, Q4 has had a worse margin than Q3, and we’re about to reverse that trend.

今年我們也受到收購 Doma 的影響；在我們把 Doma 整合進 Opendoor 的期間，這會對貢獻毛利造成暫時性的拖累。但我們預期會打破 Opendoor 過去的歷史趨勢，讓第四季（Q4）的貢獻毛利高於第三季（Q3）。在 Opendoor 成為上市公司以來的每一年，Q4 的毛利都比 Q3 更差，而我們即將扭轉這個趨勢。

But going back to the main point. In Q2, we got to a contribution margin zone we told you we’d aim for, and we have proven that we can run the company here. Okay, so acquisition volume is tracking to where we want it to be, and margin is within the range we told you we would have to be in. That leaves OpEx and financing.

不過回到重點。在 Q2，我們達到了先前告訴各位我們要努力達成的貢獻毛利區間，而且我們已經證明公司可以在這個區間內運作。好，所以收購量（acquisition volume）正朝我們希望的方向推進，毛利也在我們說必須達到的範圍內。剩下的就是營業費用（OpEx）和融資。

Let’s talk about OpEx. So Opendoor’s OpEx includes marketing, variable operations, which we call just operations in our financials, and fixed operations. So most of our costs happen when we buy and renovate homes. So it’s useful to look at these costs in relation to our acquisition numbers since that is the variable that scales them, right?

我們來談談營業費用（OpEx）。Opendoor 的 OpEx 包括行銷、可變動營運（我們在財報中通常就稱為營運），以及固定營運。我們的大部分成本發生在我們買房與翻修房屋的時候。因此，把這些成本與我們的收購數量相對照來看會很有用，因為收購數量是會帶動它們規模變化的變數，對吧？

So we have this idea called acquisition GMV, which is roughly the revenue we expect to get from homes we’ve closed on. Our homes have been selling right around the $375,000 mark, so let’s use that assumption so I don’t have to leak our internal model to the world. It’ll also make the math easier. So acquisition GMV is acquisitions times $375,000.

所以我們有一個概念叫做收購 GMV（acquisition GMV），大致上代表我們預期從已完成交割的房屋中取得的收入。我們的房屋售價大約落在 37.5 萬美元左右，所以我們就用這個假設，這樣我就不用把我們的內部模型洩漏給全世界。這也會讓數學更容易。因此，收購 GMV 就是收購數量乘以 37.5 萬美元。

With that in mind, let’s talk about the three parts of our OpEx. First, marketing. Marketing is our cost of customer acquisition per home we buy. We do basically no marketing when we sell homes. And the last time we signed more than 6,000 contracts in a quarter, our marketing spend was over $80 million.

在此基礎上，我們來談 OpEx 的三個部分。第一，行銷。行銷是我們每買下一間房的客戶取得成本。我們在賣房時基本上幾乎不做行銷。而上一次我們在單季簽下超過 6,000 份合約時，我們的行銷支出超過 8,000 萬美元。

This quarter, it was $5 million, not 5-0, just 5. I know that sounds crazy, but yes, Morgan actually wrote the book on this. And yes, this is one part of Opendoor where I think we’re executing 10 out of 10. It took us a minute, but we have found our groove. Our marketing as a percentage of our acquisition GMV has gone from 1.6% under Opendoor One to 0.3%.

本季是 500 萬美元，不是 5,000 萬，就是 500 萬。我知道這聽起來很瘋狂，但沒錯，Morgan 其實在這方面寫過「教科書」。而且沒錯，這是 Opendoor 我認為我們執行力可以打 10 分滿分的其中一個領域。我們花了一點時間，但我們已經找到節奏了。我們的行銷費用占收購 GMV 的比例，已從 Opendoor One 時期的 1.6% 降到 0.3%。

Let’s say we’ll not be that good forever. Let’s assume we’ll get a little worse. Let’s assume we’ll have 0.5% of acquisition GMV on marketing. Just remember that, okay?

假設我們不可能永遠都這麼厲害。假設我們會稍微變差一點。假設行銷費用會是收購 GMV 的 0.5%。請記住這個數字，好嗎？

Next, let’s talk about variable operations. Variable operations are mostly the costs we incur when we’re buying and renovating our houses, right? This is the human and system cost of underwriting, buying, and renovating a home. So variable ops have declined from 2.7% of acquisition GMV to 0.9% of acquisition GMV.

接下來談可變動營運（variable operations）。可變動營運主要是我們在買房與翻修房屋時產生的成本，對吧？這是對房屋進行核保（underwriting）、購買與翻修的人力與系統成本。因此，可變動營運費用已從收購 GMV 的 2.7% 降到收購 GMV 的 0.9%。

If you look at acquisition contract GMV, this number is already at 50 basis points. But look, don’t give us the benefit of this doubt. There’ll be some contracts that will fall through, so let’s mark this up. Let’s say it’ll be 70 basis points on acquisition GMV.

如果你看的是收購合約 GMV（acquisition contract GMV），這個數字已經是 50 個基點。但你看，我們不要假設對我們最有利的情況。會有一些合約最後無法成交，所以我們把它往上調。就假設它會是收購 GMV 的 70 個基點。

Why has this number gone down so much? The answer is simple, and Vu would tell you, we are among the best users of AI in tech. We have fundamentally re-engineered our business around AI automation across underwriting and operations.

為什麼這個數字會下降這麼多？答案很簡單，Vu 也會告訴你：我們是科技業中最擅長使用 AI 的公司之一。我們已經以 AI 自動化為核心，從核保到營運，徹底重新設計了我們的業務。

Look, in Q3 last year, the people who managed renovations of our homes, our HPMs, they carried three renovations per person per month. Right now, they’re carrying around 10. By the end of this year, they will carry around 20.

你看，在去年第三季（Q3），負責管理房屋翻修的同仁，也就是我們的 HPM，每人每月大約只能同時管理 3 個翻修案。現在，他們大約能管理 10 個。到今年年底，他們將能管理大約 20 個。

In Q3 last year, our pricing underwriting team could handle around 20 underwrites per person per day. They can now do over 50. They’ll be able to handle around 100 by the end of the years.

在去年第三季（Q3），我們的定價核保團隊每人每天大約能處理 20 件核保。現在他們可以做超過 50 件。到今年年底，他們將能處理大約 100 件。

A couple of years ago, more than 80% of homes that we bought required an Opendoor employee to visit them before we even made an offer. Today, that percentage is below 20%, and it’ll go down to below 10% by the end of this year. So 70 basis points, just remember that.

幾年前，我們買下的房屋中，有超過 80% 在我們甚至提出報價之前，都需要 Opendoor 的員工親自到現場看屋。今天，這個比例已低於 20%，而且到今年年底會降到 10% 以下。所以，70 個基點，請記住這個數字。

The last component is our fixed operations. These are the people that build the machine that runs Opendoor. Last year, there were a lot of consultants in G&A. Now it’s a lot of engineers and data scientists writing code. One of the beautiful things about code is that it scales really well. Spinning up a server is easier in hiring a new consulting firm. And you can see this in our numbers.

最後一個組成部分是固定營運（fixed operations）。這些人是在打造並維運驅動 Opendoor 運作的機器。去年，在一般與行政（G&A）裡有很多顧問。現在則有很多工程師與資料科學家在寫程式。程式碼有個很美的特性：它非常容易擴展。啟動一台伺服器比聘請一家新的顧問公司更容易。而你也可以從我們的數字中看到這一點。

In Q3 last year, this number was over 8% of our acquisition GMV and over 6% of our acquisition contract, right? In Q2, this was 2.1% of acquisition GMV and 1.4% of acquisition contract GMV.

去年第三季，這個數字超過我們收購 GMV 的 8%，也超過我們收購合約的 6%，對吧？第二季，這是收購 GMV 的 2.1%，以及收購合約 GMV 的 1.4%。

Let’s say it’ll be somewhere between those two numbers. Let’s pick, I don’t know, 1.7%. So 50 basis points on marketing, 70 basis points on variable operations, and 1.7% on fixed OpEx. Add all those up and it’s 2.9% on our acquisition GMV, which doesn’t give us the benefit of our increased acquisitions that we’re making right now.

假設它會落在這兩個數字之間的某個位置。我們就選一個，我不知道，1.7%。所以行銷 50 個基點、可變營運 70 個基點，以及固定營業費用（OpEx）1.7%。把這些加總起來，就是我們收購 GMV 的 2.9%，而這還沒有把我們目前正在增加的收購量所帶來的好處算進去。

If you take the dollars spent and compare them to our $9 billion run rate, that’s 2.4%. That’s less than the low end of 3% to 4% range I told you we would need to get to become ANI profitable.

如果把花出去的美元金額拿來跟我們 90 億美元的年化運行率相比，那就是 2.4%。這低於我告訴你們、我們要達到 ANI 獲利所需的 3% 到 4% 區間下緣。

Okay. The last component here is interest. Look, we finance the homes we buy. So this scales with how many homes we buy and how long we own them. At the speed that we turn inventory right now, that’s about three times a year. So net interest runs a little above 2% on our revenue.

