Amplitude, Inc. (AMPL) 2026 Q1 法說會逐字稿

(audio in progress) -- additional information regarding these non-GAAP financial measures and a reconciliation between these GAAP and non-GAAP financial measures are included in our earnings press release and the supplemental financial information which can be found on our Investor Relations website at investors.Amplitude.com/.

（音訊進行中）——關於這些非 GAAP 財務衡量指標的其他資訊，以及這些 GAAP 與非 GAAP 財務衡量指標之間的調節表，已包含於我們的財報新聞稿與補充財務資訊中，可於我們的投資人關係網站 investors.Amplitude.com/ 查閱。

With that, I hand over the call over to Spenser.

接下來，我把電話會議交給 Spenser。

Good afternoon, everyone, and welcome to Amplitude's first quarter 2026 earnings call.

各位下午好，歡迎參加 Amplitude 2026 會計年度第一季財報電話會議。

Today, I'll cover three things. First, our Q1 results. Second, how AI is reshaping the software development lifecycle. And third, a deep dive into our latest AI products and the customers putting them to work.

今天我將涵蓋三件事。第一，我們第一季的業績。第二，AI 如何重塑軟體開發生命週期。第三，深入介紹我們最新的 AI 產品，以及正在實際導入使用的客戶。

Let me start with the numbers.

我先從數字開始。

Q1 revenue was $94 million, up 17% year-over-year. Annual recurring revenue was $374 million, up 17% year-over-year and up $9 million from last quarter. Non-GAAP operating loss was $3.1 million. Customers with more than $100,000 in ARR grew to $727,000, an increase of 18% year-over-year.

第一季營收為 9,400 萬美元，年增 17%。年度經常性收入（ARR）為 3.74 億美元，年增 17%，較上季增加 900 萬美元。非 GAAP 營業虧損為 310 萬美元。ARR 超過 10 萬美元的客戶數成長至 727 家，年增 18%。

Our progress in expanding the enterprise and growing our multi-product footprint continued in the first quarter. Dollar-based net expansion improved sequentially to 106%. This reflects continued strength in our core business as we expand the capabilities of our platform to help the next generation of builders understand, improve, and grow their digital products.

我們在拓展企業客群並擴大多產品滲透率方面的進展，在第一季持續推進。以美元計價的淨擴張率（Dollar-based net expansion）按季改善至 106%。這反映出我們核心業務的持續強勁，同時我們也在擴展平台能力，協助下一代建置者理解、改善並成長其數位產品。

I am focused on aggressively transforming Amplitude into an AI company. In Q1, we made broader changes to the leadership within go-to-market to remove layers and become a more technical team. Nate Crook is now our Chief Commercial Officer overseeing sales, customer success, revenue operations, and enablement. Nate and the team now own the entire path from landing a customer to ensuring they succeed long-term.

我正專注於積極將 Amplitude 轉型為一家 AI 公司。第一季，我們在 go-to-market 團隊的領導架構上做了更廣泛的調整，去除層級、打造更具技術性的團隊。Nate Crook 現任我們的首席商務長（Chief Commercial Officer），負責銷售、客戶成功、營收營運與賦能（enablement）。Nate 與團隊現在負責從取得客戶到確保其長期成功的完整路徑。

We restructured customer success and marketing to match customer buying trends. Customer success now has fewer handoffs and deep tech coverage with four deployed engineers. Marketing is now oriented around AI-native storytelling.

我們也重整了客戶成功與行銷部門，以符合客戶的採購趨勢。客戶成功團隊現在交接更少，並由四位派駐工程師提供深度技術覆蓋。行銷則轉向以 AI 原生的敘事方式為核心。

We welcomed Gab Menachem as Chief Product Officer last month. Gab is a serial founder who built Loom Systems, an AIOps company acquired by ServiceNow. Loom Systems analyzed log data across cloud and on-prem, similar to what Amplitude does for behavioral data. Gab then spent six years scaling ServiceNow's IT operations management business to more than $1 billion in revenue. I'm excited about that combination of founder DNA and enterprise experience at scale. Gab is part of a growing group of founders we brought into Amplitude over the past 18 months to lead our AI transformation.