好。這裡最後一個組成項是利息。你看，我們會為我們買下的房屋提供融資。所以這會隨著我們買多少房、以及我們持有它們多久而擴大或縮小。以我們目前的庫存周轉速度來看，大約一年三次。因此，淨利息約略高於我們營收的 2%。

We’re working on lots of things that will lower this, and I think there’s a lot of upside here. But don’t give us the benefit of the doubt. Let’s assume they just stay where they are. So what would happen if we just froze the company? Like if we pretend that we don’t improve anything, no new products, no funnel improvements, no pricing model updates, no rate cuts, no macro rescue. No one has to be a hero. We just keep doing exactly what we did last week.

我們正在推動很多事情來降低這個數字，我認為這裡有很大的上行空間。但先不要給我們「疑點利益」。就假設它們就維持在目前的水準。那如果我們把公司「凍結」會怎樣？就像假裝我們什麼都不改善：沒有新產品、沒有漏斗改善、沒有定價模型更新、沒有降息、沒有宏觀環境的救援。不需要任何人當英雄。我們就照著上週做的事情一模一樣地繼續做。

In fact, the math I just showed you assumes that we would get worse at marketing and operations, which we for sure won’t. But what happens if that happens? Current volumes, current margins, worse cost structures, current financing costs. Run those numbers forward, and this is the back of a napkin math, right? At roughly 6,000 transactions per quarter, Opendoor reaches adjusted net income profitability on a run rate basis heading into next year. No new assumptions, just today’s company carried forward.

事實上，我剛剛給你們看的數學假設我們在行銷和營運上會變得更差，而我們肯定不會。但如果真的發生了會怎樣？目前的量、目前的毛利、更差的成本結構、目前的融資成本。把這些數字往前推，這就是餐巾紙背面的粗算，對吧？在每季大約 6,000 筆交易的水準下，Opendoor 以年化運行率口徑在明年初將達到調整後淨利（Adjusted Net Income）獲利。不需要新的假設，只是把今天的公司狀態延續下去。

Look, I told you this on our very first call. I’ve said it on every call since then. Adjusted net income breakeven on a 12-month go-forward basis by the end of 2026. When I said this, there were more than a few things that needed to go right, but we moved fast, we shipped, we took charge of our own company every single week, and every quarter, the math became more and more obvious.

你看，我在我們第一次電話會議就跟你們說過。從那之後每一次電話會議我都說過。到 2026 年底，以未來 12 個月前瞻基礎達到調整後淨利（ANI）損益兩平。我說這句話時，需要有不少事情順利發生，但我們動作很快、我們交付成果、我們每週都掌控自己的公司，而每一季，這套數學就變得越來越明顯。

Yang, our Chief Investment Officer, he runs our pricing and meme teams at Opendoor. He has a simple way of putting these things. He likes it so much that he actually bought a T-shirt at a nerd convention and wears it to the office. The T-shirt says, it’s just math.

我們的投資長 Yang，他在 Opendoor 負責我們的定價和 meme 團隊。他有一個很簡單的說法來總結這些事。他喜歡到甚至在一個宅宅展會上買了一件 T 恤，然後穿來辦公室。那件 T 恤上寫著：這只是數學。

For years, one question has followed this company everywhere it went. Will Opendoor become profitable? As of today, that question is boring. As things stand right now, Opendoor will become ANI profitable. It’s just math. Sweetheart, after this call, I am coming home.

多年來，有一個問題不管這家公司走到哪裡都如影隨形。Opendoor 會獲利嗎？截至今天，這個問題已經變得很無聊。以目前的狀況來看，Opendoor 會達到 ANI 獲利。這只是數學。親愛的，這通電話結束後，我就回家。

So the last few minutes have sounded like a confident CEO, 10 months into a turnaround, holding a napkin that says everything is working. But let me tell you what’s not on the napkin. Here’s the uncomfortable truth. Nothing about this has been easy. Turnarounds are really, really hard, and there’s obviously a bit of a survivor bias here, right? Everyone knows about the ones that worked in retrospect.

所以，過去幾分鐘聽起來像是一位自信的 CEO，在轉型 10 個月後，拿著一張寫著「一切都在奏效」的餐巾紙。但我想告訴你們，餐巾紙上沒有寫的是什麼。這裡有個令人不舒服的真相。這一切沒有任何一部分是容易的。轉型真的、真的很難，而且顯然這裡也有一點幸存者偏差，對吧？大家回頭看都只知道那些成功的案例。

But why will this one work? We’re 10 months now into this process and really proud of what we’ve done. If we freeze the company, I just told you we would become ANI profitable even if the macro keeps punching us in the face. Look, we’re going to become ANI profitable on the path to fulfilling our mission and becoming a meaningful company for this country.

但為什麼這一次會成功？我們進入這個流程已經 10 個月了，對我們所做的一切感到非常自豪。如果我們把公司凍結，我剛剛告訴你們，即使宏觀環境持續狠狠打擊我們，我們也會達到 ANI 獲利。你看，我們會在履行使命、並成為對這個國家有意義的公司這條路上，達到 ANI 獲利。

But there’s something else that I need to tell you. The math works, and we’re going to become ANI profitable. We can very clearly see that right now. But that does not mean that everything between here and there will be just perfectly smooth. Turnarounds are hard, and they surprise you sometimes. Sometimes these surprises are good, and I want to talk about two of them.

但還有另一件事我需要告訴你們。這套數學是成立的，我們會達到 ANI 獲利。我們現在非常清楚地看得到這一點。但這不代表從現在到那時之間的一切都會完全順風順水。轉型很難，而且有時會給你驚喜。有些驚喜是好的，我想談其中兩個。

First is seasonality. Opendoor’s business has had a real seasonality to it. We’ve traditionally been a company that’s had feast in Q2 and famine the rest of the year. In fact, every year since we’ve been public, other than 2023, our margin degradation between Q2 and Q3 has averaged almost 500 basis points.

第一個是季節性。Opendoor 的業務一直具有明顯的季節性。傳統上，我們是一家第二季大豐收、其餘時間相對清淡的公司。事實上，自從我們上市以來，除了 2023 年之外，每一年我們在第二季到第三季之間的毛利惡化平均接近 500 個基點。

Look, this year we haven’t killed off seasonality entirely, so it won’t be zero, but it’ll be way, way, way less. Folks won’t appreciate why this is a big deal, and they won’t appreciate it for a while. But I think of everything we have done this quarter. This compression may be the most important thing for the long-term health of the company. That’s the first thing.

你看，今年我們還沒有把季節性完全消除，所以不會是零，但會小非常非常非常多。大家不會立刻理解為什麼這很重要，而且一段時間內也不會真正體會。但回想我們這一季做的所有事情。這種壓縮可能是公司長期健康最重要的一件事。這是第一點。

The second one is the embedded impact of things that have already happened but don’t look like they have. When a seller signs a contract with us, that home becomes revenue a few months later, right? We buy, we fix, we renovate, we list, we sell. And we can’t really skip a step in between, right? It just takes time. This means I always live a few months in the future.

第二個是那些已經發生、但看起來好像還沒發生的事情所帶來的內嵌影響。當賣方與我們簽約時，那間房子會在幾個月後才變成營收，對吧？我們買下來、修繕、翻新、掛牌、出售。而且我們中間其實不能真的跳過任何一步，對吧？這就是需要時間。這意味著我總是活在幾個月後的未來。

Every week, contracts turn into homes, which turn to listings, which turn to closings. What I see today in our acquisition contracts turns into GAAP revenue a few months from now based on our conversion rates. It’s why we put our weekly contracts on accountable.opendoor.com, so you can see what we see in real time a few months sooner.

每一週，合約會變成房屋，房屋會變成掛牌，掛牌會變成交割完成。我今天在收購合約中看到的，會依照我們的轉換率，在幾個月後轉成 GAAP 營收。這也是為什麼我們把每週合約數放在 accountable.opendoor.com 上，讓你們能比其他人早幾個月、即時看到我們所看到的。

C.S. Lewis has a famous metaphor about watching a horse grow wings. As a father of daughters, I’m a bit of an expert on this topic. This is technically a Pegasus, not a unicorn. The horse in transition for the first little while would look a little odd. This is a horse that was running fast, and it’s going to run a little awkwardly right now because he’s growing these bumps on his backs that are going to become wings, but aren’t quite yet. He doesn’t take off until the wings grow.

C.S. Lewis 有一個著名的隱喻，談的是看著一匹馬長出翅膀。身為女兒們的父親，我在這個主題上算是有點專家。嚴格來說，這在技術上是飛馬（Pegasus），不是獨角獸。那匹馬在轉變的最初一小段時間，看起來會有點怪。這是一匹原本跑得很快的馬，而他現在會跑得有點彆扭，因為他的背上長出一些隆起，將來會變成翅膀，但還沒完全長成。直到翅膀長好，他才會起飛。

Silicon Valley has spent the last 20 years chasing unicorns. And we’re coining a new category, the Pegasus. Not a company that was magical from the beginning, but a company that had to grow its wings in public. This is what transformations look like midstream. The changes are real before the financial statements catch up.

矽谷在過去 20 年一直在追逐獨角獸。而我們正在創造一個新類別：飛馬。不是一開始就很神奇的公司，而是一家必須在眾目睽睽之下長出翅膀的公司。這就是轉型在進行到一半時的樣子。在財務報表追上之前，變化就已經是真實發生的。

That awkwardness is part of process. Opendoor’s really starting to feel that way to me. Awkward, awkward, awkward, flight. We still have some awkward growing pains. But our wings are growing, and it really, really feels like this thing is taking off.