上個月我們迎來 Gab Menachem 擔任首席產品長（Chief Product Officer）。Gab 是連續創業者，曾創辦 Loom Systems——一家被 ServiceNow 收購的 AIOps 公司。Loom Systems 分析雲端與地端（on-prem）環境的日誌資料，這與 Amplitude 對行為資料所做的事情相似。之後 Gab 花了六年時間，將 ServiceNow 的 IT 營運管理業務擴展到超過 10 億美元營收。我對這種「創業者 DNA」與大規模企業經驗的結合感到非常振奮。Gab 也是我們在過去 18 個月引進的一群創辦人領袖之一，將帶領 Amplitude 的 AI 轉型。

A few weeks ago, we ran AI Week at Amplitude. We paused normal work across the entire company so that every function could build and ship AI-powered workflows to reimagine their daily jobs and functions. It is much more important for Amplitude's talent to be AI native over the next year than any short-term initiative in the business. The team shipped hundreds of amazing demos, including automatically creating custom demo websites per customer, automating part of the quarter close process, and automating how we create new creative assets in marketing.

幾週前，我們在 Amplitude 舉辦了 AI Week。我們暫停全公司日常工作，讓每個職能都能打造並上線 AI 驅動的工作流程，重新想像其日常工作與職能。對 Amplitude 而言，未來一年讓人才成為 AI 原生，比任何短期的業務計畫都更重要。團隊交付了數百個精彩的 demo，包括為每位客戶自動建立客製化 demo 網站、自動化季度結帳流程的一部分，以及自動化行銷中新創意素材的產出方式。

Yesterday, we announced a strategic partnership with Statsig. Amplitude, as part of this partnership, Amplitude will take on Statsig's brand and customers. We will also maintain and develop the current Statsig platform across the cloud and data warehouse, including support for all existing Statsig customers. Amplitude will also begin building a more integrated roadmap for the future of Amplitude and Statsig platforms together. We will work closely with the Statsig team at OpenAI during this transition.

昨天，我們宣布與 Statsig 建立策略合作夥伴關係。作為此合作的一部分，Amplitude 將承接 Statsig 的品牌與客戶。我們也將在雲端與資料倉儲環境中維護並開發現有的 Statsig 平台，並支援所有現有 Statsig 客戶。Amplitude 也將開始為 Amplitude 與 Statsig 平台的共同未來建立更整合的產品路線圖。在此過渡期間，我們將與 OpenAI 的 Statsig 團隊密切合作。

As context for the move, AI has dramatically lowered the barrier to building and shipping software, boosting productivity for experienced engineers and enabling non-traditional roles to become AI builders. While teams can generate more code than ever before, the software development lifecycle remains bottlenecked in many other places. AI builders are generating code faster than they can understand its impact. The challenge is now evaluating code before it's released, tracking what's working after release, knowing when you need to roll things back, and turning behavioral signals into what to build next.

作為此舉的背景，AI 大幅降低了建置與交付軟體的門檻，提升資深工程師的生產力，也讓非傳統職能成為 AI 建置者。雖然團隊能生成比以往更多的程式碼，但軟體開發生命週期在許多其他環節仍然受限。AI 建置者生成程式碼的速度，已快過他們理解其影響的速度。現在的挑戰在於：在發布前評估程式碼、發布後追蹤哪些有效、知道何時需要回滾，以及把行為訊號轉化為下一步該建置什麼。

Amplitude is the market leader and is focused on giving the best behavioral insights to product managers. Statsig has reinvented experimentation and feature management and done an amazing job with data leaders with its warehouse native capabilities. Together, we can accelerate the software development lifecycle. We now offer organizations access to these same capabilities that the world's most advanced AI companies use today.

Amplitude 是市場領導者，專注於為產品經理提供最佳的行為洞察。Statsig 重新定義了實驗與功能管理，並以其倉儲原生（warehouse native）能力在資料領導者族群中表現出色。結合雙方，我們能加速軟體開發生命週期。我們現在讓各組織也能使用到與全球最先進 AI 公司今日所使用的同等能力。

Initial customer feedback has been promising. Many of our existing customers have already expressed interest in the Statsig product. The pace at which Amplitude builds and ships products continues to accelerate. Over 90% of the code our team ships today is written by AI. I want to show you four quick demos today, each one reflecting a different dimension of what it means to close the product development loop.

初步的客戶回饋相當正面。我們許多既有客戶已表達對 Statsig 產品的興趣。Amplitude 建置與交付產品的速度持續加快。我們團隊目前交付的程式碼中，超過 90% 由 AI 撰寫。今天我想快速展示四個 demo，每一個都反映了「閉合產品開發迴路」所代表的不同面向。

I want to start with agent analytics. Everyone building agents has one big question, are they working? With agent analytics, customers get complete visibility into every agent interaction. See every conversation's full thread, what the user asked, what the agent responded, which model was used, how many tokens it burned, and how long it took to complete.

我想先從代理（agent）分析開始。每個在打造代理的人都有一個大問題：它們真的有在運作嗎？透過代理分析，客戶可以完整掌握每一次代理互動。你可以看到每段對話的完整串、使用者問了什麼、代理回了什麼、使用了哪個模型、消耗了多少 token，以及完成所花的時間。

Once the conversation completes, evaluators automatically run and judge your agent's performance across dimensions like user satisfaction, agent confusion, response quality, and task completion. This happens on every conversation and is fully customizable, so you can build evaluators specific to your use case.

當對話結束後，評估器會自動執行，並從使用者滿意度、代理困惑程度、回覆品質與任務完成度等維度，判定你的代理表現。這會發生在每一次對話上，且可完全自訂，因此你可以建立符合自身使用情境的專屬評估器。

Then you can put all your agent analytics data together with your customer data. You can see how your agent's performance directly connects to real customer events, like what impact an original agent interaction can have on a customer later completing a purchase.

接著，你可以把所有代理分析資料與你的客戶資料整合在一起。你能看到代理表現如何直接連結到真實的客戶事件，例如一次最初的代理互動，如何影響客戶之後完成購買。

We have been shipping faster around our Amplitude agents. We've added productivity updates to our global agent, including voice-to-text input for natural language prompting, image upload for providing deeper context, searchable chat history, and conversation history across projects.

我們在 Amplitude agents 的交付速度也更快了。我們為全球代理加入了生產力更新，包括語音轉文字輸入以進行自然語言提示、圖片上傳以提供更深層的脈絡、可搜尋的聊天紀錄，以及跨專案的對話歷史。

In addition to that, we've also added memory, so agents now monitor when they're corrected or directed in specific ways and save that for the future. For example, a weekly active user in Amplitude is a user who saves a chart, not someone who simply logs in. The agent, after telling the agent this, it remembers it for future analysis instead of needing to be corrected every time. 90% of these memories are automatically created as people use agents, so agents get smarter and better the more people use them.

除此之外，我們也加入了記憶功能，因此代理現在會在被糾正或被特定方式指示時進行監測，並將其保存以供未來使用。舉例來說，在 Amplitude 中，每週活躍使用者是指「儲存圖表」的使用者，而不是僅僅登入的人。當你告訴代理這件事之後，它會在未來分析中記住，而不需要每次都被糾正。這些記憶中有 90% 會在使用者使用代理的過程中自動建立，因此使用的人越多，代理就會越聰明、越好用。

We also have MCP connectors built directly into agents. Agents are incredible at analysis, but the connection to action is broken. A human still needed to file the linear ticket, do the write-up in Notion, or read the Slack channel for context. Not anymore with MCP connectors. All of these actions can be triggered automatically. Non-event data can now be paired with Amplitude's data, connect financial information from BigQuery, and quantify the real cost benefit of an experiment or with GitHub and Amplitude, retroactively track how specific releases affect error rates, session length, or feature adoption.

我們也把 MCP 連接器直接內建到代理中。代理在分析方面非常強，但「從洞察到行動」的連結一直是斷裂的。過去仍需要人去建立 Linear 工單、在 Notion 撰寫說明，或閱讀 Slack 頻道以取得脈絡。有了 MCP 連接器，就不再需要了。所有這些動作都可以自動觸發。非事件（non-event）資料現在也能與 Amplitude 的資料配對，例如從 BigQuery 連接財務資訊，量化一次實驗的真實成本效益；或透過 GitHub 與 Amplitude，回溯追蹤特定版本發布如何影響錯誤率、工作階段長度或功能採用率。

Now our agents can run and connect to all your data sources to surface insights and deliver the context wherever it's needed. Institution customers experienced, they had agents running on Amplitude surfacing insights for a new interface rollout they were planning. One of those agents surfaced pages that were indexed incorrectly before they went live without being instructed to find the inaccuracies. That helped the team avoid serving incorrect data to customers without even being asked. That is what it looks like when the loop closes on the right side automatically. People don't check dashboards. The system catches the problem before it become one.