那種尷尬感是過程的一部分。Opendoor 對我來說真的開始有那種感覺了。尷尬、尷尬、尷尬，然後起飛。我們仍然有一些尷尬的成長陣痛。但我們的翅膀正在長出來，而且真的、真的感覺這件事正在起飛。

Speaking of things that still look a little awkward but are starting to grow wings, let’s talk about mortgages. Look, when you buy a home, you’re actually buying two separate things, the house and the money. Two different work streams, two different sets of people, two timelines that need to automagically merge, and 100 different ways that can kill the deal.

說到那些看起來仍有點尷尬、但開始長出翅膀的事情，我們來談談房貸。你看，當你買房時，你其實是在買兩個分開的東西：房子和錢。兩條不同的工作流程、兩組不同的人、兩條需要「自動魔法般」合併的時間線，以及 100 種可能讓交易告吹的方式。

We’re collapsing these two things into one integrated transaction. But why does this matter? It matters because friction. Friction destroys the process, and getting rid of it expands our margin, reduces risk, and builds a real flywheel between our buying engine and our selling engine. The best place to sell a home becomes the best place to buy one.

我們正在把這兩件事收斂成一筆整合式交易。但為什麼這很重要？因為摩擦。摩擦會摧毀流程；消除摩擦能擴大我們的利潤率、降低風險，並在我們的買方引擎與賣方引擎之間建立真正的飛輪效應。最適合賣房的地方，也會成為最適合買房的地方。

At our core, our job is simple: remove friction from homeownership process. And there are two kinds of friction in residential real estate. There’s a friction that holds people in place. Sellers are stuck. They’re stuck because of price uncertainty, repair headaches, and all the traditional pain that goes along with the timeline of selling a home. We’ve solved that first with our Offer product. Our core Offer product gives you near instant certainty, but unsticking the sellers is only half the trade.

在核心上，我們的工作很簡單：移除購屋流程中的摩擦。而住宅不動產有兩種摩擦。有一種摩擦會把人「卡」在原地。賣方被卡住了。他們被卡住，是因為價格不確定性、修繕的麻煩，以及所有伴隨賣房時間線而來的傳統痛點。我們先用 Offer 產品解決了第一種摩擦。我們的核心 Offer 產品能讓你幾乎即時獲得確定性，但讓賣方脫困只完成了交易的一半。

Once the buyer enters the picture, there’s a whole new type of friction. This is like the rate shock that acts as drag that’s already in motion. A 7% mortgage rate slows the deal down, or it kills it entirely. That’s exactly the friction that our mortgage product eliminates. Our core product allows sellers to move on. Our mortgage product allows buyers to move in. And this isn’t theoretical.

一旦買方進入畫面，就會出現全新類型的摩擦。這就像利率衝擊，成為已在運行中的阻力。7% 的房貸利率會讓交易變慢，甚至直接把它殺死。這正是我們房貸產品要消除的摩擦。我們的核心產品讓賣方能往前走。我們的房貸產品讓買方能搬進去。而這不是理論。

Look at the early numbers. In Colorado, our first launch market, more than half of our scheduled closes are going to be financed through Opendoor Home Loans. In Texas, just six weeks after launch, we’re already at nearly one in five, and that’s before rolling out FHA, VA, or adjustable rate products.

看看早期數據。在我們第一個上線的市場科羅拉多，超過一半的預定交割將透過 Opendoor Home Loans 融資。在德州，上線僅六週，我們就已接近五分之一，而且這還是在尚未推出 FHA、VA 或可調利率產品之前。

Look, to be clear, each state we launch will have its own dynamicsm right? Texas shows where a market can be in just six weeks, and Colorado is where a market can be with some seasoning. Neither of these is ceilings. These numbers are going to bounce around as we scale, and that’s normal. What matters is the direction and the underlying physics. There’s something big that I think people are missing here.

先說清楚，我們每個上線的州都會有各自的動態，對吧？德州顯示一個市場在短短六週內可以到達的狀態，而科羅拉多則是市場經過一段時間沉澱後可能呈現的樣子。這兩者都不是天花板。隨著我們擴張，這些數字會上下波動，這很正常。重要的是方向與底層的物理法則。我認為有件很大的事，大家正在忽略。

The frame of this has always kind of been wrong, right? This is not an adjacent service. This is not an attached play. If you’re evaluating the mortgage as an extra fee on the side, sure, it adds margin, but you’re missing the bigger picture.

大家一直以來的框架其實有點不對，對吧？這不是一個相鄰服務。這不是一個附加式打法。如果你把房貸當作旁邊額外收的一筆費用來評估，當然，它會增加利潤率，但你會錯過更大的圖景。

Mortgage is the other half of the coin. Our market maker needs both sides to clear. Our Offers create sellers; our mortgages create buyers.

房貸是硬幣的另一面。我們的做市機制需要兩邊都能成交。我們的 Offers 創造賣方；我們的房貸創造買方。

And you can look, ask a fair question. We tried mortgages before, and it didn’t work. Why will this time be different? I like this question that I just asked because it gives me a chance to be nerdy. You see, I love studying the history of companies, and I want to tell you a story that I tell our product managers. I call it good Sears and bad Sears.

你也可以問一個合理的問題。我們以前做過房貸，但沒有成功。為什麼這次會不一樣？我喜歡我剛剛自己提出的這個問題，因為它讓我有機會 nerdy 一下。你看，我很愛研究公司的歷史，我想講一個我會跟我們產品經理說的故事。我把它叫作「好的 Sears」和「壞的 Sears」。

For years, I’ve carried this metal card in my wallet, like actually my wallet. I love the history of money. I wrote a book about this. So it’s less odd for me to carry an old, unusable charge card than almost anyone else. But this is the Sears revolving charge card. This one was issued in 1951 to help fund customers buy stuff from Sears.

多年來，我一直在錢包裡放著一張金屬卡，真的就是我的錢包。我很喜歡貨幣的歷史。我還為此寫過一本書。所以，相較於幾乎任何人，我帶著一張老舊、不能用的記帳卡其實沒那麼奇怪。但這張是 Sears 的循環信用記帳卡。這張卡在 1951 年發行，用來協助資助顧客向 Sears 購買商品。

Sears eventually allowed other merchants to accept this card, and that business inside Sears went to become one of the most successful financial services products ever launched. And eventually, it became worth more than all of Sears itself. That business was eventually spun out, and we now call it Discover. That was good Sears, financial infrastructure born inside a transaction.

Sears 後來允許其他商家也接受這張卡，而 Sears 內部的這項業務，最終成為史上最成功的金融服務產品之一。而且最後，它的價值甚至超過整個 Sears 本身。這項業務後來被分拆出去，我們現在稱它為 Discover。那是「好的 Sears」：在交易之中誕生的金融基礎設施。

A few decades later, Sears bought Dean Witter, and they put it inside their department stores so that a mother of three who came in for a new fridge could also leave with a mutual fund. The idea was just absurd. It was two products sharing a roof but not a purpose. Sears had no competitive edge, nothing that would make this product special. A bunch of number crunchers tried to create ROI through attach. The whole thing failed. That was bad Sears.

幾十年後，Sears 收購了 Dean Witter，並把它放進百貨門市裡，讓一位帶著三個孩子、進來買新冰箱的媽媽，也能順便帶走一檔共同基金。這個想法簡直荒謬。那是兩個產品共用同一個屋簷，卻不共享同一個目的。Sears 沒有任何競爭優勢，沒有任何能讓這個產品變得特別的東西。一群只會算數的人試圖透過「附加」來創造 ROI。整件事失敗了。那是「壞的 Sears」。

Look, the main question is this. There is a difference between bolting something on and building something in. During my time at Shopify, we built Shopify Capital, the good Sears way. It didn’t work because we cross-sold merchants. It worked because the platform already saw every merchant in real time. We didn’t need a loan application. We already had the underlying data, right? Financial product wasn’t bolted on. It was born inside the transaction.

你看，主要問題是這個。把某個東西硬加上去，和把某個東西內建進去，是有差別的。我在 Shopify 的時候，我們用「好的 Sears」方式打造了 Shopify Capital。它之所以成功，不是因為我們向商家交叉銷售。而是因為平台本來就能即時看到每一位商家的狀況。我們不需要貸款申請。我們已經有底層數據了，對吧？金融產品不是被硬加上去的。它是在交易之中誕生的。

Opendoor 1.0’s mortgage, honestly, every mortgage product on the market is bad Sears. You buy a house, then someone awkwardly tries to sell you a loan. It just adds friction, and it fails because it deserves to fail. These products that fail aren’t about the customer. They’re about companies wanting margin. The product we’re building today, it sits inside the process from day one, built in, not bolted on. That changes both the customer experience and our unit economics. And here’s why.