現在，我們的代理可以執行並連接到你所有的資料來源，以挖掘洞察，並在需要的地方提供脈絡。有機構型客戶分享，他們讓代理在 Amplitude 上運行，為其規劃中的新介面上線提供洞察。其中一個代理在未被指示要找出不一致之處的情況下，就在上線前發現有些頁面被錯誤索引。這幫助團隊避免向客戶提供錯誤資料，甚至不需要被要求。這就是當迴路在右側自動閉合時的樣子。人們不會一直看儀表板；系統會在問題變成問題之前就把它抓出來。

We're also building new AI products to expand our platform for customers. Our most recent launch was AI Assistant. AI Assistant is a chatbot that answers customers' questions in real time like Intercom Fin. It's tied into Amplitude, so it can know who users are, where they've been previously, and where they are right now. If users want to know how to accomplish a task, instead of giving text instructions, it can create a visually guided tour that walks users through the interface.

我們也正在打造新的 AI 產品，以擴展我們為客戶提供的平台能力。我們最近一次的發布是 AI Assistant。AI Assistant 是一個即時回答客戶問題的聊天機器人，類似 Intercom Fin。它與 Amplitude 串接，因此能知道使用者是誰、先前去過哪裡，以及他們此刻在哪裡。如果使用者想知道如何完成某項任務，它不只提供文字指引，還能建立視覺化導覽，帶著使用者一步步走過介面。

Here, I'm asking how to integrate with Slack, and it's triggering a guide that helps me do so. It shows me where to click on the screen and guides me through the process. This is live for customers to purchase today, and it's a great way to highlight how we're using AI to infuse context and understanding of the user for our customers. The last demo I want to show you today is our command line interface wizard. AI builders need an automated installation of Amplitude. That is why we built the CLI Wizard.

在這裡，我在詢問如何與 Slack 整合，而它會觸發一份指南來協助我完成。它會告訴我在螢幕上要點哪裡，並引導我走完整個流程。這項功能今天已可供客戶即時購買使用，這也是一個很好的方式，凸顯我們如何運用 AI，為客戶注入情境與對使用者的理解。我今天最後想展示的示範是我們的命令列介面精靈。AI 建置者需要自動化安裝 Amplitude。這就是我們打造 CLI 精靈的原因。

Setting up Amplitude used to be a sticking point for some users in the past. Now with our CLI Wizard, it's one line of code in the terminal. The rest is done for them. The CLI Wizard package runs against any code base, any programming language, and it instruments Amplitude for you. It adds SDKs, creates the taxonomy, and instruments all events, and configures MCP. It will even create an initial dashboard for you. What used to take weeks now takes a few minutes. All initial setup into one action, dead simple install for humans. After this, we're going to give the ability for agents to install Amplitude automatically in the cloud. There is now no barrier to installing Amplitude.

過去，設定 Amplitude 對部分使用者而言曾是一個卡點。現在有了我們的 CLI 精靈，在終端機只需要一行程式碼，其餘都會替他們完成。CLI 精靈套件可在任何程式碼基底、任何程式語言上執行，並為你完成 Amplitude 的埋點。它會加入 SDK、建立分類體系（taxonomy）、為所有事件進行埋點，並設定 MCP。它甚至會為你建立一個初始儀表板。過去需要數週的工作，現在只要幾分鐘。把所有初始設定濃縮成一次動作，對人類而言是極度簡單的安裝。接下來，我們還會提供讓代理（agents）在雲端自動安裝 Amplitude 的能力。現在安裝 Amplitude 已經沒有任何門檻。

Let me tell you about a few customers who are putting this to work. Granola is one of the fastest-growing AI companies out there. They came to Amplitude before they had actually -- before they had even launched because they wanted to understand from day one whether what they were shipping was actually working for users. Today, more than half the company uses Amplitude every day. And actually, I think Amplitude is a Granola user. They ship new features fast and rely on real-time behavioral signals to decide what to do next. They have grown with us horizontally and use the full platform. Granola is what a next-generation software company looks like. No separate analytics team, no weekly reporting cycle. The loop from ship to learn runs continuously, and Amplitude is the infrastructure that makes it possible.

我來跟你分享幾位正在把這套方案用起來的客戶。Granola 是目前成長最快的 AI 公司之一。他們在實際——在他們甚至還沒正式上線之前就來找 Amplitude，因為他們希望從第一天起就能了解他們交付的內容對使用者是否真的有效。如今，公司超過一半的人每天都在使用 Amplitude。而且其實，我覺得 Amplitude 也是 Granola 的使用者。他們快速推出新功能，並依賴即時的行為訊號來決定下一步要做什麼。他們在橫向上與我們一起成長，並使用完整平台。Granola 展現了下一代軟體公司的樣貌：沒有獨立的分析團隊，沒有每週一次的報告週期。從發布到學習的迴圈持續運轉，而 Amplitude 正是讓這一切成為可能的基礎設施。

Smartsheet is an intelligent work management platform that helps enterprises unite people, data, and AI to turn strategy into results. As Smartsheet accelerated its push into AI-driven experiences, the team faced a real bottleneck. Their product managers were entirely dependent on the BI team for every single insight. A question as basic as how many people used this feature last month and what does this mean for retention could take weeks to answer. Today, with Amplitude analytics, feature experimentation, and guides and surveys, Smartsheet's product managers, engineers, designers, and researchers have that answer instantly. They've used those insights to identify and fix drop-off in their onboarding funnel with a direct, measurable impact on retention. As Smartsheet invests in AI, Amplitude gives them the velocity to understand whether new experiences are working at the speed their ambitions demand.

Smartsheet 是一個智慧化工作管理平台，協助企業整合人員、資料與 AI，將策略轉化為成果。隨著 Smartsheet 加速推進 AI 驅動的體驗，團隊遇到了一個真正的瓶頸：他們的產品經理在每一個洞察上都完全依賴 BI 團隊。像是「上個月有多少人使用這個功能？這對留存意味著什麼？」這樣基本的問題，可能都要花上數週才能回答。如今，透過 Amplitude 的分析、功能實驗，以及指南與調查，Smartsheet 的產品經理、工程師、設計師與研究人員可以即時得到答案。他們也運用這些洞察找出並修正新手引導漏斗中的流失點，對留存帶來直接且可衡量的影響。當 Smartsheet 投資 AI 時，Amplitude 讓他們能以其雄心所要求的速度，理解新體驗是否奏效。

Astra Tech chose Amplitude as its partner to support Bottom and Bottom Money's evolution into an AI-native fintech super app for over 150 million users across 150 countries. Bottom uses insights from Amplitude to optimize fintech entry points, pinpoint critical journey drop-offs, and establish clear engagement baselines for Bottom AI across user segments, usage patterns, and downstream actions. With cross-functional teams in growth, design, tech, using those insights to steer a completely revamped Bottom to reposition itself as a fintech-led communications platform.

Astra Tech 選擇 Amplitude 作為合作夥伴，支持 Botim 與 Botim Money 演進為 AI 原生的金融科技超級 App，服務遍及 150 個國家、超過 1.5 億名使用者。Botim 使用來自 Amplitude 的洞察來優化金融科技入口、精準定位關鍵旅程的流失點，並為 Botim AI 在不同使用者分群、使用模式與後續行為上建立清晰的互動基準。其成長、設計、技術等跨職能團隊運用這些洞察，引導全面改版的 Botim，將自身重新定位為以金融科技為核心的通訊平台。

I'm very excited to share the business impact we had with them. Their revamp across services, including international transfer, local transfer, ad funds, and gold, Astra Tech increased fintech service entries by 4%, lifted engagement from top offers in 4U by up to 3%, and grew fintech transacting users by three times. That happened all within a span of nine months of working with Amplitude.

我非常興奮能分享我們與他們合作帶來的業務影響。透過涵蓋多項服務的改版，包括國際匯款、本地轉帳、廣告資金與黃金，Astra Tech 將金融科技服務入口提升了 4%，把 4U 中頂級優惠帶來的互動提升最高達 3%，並讓進行金融交易的使用者成長至 3 倍。這些成果都發生在與 Amplitude 合作的九個月內。

I want to note that the companies on the bleeding edge of the tech industry are Amplitude customers. That's because the faster you build, the more you need to know what to build next. AI natives understand that better than anyone. This underscores the long-term case for Amplitude. AI makes what we do more critical than ever. We are set up to close the right side of the product development loop, and we have the platform, the customers, the leadership, and the conviction to see it through. I'm extraordinarily excited for what's next.