老實說，Opendoor 1.0 的房貸——其實市面上每一種房貸產品都很糟，就像爛掉的 Sears。你買了房子，然後就有人很尷尬地想賣你一筆貸款。這只會增加摩擦，而且它失敗是因為它活該失敗。這些會失敗的產品並不是以客戶為中心。它們是為了公司想要賺取利差。我們今天正在打造的產品，從第一天起就嵌入流程之中，是內建的，不是事後硬加上去的。這同時改變了客戶體驗與我們的單位經濟效益。原因如下。

People think mortgages are special. They’re really, really not, right? The legacy mortgage industry carries 65 to 85 basis points of yield in every single loan just to feed the pork barrel buffet of people taking margin. The legacy mortgage industry pays thousands of dollars to acquire each buyer. And this chain exists only because it has always existed.

人們以為房貸很特別。其實一點也不特別，真的。傳統房貸產業在每一筆貸款裡都夾帶 65 到 85 個基點的收益率，只為了餵飽一整條靠抽利差吃飯的人。傳統房貸產業為了取得每一位買家，要付出數千美元的獲客成本。而這條鏈條之所以存在，只因為它一直以來都存在。

Look, we didn’t ask how we could make money on mortgage. We asked the following question, if our goal was to offer the lowest rate possible, what would be built? So we built our own point of sale and our own loan origination system. And we offer only the most common loans; no bells, no whistles. You can have any flavor of ice cream you want, so long as it’s vanilla.

你看，我們不是在問：要怎麼在房貸上賺錢？我們問的是：如果我們的目標是提供可能最低的利率，那應該打造什麼？所以我們打造了自己的銷售端系統（point of sale）以及自己的貸款起案系統（loan origination system）。而且我們只提供最常見的貸款；沒有花樣、沒有噱頭。你想要什麼口味的冰淇淋都可以——只要是香草口味。

If you want a fancy mortgage with lots of features, please use a bank. Do not use Opendoor Home Loans. But if you want a regular mortgage for a regular home, we are going to be your best bet. We have automated so much of the process that our loan officers can handle 50 mortgages a month against about 10 in the industry.

如果你想要一筆功能很多、很花俏的房貸，請去找銀行。不要使用 Opendoor Home Loans。但如果你想要的是一般住宅的一般房貸，我們會是你最好的選擇。我們把流程自動化到很高程度，因此我們的貸款專員每月可以處理 50 筆房貸，而業界大約是 10 筆。

Oh, and we’ve spent $0 on marketing. We took all the money we saved and passed them to home buyers. We’re not new at this, right? Home builders have figured this out decades ago. A great mortgage is a more powerful force than a price cut, and we have figured out the same thing.

另外，我們在行銷上花了 0 美元。我們把省下來的錢全部回饋給購屋者。我們不是新手，對吧？建商在幾十年前就想通這件事了。一筆很棒的房貸，比降價更有力量，而我們也得出了同樣的結論。

But because we turn inventory a few times a year, the math is even better for us. For a market maker, a lower mortgage rate is the thing that speeds everything else up. It removes friction and means that we get more turns a year. It frees up capital so we can buy the next house, and the next, and the next. We haven’t invented new math here. We have just bent it in a way that a standalone lender structurally can’t.

但因為我們一年能把庫存周轉好幾次，所以對我們來說數學更漂亮。對做市商而言，更低的房貸利率就是能讓其他一切加速的關鍵。它消除摩擦，意味著我們一年可以有更多次周轉。它釋放資本，讓我們能買下一間房、再下一間、再下一間。我們沒有發明新的數學。我們只是把它彎成一種獨立放貸機構在結構上做不到的方式。

Look, I’ve spent most of my career building financial services products. I’ve seen this work. Shop payments went from not existing to being the largest installment product on the internet in a year. Standalone lenders are constantly fighting gravity on customer acquisition. For us, the mortgage is just a second wing. Once both wings are locked in place, the physics switch from drag to lift.

你看，我職涯大部分時間都在打造金融服務產品。我看過這套方法奏效。Shop Payments 從不存在，到在一年內成為網路上最大的分期產品。獨立放貸機構在獲客上一直是在對抗地心引力。對我們而言，房貸只是第二個機翼。當兩個機翼都鎖定到位後，物理效應就會從阻力變成升力。

And mortgage wasn’t the only product that had lots of launches. We also deprecated five bespoke tools and multiple microservices and unified them all into Opendoor’s internal God view, Ops Hub. A single piece of software, that’s our command and control center, gives us end-to-end view of every home.

而房貸也不是唯一一個有大量上線的產品。我們也淘汰了五個客製化工具與多個微服務，並把它們全部整合到 Opendoor 內部的「上帝視角」——Ops Hub。這是一套單一軟體，作為我們的指揮與控制中心，讓我們能端到端掌握每一棟房屋的全貌。

We opened Opendoor up to agents. We launched Opendoor 2.0 for agents, with a new structure that puts our offers directly inside the tools agents already use, tools like RealScout, SOLD.com, and Movoto. And our new partner API launched, which means integrations into Opendoor now take days, not months. We rebuilt Opendoor iOS app right from the ground up. We held it to one bar, excellence against apps people use every day. Not good enough for a real estate app.

我們把 Opendoor 開放給房仲。我們為房仲推出 Opendoor 2.0，採用新的架構，將我們的報價直接嵌入房仲已在使用的工具中，例如 RealScout、SOLD.com 和 Movoto。我們也上線了新的合作夥伴 API，這代表與 Opendoor 的整合現在只需要幾天，而不是幾個月。我們從零開始徹底重建了 Opendoor iOS App。我們只用一個標準來要求它：要能與人們每天使用的 App 相比的卓越水準。而不是「對房地產 App 來說夠好」就算了。

We launched a unified dashboard that serves as a user’s home within Opendoor. Whether you’re selling to us, buying from us, or both, there’s now one place for your entire relationship with Opendoor.

我們推出了統一的儀表板，作為使用者在 Opendoor 裡的「家」。不論你是賣房給我們、向我們買房，或兩者皆是，現在都有一個地方可以管理你與 Opendoor 的整段關係。

We put machine learning to work on underwriting. Our auto underwriter now handles roughly one in five valuations, and we’re testing models that value entire segments of homes with almost no manual review at all.

我們把機器學習用在核保上。我們的自動核保系統現在大約處理五分之一的估值，我們也在測試能對整個房屋區段進行估值、幾乎不需要任何人工審查的模型。

We turned Chloe, our AI assistant, into a real part of our sales teams. She converts three times what she did in January at a third of the cost. She works 168 hours a week, takes no vacations or bathroom breaks, and is completely transparent about being a bot. But people love using her because she is good.

我們把 Chloe——我們的 AI 助理——變成銷售團隊真正的一部分。她的轉換率是 1 月的三倍，而成本只有三分之一。她每週工作 168 小時，不休假、不上廁所，而且完全透明地表明自己是機器人。但人們很愛用她，因為她真的做得很好。

We became faster under the hood. Seller dashboards now load three times faster, contract changes went from 4 hours to 10 minutes, and we retired legacy systems that had been slowing us down for a better part of a decade. We launched a new set of alerts for our buyers, including nearby listing matches, price drop notifications, and recommended homes. And that’s just a small sample of what we’ve shipped in the past quarters.

我們在底層也變得更快。賣方儀表板載入速度提升三倍，合約變更從 4 小時縮短到 10 分鐘，我們也汰換了那些拖慢我們將近十年的舊系統。我們為買方推出一套新的提醒功能，包括附近物件匹配、降價通知與推薦房源。而這只是我們過去幾個季度交付成果的一小部分。

Look, in my first call, I told you we would drive Opendoor to ANI profitability. I said drive because we didn’t know exactly how we would do it, and we didn’t really have any proof points. Today, we do. We’ve shown that if you ship great products every week and if you keep tilting the world toward homeowners, the score just kind of takes care of itself.

你看，在我第一次電話會議時，我告訴各位我們會推動 Opendoor 達到 ANI 獲利。我說「推動」，是因為我們當時並不確定要怎麼做到，也沒有真正的佐證。今天，我們有了。我們已經證明：只要每週持續推出優秀產品，並且不斷把世界往屋主那一邊傾斜，成績自然就會水到渠成。

With that, I’ll hand it over to Christine to tell you about the numbers. Christy?

接下來，我把時間交給 Christine，請她帶各位看數字。Christy？

Thank you, Kaz. Three things to know about Q2 before we get into the details. We grew homes acquired by 77% quarter over quarter and 149% year over year. We also closed the quarter with another 2,310 homes under contract. A year ago, that number was 393.

謝謝你，Kaz。在進入細節之前，先掌握第二季（Q2）三件事。我們的收購房屋數量季增 77%、年增 149%。本季結束時，另有 2,310 間房屋已簽約在案。一年前，這個數字是 393。

Contribution profit was $51 million, up 59% quarter over quarter and 22% more homes sold. Contribution margin was the highest we’ve reported in two years. A year ago, we purchased 1,757 homes and spent $15 million on operations expense. This quarter, we purchased 4,378 homes and spent $14 million, 2.5 times the volume on less variable cost.

貢獻毛利為 5,100 萬美元，季增 59%，且售出房屋數增加 22%。貢獻毛利率是我們兩年來報告過的最高水準。一年前，我們買進 1,757 間房屋，營運費用支出 1,500 萬美元。本季我們買進 4,378 間房屋，支出 1,400 萬美元——在變動成本更低的情況下，量能是 2.5 倍。

Volume, margin, operating leverage, these are the three management objectives on our path to profitability. The table in our earnings release shows where each one stands. Let’s walk through them now.

成交量、毛利率、營運槓桿，這三項是我們邁向獲利之路的管理目標。我們的財報新聞稿中的表格顯示了每一項目前的進展。我們現在就逐一說明。

First, scale acquisitions. We purchased 4,378 homes in the quarter and saw revenue growth of 23% quarter over quarter to $883 million. Acquisition contracts reached 6,908, up from 5,136 in Q1. Marketing spend moved in the opposite direction from $19 million to $5 million quarter over quarter.

第一，擴大收購規模。本季我們購買了 4,378 間房屋，營收較上季成長 23% 至 8.83 億美元。收購合約達到 6,908 份，高於第一季的 5,136 份。行銷支出則朝相反方向變動，較上季從 1,900 萬美元降至 500 萬美元。

Second, improve unit economics and resale velocity. Contribution margin has climbed from a low of 1% three years ago to 5.8% in Q2, landing in the middle of our 5% to 7% range we guided to. Aged inventory, which we define as homes listed for greater than 120 days, has fallen from 51% over that same span to 9% now.

第二，改善單位經濟效益與轉售速度。貢獻毛利率已從三年前 1% 的低點攀升至第二季的 5.8%，落在我們先前指引的 5% 至 7% 區間中段。逾期庫存（我們定義為掛牌超過 120 天的房屋）在同一期間已從 51% 降至目前的 9%。

Third, build operating leverage. Trailing 12-month operations expense was 1.6% of revenue, up from 1.3% in Q1. That said, operations expense actually declined from $52 million to $51 million. The increase in the ratio reflects last year’s higher revenue quarters rolling out of the trailing window, not any change in our cost discipline. On a per acquisition basis, operations expense was $3,000 this quarter, down from $5,000 in Q1 and $8,400 a year ago.

第三，建立營運槓桿。過去 12 個月的營運費用占營收比重為 1.6%，高於第一季的 1.3%。不過，營運費用金額實際上從 5,200 萬美元降至 5,100 萬美元。比率上升反映的是去年營收較高的季度逐步退出過去 12 個月的計算區間，而非我們成本紀律有任何改變。以每筆收購計算，本季營運費用為 3,000 美元，低於第一季的 5,000 美元以及一年前的 8,400 美元。

Fixed operating expenses were $35 million, up $2 million from Q1, funding our investments in AI and engineering.

固定營運費用為 3,500 萬美元，較第一季增加 200 萬美元，用於支持我們在 AI 與工程方面的投資。

Turning to the balance sheet. We ended the quarter with $896 million in cash and cash equivalents. We deliberately put capital to work rebuilding a larger, higher quality book, growing our inventory by more than $700 million during the second quarter, funded largely by our non-recourse asset-backed facilities. We held 5,459 homes in inventory at quarter end, with another 2,310 homes already under contract to purchase.

接著談資產負債表。我們在季末持有 8.96 億美元的現金及約當現金。我們刻意投入資本以重建更大、品質更高的資產組合，第二季庫存增加超過 7 億美元，主要由我們的無追索權資產擔保融資工具提供資金。季末我們持有 5,459 間房屋庫存，另有 2,310 間房屋已簽約待購。

One point on capital. As our Cash Now, More Later product scales, it provides a more efficient way to grow. It commits less capital per home than our core product, giving us a way to add acquisitions without growing our balance sheet at the same pace.

關於資本再補充一點。隨著我們的 Cash Now, More Later 產品規模擴大，它提供了更有效率的成長方式。相較於核心產品，它每間房屋所需投入的資本更少，讓我們能在不以同樣速度擴張資產負債表的情況下增加收購量。

Now the guidepost for Q3. Acquisitions. You can continue to track our contracts on accountable.opendoor.com. Revenue. We expect revenue to increase at least 20% year over year. Contribution profit. We expect contribution profit dollars to more than double year over year and contribution margin to be around 4% to 4.5%, reflecting typical seasonal trends.

接著是第三季的指引重點。收購方面：你們可以持續在 accountable.opendoor.com 追蹤我們的合約數。營收方面：我們預期營收年增至少 20%。貢獻利潤方面：我們預期貢獻利潤金額將較去年同期增加一倍以上，貢獻毛利率約為 4% 至 4.5%，反映典型的季節性趨勢。

Contribution margin has fallen from Q2 to Q3 in every year we have been public by an average of 470 basis points, excluding 2023, when 2Q CM was negative. We expect a decline well below that average and the best Q2 to Q3 relative performance we have delivered outside of 2023.

自我們上市以來，每一年貢獻毛利率都會從第二季降到第三季，平均下降 470 個基點（不含 2023 年，因為 2023 年第二季的貢獻毛利率為負）。我們預期本次降幅將明顯低於該平均值，並將創下除 2023 年以外我們所交出的最佳第二季到第三季相對表現。

Adjusted EBITDA. We expect to be adjusted EBITDA profitable on a 12-month go-forward basis as of the second quarter of 2026. In Q3, we expect a modest increase in adjusted operating expense, driven mostly by the holding costs that come with growing our inventory and a small increase in marketing.

調整後 EBITDA。我們預期自 2026 年第二季起，以未來 12 個月滾動基礎衡量，將達到調整後 EBITDA 獲利。在第三季，我們預期調整後營運費用將小幅增加，主要由庫存成長所伴隨的持有成本，以及行銷的小幅增加所帶動。

Our commitment has not changed. We expect to be adjusted net income positive by the end of this year, measured on a 12-month go-forward basis. A quarter ago, we told you exactly what this quarter would look like. And despite a real estate market that remains challenged, we did what we said we would do.

我們的承諾沒有改變。我們預期在今年年底前，以未來 12 個月滾動基礎衡量，調整後淨利將轉為正值。一季前，我們就已精準告訴各位本季會是什麼樣子。即便房地產市場仍具挑戰，我們也做到了我們說過會做的事。

With that, Michael, I’ll turn it over to you for questions.

接下來，Michael，我把時間交給你進行提問。

Thanks, Christy. Our first question comes to us via video submission from Lorraine Wang.

謝謝你，Christy。我們的第一個問題是透過影片投稿，由 Lorraine Wang 提出。

(video starts)

（影片開始）

Hi, everyone. My name is Lorraine Wang, a long-term Opendoor shareholder. Thank you for taking my question.

大家好。我叫 Lorraine Wang，是 Opendoor 的長期股東。謝謝你們讓我提問。

When long-term shareholders look back two years from now, what will be the one or two changes inside the business you hope they will point to and say, that’s when Opendoor truly became a different company. Thank you.

當長期股東在兩年後回顧時，你希望他們會指出公司內部哪一到兩項改變，並說：那就是 Opendoor 真正成為一家不同公司的時刻。謝謝。

(video ends)

（影片結束）

Thanks, man. Well, thanks for being a long-term shareholder. People like you are the reason this company survived so that we could get a chance to turn around. I generally think of our everyday shareholders as my boss. So I’m going to give you the same answer I would give if you were my boss.

謝謝，夥伴。也謝謝你作為長期股東。像你這樣的人，正是這家公司能撐過來、讓我們有機會扭轉局面的原因。我通常把我們日常的股東視為我的老闆。所以我會給你同樣的答案，就像你真的是我老闆一樣。

I won’t give you a number because it won’t be a number or a feature we launch. It just won’t. It’ll be this. We stopped being a real estate company with a website and became a product company that happens to be in real estate. Let me tell you what I mean.

我不會給你一個數字，因為那不會是一個數字或我們推出的某個功能。真的不會。會是這件事。我們不再是一家有網站的房地產公司，而是成為一家剛好身處房地產領域的產品公司。我來解釋我的意思。

For most of this company’s life, it seems like every decision we made got filtered through a spreadsheet. Mortgage rates, HPA, spreads, conversion, unit economics. I look at these things every single day. They are very real, but that’s the business. It’s not the company. The company is actually the product and the customers.

在這家公司大部分的歷史中，感覺我們做的每個決策都要先經過試算表的篩選。房貸利率、房價升幅（HPA）、利差、轉換率、單位經濟效益。我每天都在看這些。它們非常真實，但那是「生意」。那不是「公司」。公司其實是產品與客戶。

When you run a company like a financial portfolio, every individual trade, it makes sense on paper, but the end-to-end user experience, it just becomes completely broken. You end up with something that is barely functional but a pain to use. You end up with a bank.

當你用管理金融投資組合的方式經營公司，把每一筆交易都拆開來看，紙面上都說得通，但端到端的使用者體驗就會徹底崩壞。最後你會得到一個勉強能用、但用起來很痛苦的東西。你會得到一家銀行。

Look, you don’t build for a future that doesn’t yet exist in spreadsheets. So the shift is this. Our primary goal, the reason we come to work every day, is to build something people want. Our unit economics are downstream from that, right? When people use our products, when we eliminate friction, when we help our users, conversion goes up, margin goes up, and everything follows.

你看，你不會為了試算表裡尚不存在的未來去打造產品。所以轉變是這樣的。我們的首要目標、我們每天來上班的原因，是打造人們想要的東西。我們的單位經濟效益是由此而來的下游結果，對吧？當人們使用我們的產品、當我們消除摩擦、當我們幫助使用者，轉換率會上升、毛利率會上升，然後一切都會隨之而來。

Our product is better today, but it’s still not where it should be, right? This is the largest transaction of most folks’ life, and for most of them, it’s also the worst. There’s like no reason why both those facts have to be true.