我想指出，站在科技產業最前沿的公司都是 Amplitude 的客戶。因為你建置得越快，就越需要知道下一步該建置什麼。AI 原生公司比任何人都更理解這一點。這也凸顯了 Amplitude 的長期投資論點。AI 讓我們所做的事情比以往任何時候都更關鍵。我們已準備好補上產品開發迴圈的右半邊，我們擁有平台、客戶、領導力與信念去把它完成。我對接下來的發展感到無比興奮。

Now, over to Andrew to walk you through the financials.

接下來交給 Andrew，帶大家看財務表現。

Thank you, Spenser, and good afternoon, everyone.

謝謝你，Spenser，各位下午好。

The first quarter was solid with incremental improvement and our dollar-based net retention to 106%, multi-product accounts for more than 77% of our ARR, and our ARR growth was 17%. We've made our guidance on both top and bottom line, and we are combining the best of Statsig with Amplitude.

第一季表現穩健並有逐步改善：以美元計的淨留存率提升至 106%，多產品客戶占我們 ARR 的 77% 以上，而 ARR 成長率為 17%。我們在營收與獲利（上、下線）指引上都達標，並且正在把 Statsig 與 Amplitude 的優勢結合起來。

Reflecting on Q1, there are many changes in our go-to-market team. We've introduced a number of new AI products, and Amplitude has been implementing a host of new AI-based workflows to drive efficiency. We are at a moment of transformation. We are transforming the value our customers receive, we are transforming how we deliver value, and we are transforming our organization from the ground up. We've done. This while continuing to execute on our core business.

回顧第一季，我們的 go-to-market 團隊有許多變化。我們推出了多項新的 AI 產品，而 Amplitude 也正在導入一系列以 AI 為基礎的新工作流程以提升效率。我們正處於轉型時刻：我們正在改變客戶獲得的價值、改變我們交付價值的方式，並從根本改造我們的組織。同時，我們也持續在核心業務上穩健執行。

We are leveraging AI at scale across our organization and helping customers unlock incremental value faster. No longer is a good piece of code with a friendly UI good enough. We must deliver customer valued outcomes. We are focused on becoming a true partner with our customers to understand how to apply the technology in the most effective ways. We are building on a decade of understanding, context, but delivering this knowledge through our services, our platform, and our know-how.

我們在整個組織中大規模運用 AI，並協助客戶更快解鎖額外價值。如今，只有一段好的程式碼加上友善的 UI 已經不夠了；我們必須交付客戶重視的成果。我們專注於成為客戶真正的夥伴，理解如何以最有效的方式應用這項技術。我們建立在十年累積的理解與情境之上，並透過我們的服務、平台與專業能力（know-how）來交付這些知識。

The speed of change has accelerated, and we're leaning into that moment. We're seeing increased usage of our AI agents, along with data ingested into our platform. This has created some headwinds in our cost to serve, but it's also aligned to our monetization strategy. Adapting quickly and delivering greater value to our customers will be the advantage of the next generation of winners in software, which is why we've made changes to our products, pricing, and internal operations.

變化的速度正在加快，而我們正積極把握這個時刻。我們看到 AI 代理的使用量增加，同時匯入平台的資料量也在上升。這使我們的服務成本面臨一些逆風，但也與我們的變現策略一致。能夠快速適應並為客戶交付更大價值，將成為下一代軟體贏家的優勢；因此我們已對產品、定價與內部營運做出調整。

Taking on the Statsig business is another great example of our ability to be flexible and act quickly. By combining Statsig's industry-leading warehouse native experimentation with Amplitude's best-in-class analytics platform, we're expanding our total addressable market and meeting customers where their data needs are. We will build this business to be incremental and accretive to our core business.

承接 Statsig 業務是我們具備彈性並能快速行動的另一個絕佳例子。透過結合 Statsig 業界領先、以資料倉儲為原生（warehouse native）的實驗能力，以及 Amplitude 同級最佳的分析平台，我們正在擴大可服務的總市場（TAM），並在客戶的資料需求所在之處與他們相遇。我們將把這項業務打造為對核心業務具增量貢獻，並且具增益性（accretive）。

Spenser highlighted some of the changes our team has undergone, and we're instrumenting the business for long-term scale and efficiency, so that driving business growth continues to result in greater leverage. That being said, our goals as a business remain steady. We want to grow our enterprise business, expand our multi-product footprint, and deliver great value for our customers.