我們今天的產品更好了，但仍然還沒到它應該達到的水準，對吧？這是大多數人一生中最大的一筆交易，而對多數人來說，它也同時是最糟的一次體驗。這兩件事沒有理由一定要同時成立。

Two years from now, I hope to point back to this. We decided to build an excellent product. We took the parts of the product that were bad every single day, and we fixed them. Then you’ll wake up one day, and you’ll see that all the work we have done to fix our infrastructure, to build a better product, to build a mortgage system that is just awesome, will merge together, and it’ll make the process of buying and selling a home as easy as using Shop Pay.

兩年後，我希望能回頭指向這件事。我們決定打造一個卓越的產品。我們把產品中每天都很糟的部分找出來，並把它們修好。然後有一天你醒來會發現，我們為了修復基礎設施、打造更好產品、打造一套非常棒的房貸系統所做的所有工作，會融合在一起，讓買賣房屋的流程像使用 Shop Pay 一樣簡單。

All of it will have started on this. We build products, not spreadsheets.

這一切都將從這件事開始。我們打造的是產品，不是試算表。

Matthew S. asks, are we still on track for profitability by the end of this year?

Matthew S. 問：我們是否仍按計畫在今年年底前達成獲利？

Yes, yes, yes, yes. Look, every single quarter, we’ve done what we said we would do. We have come here, we’ve said what we’re going to do, and then we’ve done it. We’re going to become ANI profitable on a 12-month go-forward basis at the end of this year.

是的，是的，是的，是的。你看，每一個季度，我們都做到了我們說過要做的事。我們來到這裡，我們說我們要做什麼，然後我們就做到了。我們將在今年年底，以未來12個月滾動基礎達成 ANI 獲利。

I told you we were going to do it on our first earnings calls. I sat here, and I said it. We’re going to become ANI profitable on a 12-month go-forward basis at the end of this year. And we’re now giving you the math to back this up. The answer is yes. Was that clear?

我在第一次財報電話會議就告訴過你們我們會做到。我坐在這裡，我說過。我們將在今年年底，以未來12個月滾動基礎達成 ANI 獲利。而我們現在也把數學算給你們看來支撐這點。答案是肯定的。這樣清楚嗎？

Clear to me. Angelo E. asks, you have described letting buyers and sellers transact directly on Opendoor without Opendoor owning the home. When could this launch? And what revenue and contribution profit per transaction are achievable?

對我來說很清楚。Angelo E. 問：你們描述過讓買方與賣方在 Opendoor 上直接交易，而不需要 Opendoor 持有房屋。這何時可能上線？以及每筆交易可達成的營收與貢獻利潤是多少？

Angelo, thanks for the question, man. Let me address the economics of this upfront by saying, I’m not going to talk about revenue or contribution profit on a product that we haven’t launched yet. This is the same answer I gave to Lorraine.

Angelo，謝謝你的問題，老兄。我先從經濟性角度說明：對於我們尚未推出的產品，我不會談營收或貢獻利潤。這和我給 Lorraine 的答案一樣。

Great products don’t start with a margin target. We build for the mission,and we build for the user, and economics follow, right? Our mission is to tilt the world in favor of homeowners.

偉大的產品不是從毛利目標開始的。我們為使命而打造，也為使用者而打造，然後經濟效益自然會跟上，對吧？我們的使命是讓世界更偏向房主。

Today, Americans pay 10% to 12% transacting on the biggest asset they own, and the process sucks. That’s the problem that we need to fix. And we’ll get there in three steps, right? I come from the e-commerce world, so I like using that language. I say 1P, 2P, 3P.

今天，美國人在交易自己擁有的最大資產時，要支付 10% 到 12% 的成本，而且流程很糟。這就是我們需要解決的問題。而我們會分三步走到那裡，對吧？我來自電商世界，所以我喜歡用那套語言。我說 1P、2P、3P。

Step one, 1P. We buy the homes ourselves, and we sell them. This is our regular core cash offer product. Step two, 2P. This is Cash Now, More Later. We still buy your home, but it’s a different way of owning it. It’s capital-light for us with less risk, with alignment with the seller.

第一步，1P。我們自己買房，然後再賣出。這就是我們常規的核心現金報價產品。第二步，2P。這是「先拿現金，之後再拿更多」（Cash Now, More Later）。我們仍然會買你的房子，但持有方式不同。對我們而言更輕資本、風險更低，並且與賣方利益一致。

Step three is 3P. That’s what you’re describing. Buyers and sellers transacting directly with each other using all of our services and tools without using any of our balance sheet. Your home, your transaction, our platform. We’re on step two right now, we’re not going to move to step three until we are very good at step two.

第三步是 3P。這就是你所描述的。買方與賣方彼此直接交易，使用我們所有的服務與工具，但不動用我們的資產負債表。你的房子、你的交易、我們的平台。我們現在在第二步；在我們把第二步做得非常好之前，我們不會進入第三步。

But look, for what it’s worth, I think people overestimate how hard step three is and underestimate how important steps one and two are on the way to step three. Just look at Amazon. What did Amazon do? First, they sold their own stuff in their own warehouses, 1P, right? They sold the things that they sold, but it shipped from other people’s warehouses, 2P. Then other people’s stuff out of other people’s warehouses, 3P. Amazon launched other people’s stuff from Amazon’s warehouses using Amazon Pay. 3P with attached services offered by Amazon.

但你看，說句公道話，我認為人們高估了第三步有多難，卻低估了第一步和第二步在通往第三步路上的重要性。看看 Amazon 就知道了。Amazon 做了什麼？首先，他們在自己的倉庫賣自己的商品，1P，對吧？接著，他們賣他們要賣的商品，但由別人的倉庫出貨，2P。然後是別人的商品、從別人的倉庫出貨，3P。Amazon 也從 Amazon 的倉庫上架別人的商品，並使用 Amazon Pay。也就是 3P，再加上 Amazon 提供的附加服務。

Those last two steps like launched the fastest. Because when you are excellent at the first two, the other two aren’t that much work. And I have some conviction here because I came from the place where the entire business was 3P, letting people transact without owning the stuff. It works really, really, really well, but only once you’ve done the work required to earn it.

最後那兩步推出得最快。因為當你把前兩步做到極致時，後兩步其實不需要那麼多工作。我對此很有把握，因為我來自一個整個業務都是 3P 的地方，讓人們在不持有商品的情況下完成交易。它運作得非常、非常、非常好，但前提是你先完成了必須做的工作，才有資格做到那一步。

Great. Felix B., hey, Kaz, could you update on mortgage product? And how many states are using Opendoor Mortgage?

很好。Felix B.，嗨 Kaz，你能更新一下房貸產品的進展嗎？以及目前有多少州在使用 Opendoor Mortgage？

Yeah. Thanks, Felix. I spoke about this a bit already, and I want to be a little careful because our lawyers are going to give me notes after this. But look, the mortgage licensing process seems to have been designed by the same people who designed the DMV. It’s not built for speed, and it’s infuriatingly slow on purpose. It’s state by state slog, we’re working our way through it. And we’re doing faster than I think anyone expected. Genuinely faster than people thought we could do it, but slower than I want. So we expect to be licensed in around 35 to 40 states by the end of this year.

可以。謝謝你，Felix。我之前已經稍微談過一點，我得稍微謹慎些，因為我們的律師會在這之後給我意見。但你看，房貸牌照的流程看起來像是由設計 DMV 的同一群人設計的。它不是為速度而設計的，而且是刻意慢到令人抓狂。這是一個逐州推進的苦戰，我們正在一路處理。而且我們的速度比我想像中任何人預期的都快。真的比大家以為我們能做到的更快，但比我想要的更慢。因此，我們預期在今年年底前，會在大約 35 到 40 個州取得牌照。

But also like -- we’ll get there on this front. It’s like we know we will. So the real question is, how are we doing in states where we are now live? And the answer is we’re doing just excellent. Buyers are getting amazing mortgage rates, very fast, very little pain, end to end. And if they want it, they don’t have to talk to a single human being.

不過也像是——我們會在這方面達成的。就像我們知道我們一定會做到。所以真正的問題是：在我們目前已上線的州，我們表現如何？答案是：我們做得非常出色。買家拿到很棒的房貸利率、速度非常快、痛點很少、端到端完成。而且如果他們願意，他們甚至不必和任何一個真人交談。

Look, I shared the numbers earlier. It’s over half of our scheduled closes in Colorado are going to be on Opendoor Home Loans, nearly one in five in Texas. That’s only six weeks after launch, and this is a completely unoptimized product right now. We don’t offer FHA, which we will. We don’t offer VA, which we will. We don’t offer ARM, which we will, right?

你看，我之前分享過數字。在科羅拉多州，我們排程中的成交有一半以上會使用 Opendoor Home Loans；在德州則接近五分之一。這只是上線後六週，而且目前這個產品完全還沒做最佳化。我們還沒提供 FHA，但我們會提供。我們還沒提供 VA，但我們會提供。我們還沒提供 ARM，但我們會提供，對吧？

But the reason this is working is simple. There’s a structural unfairness in the American home buying process. There’s a pork barrel buffet of margin that takes money out of the hand of home buyers. It’s 65 to 85 basis points baked into every single loan. We built a mortgage product without any of these costs, and our job is to pass the savings on.

但之所以能奏效，原因很簡單。美國的購屋流程存在結構性的「不公平」。有一大桌充滿利益輸送的毛利自助餐，把錢從購屋者手中拿走。每一筆貸款都內含 65 到 85 個基點的成本。我們打造了一個不含這些成本的房貸產品，而我們的工作就是把節省下來的費用回饋給客戶。

When a default option is the cheapest and the easiest option, it just works. Now look, in the spirit of being very fully transparent, if you’re buying a $10 million penthouse in Manhattan, I would like you to use someone else. Do not use us. We will do just vanilla. But I’m very confident we’re going to do vanilla better than anyone else. It’s going to be just excellent.

當預設選項同時是最便宜、也最容易的選項時，它自然就會運作。現在你看，本著完全透明的精神，如果你是在曼哈頓買一間 1,000 萬美元的頂樓豪宅，我希望你去用別家。不要用我們。我們只做最基本的標準款。但我非常有信心，我們會把標準款做得比任何人都好。它會非常出色。

Angelo E. He’s curious, what could keep Opendoor below 2% of US home sales? Is it seller adoption, home eligibility, offer competitiveness, or buyer’s willingness to transact through Opendoor? What must change to exceed 6%? That good?

Angelo E. 他很好奇，是什麼可能讓 Opendoor 的市占率一直低於美國住宅銷售的 2%？是賣方採用度、房屋符合資格的範圍、出價競爭力，還是買方願意透過 Opendoor 交易的意願？要超過 6% 必須改變什麼？就這樣？

Man. Angelo, that’s two good questions in one quarter. It’s the same Angelo, I think, right?

天啊。Angelo，你在一個季度裡一次問了兩個好問題。我想是同一個 Angelo，對吧？

Okay. What can keep us below 2%? Honestly, these are things I think about a lot. Like every morning, I look at our share as a percentage of all US, but a lot of it’s what we can control. Let’s talk about 2% first, okay?

好。什麼會讓我們低於 2%？老實說，這些是我經常在想的事。像是每天早上，我都會看我們占全美的市占率百分比，但其中很多其實是我們能控制的。我們先談 2%，好嗎？

Take offer competitiveness. There’s two things that matter on this, right? How accurately we price the home and how good we are at our operations. So in the past, our offers were bad because we were bad at both of these things, and we just used spread to cover our asses. This year, we’ve gotten much better. We’re faster and more accurate with much more property level dispersion. The price of each individual home reflects the merit of the home and its portfolio risk to us, not some random market average.

先看出價競爭力。這裡有兩件事很重要，對吧？我們對房屋定價有多準，以及我們的營運做得有多好。所以過去我們的出價很糟，因為這兩件事我們都做得很差，然後我們就用價差來替自己遮羞。今年我們進步很多。我們更快、更準，而且在更細的物件層級上做出更大的差異化。每一間房子的價格反映的是房子本身的價值，以及它對我們投資組合的風險，而不是某個隨機的市場平均值。

And as a result, our offers are just genuinely better, and we’re converting better at the same spreads as we have at any point in our history. So seller adoption isn’t really the constraint. And the demand’s always been there. It’s just that our offers have sucked, and we couldn’t convert. We fixed that.

因此，我們的出價確實更好，而且在與我們歷史上任何時點相同的價差水準下，我們的轉換率更高。所以賣方採用度其實不是限制因素。而需求一直都在。只是我們的出價以前很爛，導致我們無法轉換。我們把這點修好了。

Home eligibility is another thing we fixed. Our buy box basically covers the entire lower 48 now, well over 90%. So that’s not the constraint. And buyer willingness is really just a function of us doing our job and not screwing up. So I don’t see 2% as a natural ceiling. I really don’t. And on our current trajectory, we’ll cross it, and if we don’t, it’ll be because we screwed something up.

房屋符合資格的範圍也是我們修好的另一件事。我們的收購範圍（buy box）現在基本上涵蓋全美本土 48 州，覆蓋率遠超過 90%。所以那也不是限制。而買方意願其實主要取決於我們把工作做好、不要搞砸。所以我不認為 2% 是自然的天花板。我真的不這麼看。以我們目前的軌跡，我們會跨過它；如果沒有，那就是我們把某些事情搞砸了。

But to get above 6%, I think two things need to happen. I don’t think I’ve talked about this before, but transparently, it’s obvious. 3P needs to be working. Cash offers are generally constrained by our risk appetite and our balance sheet, right? A marketplace isn’t.

但要超過 6%，我認為需要發生兩件事。我想我以前沒談過這個，但坦白說，這很明顯。第三方（3P）必須運作起來。現金收購通常會受到我們的風險偏好與資產負債表所限制，對吧？但市場平台不會。

The second thing that needs to happen is more of the transaction needs to happen inside Opendoor. Let me give you a real example, like insurance. Today, our closings get delayed constantly because the buyer’s insurance isn’t ready, and we can’t fund a mortgage. So that’s the step we don’t control yet. But I’m relatively confident that the same logic that applied for us to mortgage will apply to insurance. We’re not going to capture a key. We’re just going to make the friction disappear, and we’ll be very, very good.

第二件需要發生的事，是更多交易流程必須在 Opendoor 內部完成。我舉個真實例子，例如保險。今天，我們的交割經常延遲，因為買方的保險還沒準備好，而我們就無法撥款房貸。所以那是我們目前還無法掌控的一步。但我相當有信心，當初適用於房貸的同樣邏輯，也會適用於保險。我們不會去搶一把鑰匙。我們只是讓摩擦消失，而我們會做得非常、非常好。

Dae Lee, JPMorgan. Do you have any early thoughts on 2027?

Dae Lee，摩根大通。你對 2027 年有任何初步想法嗎？

Yeah. I’m incredibly bullish on 2027. I don’t want to give guidance for 2027 out, but we’re spending a lot of time doing very, very difficult work, setting ourselves up so we can have an excellent 2027.

有。我對 2027 年非常看多。我不想對 2027 年提供指引，但我們投入大量時間在做非常、非常困難的工作，為了讓我們能在 2027 年交出非常出色的表現而做準備。

I generally think we’re going to surprise people by how much of the US housing market will flow through Opendoor next year. Look, it’s very obvious we’re building a car that’s designed to go 200 miles an hour, and right now, we’re kind of testing it at 30 miles an hour, so we know it won’t fall apart as it turns. But this thing really, really wants to go faster.

我整體上認為，明年有多少美國房市交易會流經 Opendoor，我們會在這方面讓大家感到驚訝。你看，很明顯我們正在打造一台設計時速 200 英里的車，而現在我們有點像是在用時速 30 英里測試它，確保它在轉彎時不會散架。但這台車真的、真的很想跑得更快。

Another question from Dae. How is AI specifically changing your ability to scale acquisition volume and improve margins? And what changes are you expecting now that a Chief AI officer has joined the team?

Dae 的另一個問題。AI 具體如何改變你們擴大收購量規模、並提升利潤率的能力？以及在首席 AI 官加入團隊後，你們預期會有哪些改變？

Do you want to take this?

你想回答這題嗎？

I am happy to take this because I personally love the leverage that you get from AI and the environment we’ve created here at Opendoor to encourage all of us to use it. Last week, while I was in a meeting, I had an agent running a contribution margin analysis and another agent checking a tax filing that a colleague had prepared. So I literally was like in three places at once.

我很樂意回答，因為我個人非常喜歡 AI 帶來的槓桿效應，以及我們在 Opendoor 這裡打造的環境，鼓勵大家使用它。上週我在開會時，我讓一個代理（agent）在跑貢獻毛利分析，另一個代理在檢查同事準備的一份報稅申報文件。所以我真的就像同時在三個地方。

And I think that kind of applies and extrapolates to the whole organization. If you walk around our offices, it looks more like a tech lab than a real estate company, right? You see people with Terminal up, Cursor up, Claude Desktop up, doing all sorts of things. People running into meetings with their laptops half open because they don’t want to accidentally disrupt AI from its work in progress.

我覺得這也適用並可外推到整個組織。如果你在我們辦公室走一圈，看起來更像科技實驗室，而不是房地產公司，對吧？你會看到有人開著 Terminal、開著 Cursor、開著 Claude Desktop，在做各式各樣的事情。大家抱著筆電半開著衝進會議室，因為他們不想不小心打斷 AI 正在進行中的工作。

Kaz walked through a few examples of how we’re actually seeing this leverage in play. You have Home Project Managers that used to manage 3 renos at a time that are now managing 10, and by the end of the year, we expect that to be 20. We used to, 80% of our homes, you had to have someone in your house before you could actually receive an offer. We have that down to 20%, aiming to be at 10% by the end of the year.

Kaz 提了幾個例子，說明我們如何真正看到這種槓桿效應在發揮。以前一次只能管理 3 個翻修案的房屋專案經理（Home Project Managers），現在能同時管理 10 個；而到今年年底，我們預期會到 20 個。過去我們有 80% 的房子，在你實際收到報價前，必須先有人到你家看過。我們已把這比例降到 20%，目標是在年底降到 10%。

Our Chief AI Officer was a fantastic addition. And his primary objective aligns with our third management objective, which is to build operating leverage so that our costs don’t scale linearly with acquisitions. That means making sure our AI spend is efficient and productive.