Spenser 強調了我們團隊經歷的一些變化，而我們也正在為業務建立長期規模化與效率的能力，讓推動業務成長能持續帶來更高的槓桿效益。即便如此，我們作為一家公司的目標仍然穩定：成長企業客戶業務、擴大多產品覆蓋，並為客戶交付卓越價值。

This focus has enabled us to drive consolidation in the market through our platform approach, now having over 77% of our ARR coming from customers with more than two products, up 3 points from last quarter. Customers with five or more products now account for 24% of our ARR, up from 20% last quarter. We believe that as customers continue to adopt our AI products, they will naturally expand their use cases into the full suite of our platform and drive incremental upsell opportunities.

這項聚焦使我們能透過平台策略推動市場整併：目前超過 77% 的 ARR 來自使用兩個以上產品的客戶，較上季提升 3 個百分點。使用五個以上產品的客戶目前占 ARR 的 24%，高於上季的 20%。我們相信，隨著客戶持續採用我們的 AI 產品，他們將自然把使用情境擴展到平台的完整套件，並帶動額外的加購（upsell）機會。

Turning to our first quarter and full-year results.

接下來看第一季與全年業績。

As a reminder, all financial results I will be discussing with the exception of revenue are non-GAAP. Our GAAP financial results along with reconciliation between GAAP and non-GAAP can be found in our earnings press release and supplemental financials on the Investor Relations’ page of our website.

提醒一下，除營收外，我將討論的所有財務結果皆為非 GAAP。我們的 GAAP 財務結果，以及 GAAP 與非 GAAP 之間的調節表，可在我們的財報新聞稿與官網投資人關係（Investor Relations）頁面的補充財務資料中找到。

First quarter revenue was $93.5 million, up 17% year-over-year versus 10% in the first quarter of 2025. Total ARR increased to $374 million exiting the first quarter, an increase of 17% year-over-year and $9 million sequentially. Total remaining performance obligations grew 31% year-over-year to $427 million compared to 30% growth in Q1 2025.

第一季營收為 9,350 萬美元，年增 17%，相較之下 2025 年第一季為 10%。第一季結束時，總 ARR 增至 3.74 億美元，年增 17%，並較前一季增加 900 萬美元。總剩餘履約義務（remaining performance obligations）年增 31% 至 4.27 億美元，而 2025 年第一季的成長率為 30%。

Current RPO was up 20% year-over-year compared to 18% in Q1 of last year. Long-term RPO was up 60% year-over-year compared to 72% from the first quarter of last year. We had a strong quarter for both new and expansion deals in the enterprise. Platform sales were also particularly strong. 47% of our customers now have multiple products, with 77% coming from that cohort. We have made great progress on expanding our multi-product footprint within our customer base compared to a year ago when only 30% of our customers had multi-products and accounted for only 64% of our ARR.

目前 RPO 年增 20%，相較於去年第一季的 18%。長期 RPO 年增 60%，相較於去年第一季的 72%。本季在企業客戶的新單與擴張交易方面皆表現強勁。平台銷售也特別強勁。目前 47% 的客戶已採用多項產品，而 ARR 的 77% 來自該族群。相較一年前（當時僅 30% 的客戶採用多產品，且僅貢獻 64% 的 ARR），我們在擴大客戶群的多產品覆蓋率方面已取得重大進展。

The number of customers representing 100,000 or more of ARR in Q1 grew to 727,000, an increase of 18% year-over-year and up [29,000], customers since last quarter.

第一季 ARR 達 100,000 以上的客戶數增至 727,000，年增 18%，且較上季增加 [29,000] 名客戶。

In period, net dollar retention increased to 106% from 105% last quarter, led by cross-sell expansions across our customer base. We expect net dollar retention to improve over the long-term as we continue to see customers adopt multi-product. However, it may not be in a linear fashion.

本期淨美元留存率由上季的 105% 提升至 106%，主要由全體客戶群的交叉銷售擴張所帶動。我們預期，隨著客戶持續採用多產品，淨美元留存率將在長期改善；然而，其改善可能不會呈線性。

Gross margin was 75% for the first quarter, down 2 points from the first quarter of last year. This was largely driven by growth in inference costs as adoption of our AI tools by our customers outpaced our expectations. We now expect this adoption trend to continue, given the feedback we've received from our customers. In the short-term, this will cause gross margin compression, but we believe this will help us to drive greater data ingestion and monetization of our core platform over time.