我們的首席 AI 官是一個非常棒的補強。他的主要目標與我們的第三項管理目標一致，也就是建立營運槓桿，讓我們的成本不會隨著收購量線性成長。這代表要確保我們在 AI 上的支出是有效率且有產出的。

Performance and cost vary a lot depending on which model you use. So we route work to make sure the right model handles the right task instead of everyone defaulting to the last model they used. He is focused on consolidating tooling, building with fewer vendors to focus our tech stack and using scale to negotiate better terms with vendors.

不同模型的表現與成本差異很大。所以我們會做工作路由，確保正確的模型處理正確的任務，而不是讓每個人都預設用他們上一次用的那個模型。他專注於整合工具、用更少的供應商來建構，以聚焦我們的技術堆疊，並利用規模與供應商談出更好的條款。

And then education and strategy. Working with our teams to come up with elegant solutions to really challenging and unique problems and giving each team the tools and training they need to operate at the level of our best individual users and engineers. Thank you for the question, Dae.

然後是教育與策略。與各團隊合作，針對真正具挑戰性且獨特的問題提出精巧的解法，並提供每個團隊所需的工具與訓練，讓他們能以我們最優秀的個別使用者與工程師的水準來運作。謝謝你的提問，Dae。

Awesome. Andrew, Citizens. Could you talk about your progress with adjacent monetization? I want to know mortgage specifically, but can you grade your progress on product attach broadly and where you see a positive trajectory for adjacent revenue and gross profit to improve overall unit economics for transactions?

太棒了。Andrew，Citizens。你能談談你們在相鄰變現（adjacent monetization）方面的進展嗎？我特別想了解房貸，但也請你更廣泛地就產品附加（product attach）的進度給個評等，以及你在哪些方面看到相鄰收入與毛利的正向趨勢，能改善交易的整體單位經濟效益？

This is a great question. Okay, well, this isn’t Harvard, so I won’t grade inflate. I’ll just tell you the grade, I would tell the PM leading this stuff in their performance review. Fair? Okay.

這是個很棒的問題。好，那個，這裡不是哈佛，所以我不會灌水給分。我就直接告訴你分數——也就是我會在帶領這些工作的產品經理（PM）績效評核裡給的分數。可以嗎？好。

On mortgage, I’d give us a solid B-minus. But what we’ve built is insanely hard to imagine building, and we’ve done it in a hard environment just obscenely quickly, and the product is just excellent. But you don’t get participation prizes. We’ve done all the right stuff to set ourselves up, but we’re now going to go out and win. Good early results, but we have work to do.

在房貸方面，我會給我們一個紮實的 B-。但我們打造的東西，難度高到幾乎難以想像能做出來；而且在艱難的環境下，我們以快得離譜的速度完成，產品也真的非常出色。但不會因為努力就有參與獎。我們做了所有正確的事來把基礎打好，接下來我們要出去贏。早期結果不錯，但還有工作要做。

On title and escrow, I’d give us a B-plus. We’re almost certainly -- I’m like very certain about this. We have the best title and escrow product in the US. It’s not even close. No company could acquire as many homes as we do and have capacity left over. I think we can just do an order of magnitude more transactions with our existing capacity. It’s not an A yet because we aren’t yet doing a majority of title and escrow transactions in the US, so B-plus.

在產權與代管（title and escrow）方面，我會給我們 B+。我們幾乎可以確定——我對這點非常有把握。我們擁有全美最好的產權與代管產品。差距大到根本不用比。沒有任何公司能像我們一樣收購這麼多房子，還能有剩餘產能。我認為以我們現有產能，就能把交易量再做大一個數量級。還不到 A 的原因是，我們目前還沒有承作美國大多數的產權與代管交易，所以是 B+。

On insurance, I’d give us an incomplete. We basically dropped out after a semester so we could study for mortgage. We’ll pick this up next year. So I think things are going better than anyone could reasonably have expected they would go. They’re just generally going excellently.

在保險方面，我會給我們「未完成」（incomplete）。我們基本上上了一學期就先退選，好把時間拿去準備房貸。我們明年會再把這塊撿回來。所以我認為整體進展比任何人合理預期的都要好。總體來說就是進展非常順利。

Good report card. We’re coming up on time, so I think this will be the last one we’ll have. Also from Andrew at Citizens.

很好的成績單。時間差不多了，所以我想這會是我們最後一題。同樣來自 Citizens 的 Andrew。

From X/Twitter, it feels like Kaz is doing a good job of recruiting and bringing in high-level talent. Can you talk about the investment intensity and expectations that the investment community should have around the cost structure of Opendoor as the team appears to still be being built out?

從 X/Twitter 上看起來，Kaz 在招募並引進高階人才方面做得很好。隨著團隊似乎仍在擴建中，你能談談投資強度，以及投資社群對 Opendoor 成本結構應有的預期嗎？

Thank you for the question, Andrew, and thank you for the kind words on our team. It is truly incredible the talent that we’ve been able to bring in. And they say A players attract A talent. The talent just keeps compounding. These are people who are uniquely skilled in their craft, like Vu and Morgan. They’re deeply passionate about our mission and the problems that we’re solving here.

謝謝你的提問，Andrew，也謝謝你對我們團隊的肯定。我們能延攬到的人才水準真的令人驚嘆。大家常說 A 級選手會吸引 A 級人才。人才就這樣持續複利式累積。這些人對自己的專業有獨到的能力，例如 Vu 和 Morgan。他們也對我們的使命，以及我們在這裡要解決的問題充滿熱情。

Here’s how I’d frame the investment. I talked about in the prepared remarks that fixed operating expenses increased from $33 million to $35 million in the quarter, and that is primarily driven from our investments in talent and engineering and AI. But the way to think about it is that we’re spending small, controlled amounts now to build software that scales instead of spending large amounts later, ramping up headcount to address capacity constraints.

我會這樣來框定這項投資。我在事先準備的發言中提到，本季固定營運費用從 3,300 萬美元增加到 3,500 萬美元，主要是由於我們在人才、工程與 AI 上的投資。但重點在於：我們現在以小而可控的投入來打造可擴展的軟體，而不是之後再花大錢、透過大幅增加人力來解決產能瓶頸。

We’re streamlining and consolidating our SaaS tools and putting that spend into engineers and AI. Less time and money spent stitching together someone else’s software and more time spent building our own, which we can shape to solve the challenges that are unique to our business. For example, turning on thousands of utilities, turning them off and on every single month as we cycle through homes.

我們正在精簡並整併我們的 SaaS 工具，把那部分支出轉投到工程師與 AI 上。少花時間和金錢去把別人的軟體拼接起來，多花時間打造我們自己的系統，並且能依照我們業務獨特的挑戰來塑形。例如：為數千個物件開通公用事業服務，並在我們房屋周轉的過程中，每個月反覆關閉與重新開通。

There’s a cost tailwind underneath all of this. The cost of AI capability itself keeps getting more efficient. What used to take a large model and real spend now runs cheaper and often faster. So the same investment deployed effectively can buy more each quarter.

在這一切之下其實有一股成本順風。AI 能力本身的成本持續變得更有效率。過去需要大型模型與真金白銀的投入，現在運行成本更低，而且往往更快。因此，同樣的投資若能有效部署，每一季都能買到更多產出。

The return on our investment will come through in both fixed and variable OpEx as we scale. So the last time acquisition contracts were above 6,000 in a quarter, fixed OpEx was double what it is today. And as we highlighted on the call, acquisition cost -- sorry, acquisition closes more than doubled year over year, and we spent $1 million less in variable operations. These achievements are a reflection of our technical investments.

我們投資的回報，會在規模擴張時同時體現在固定與變動營運費用（OpEx）上。上一次我們單季收購簽約量超過 6,000 件時，固定 OpEx 是現在的兩倍。而且如同我們在電話會議中強調的，收購結案——抱歉，是收購成交結案量——年增超過一倍，但我們在變動營運上反而少花了 100 萬美元。這些成果反映了我們在技術上的投資。

So the expectation I’d leave you with, fixed OpEx may keep growing modestly in dollar terms as we invest, but the discipline we’re holding ourselves to is that fixed OpEx stays relatively constrained and variable costs grow slower than our volumes do.

所以我想留給各位的預期是：隨著我們持續投資，固定 OpEx 可能在金額上仍會小幅成長；但我們自我要求的紀律是，固定 OpEx 維持相對受控，而變動成本的成長速度要低於我們的交易量成長。

Hey, can I end this?

嘿，我可以做個結尾嗎？

Go for it.

請便。

Thanks, folks, for joining us for our financial open-house. Our job here is simple. Our job is to come here and tell you we did what we said we would do. We’ve now done that three quarters in a row. And we’ll come back next time and tell you again we did what we said we would do on our way to tilting the world toward homeowners.

各位，感謝參加我們的財務開放日。我們在這裡的工作很簡單。我們的工作就是來這裡告訴你們：我們做到了我們說要做的事。我們現在已經連續三個季度做到了。下次我們會再回來，並再次告訴你們：在讓世界更傾向於屋主的道路上，我們又做到了我們說要做的事。

With that, we’ll see you next time. Cheers.

那麼，我們下次見。乾杯。