第一季毛利率為 75%，較去年第一季下降 2 個百分點。主要原因是推論成本成長，因客戶對我們 AI 工具的採用速度超出預期。根據我們從客戶收到的回饋，我們現在預期此採用趨勢將持續。短期內這將造成毛利率承壓，但我們相信，長期而言這將有助於推動更多資料匯入，並提升核心平台的變現能力。

Sales and marketing expenses were 45% of revenue, in line with the first quarter from last year. Some of the increase in costs included severance costs related to our organizational changes and other activities like our go-to-market kickoff that occurred in the first quarter. We have focused our entire go-to-market team on driving value for our customers, increasing adoption organization-wide, and improving our internal processes, coverage, and expanding the buyer personas that we can sell to. These changes will take time to manifest and net new ARR, but ultimately, they will increase the health of our customer base and drive greater opportunity to grow our net dollar-based retention.

銷售與行銷費用占營收 45%，與去年第一季持平。成本增加的一部分包含與組織調整相關的資遣費用，以及第一季發生的市場推進（go-to-market）啟動活動等其他項目。我們已將整個市場推進團隊聚焦於為客戶創造價值、提升全組織的採用率，並改善內部流程與覆蓋，以及擴大我們可銷售的買方角色（buyer personas）。這些變革需要時間才會反映在淨新增 ARR 上，但最終將提升客戶基礎的健康度，並帶來更大的機會以提高我們的淨美元留存率。

R&D was 20% of revenue, up 1 point in the same period last year. We will be adding to the team to scale the Statsig opportunity and continue to support those customers.

研發費用占營收 20%，較去年同期上升 1 個百分點。我們將擴編團隊以擴大 Statsig 的機會，並持續支援這些客戶。

G&A was 13% of revenue, down 2 points from the first quarter of 2025, and we expect G&A to improve as a percentage of revenue over time. Total operating expenses were $73 million or 78% of revenue, down 1 point from the same period a year ago. Operating loss was $3.1 million or 3.3% of revenue. Net loss per share was $0.02 based on 133.3 million basic shares, compared to a net loss per share of $0.00 with 129.7 million shares a year ago.

一般與行政（G&A）費用占營收 13%，較 2025 年第一季下降 2 個百分點；我們預期 G&A 占營收比重將隨時間改善。總營業費用為 7,300 萬美元，占營收 78%，較一年前同期下降 1 個百分點。營業虧損為 310 萬美元，占營收 3.3%。每股淨虧損為 0.02 美元，係以 1.333 億股基本股數計算；相較一年前以 1.297 億股計算的每股淨虧損 0.00 美元。

Free cash flow in the quarter was a negative $13.2 million or negative 14% of revenue, compared to a negative $9.2 million or negative 12% of revenue during the same period last year.

本季自由現金流為負 1,320 萬美元，或占營收負 14%；相較去年同期為負 920 萬美元，或占營收負 12%。

We continue to be active in the open market last quarter, retiring shares against our open buyback. We have conviction in the long-term value of our platform and have used and will use our cash to minimize the impacts of dilution, while our share price continues to not align with the value we believe we're creating. Our balance sheet position remains strong and allows us the opportunity to be more aggressive in our M&A strategy to accelerate our R&D roadmap when appropriate.

我們上季持續在公開市場上回購並註銷股份，以對沖既有回購計畫下的股份。我們對平台的長期價值深具信心，並已使用且將持續使用現金以降低稀釋影響，因為我們的股價仍未反映我們認為正在創造的價值。我們的資產負債表仍然強健，並使我們在適當時機能更積極採取併購（M&A）策略，以加速我們的研發路線圖。

In Q2, we will also take into consideration bringing the Statsig customers and technology over to Amplitude as of the beginning of May. To start, we will record an additional $16 million in incremental ARR from the Statsig customer base, aligning that business to our definition of ARR. As we take on the Statsig business, we will also be investing in a transition team as we ramp an internal team to continue to provide the best support for the Statsig customers.

在第二季，我們也將考量自 5 月初起把 Statsig 的客戶與技術併入 Amplitude。首先，我們將自 Statsig 客戶基礎認列額外 1,600 萬美元的增量 ARR，使該業務與我們對 ARR 的定義一致。隨著我們承接 Statsig 業務，我們也將投資於轉換團隊，同時擴編內部團隊，以持續為 Statsig 客戶提供最佳支援。

Over time, we will scale our internal team to continue to develop the warehouse native and cloud aspects of Statsig. Additionally, there will be some pressure on gross margins for the remainder of the year as we integrate and optimize our hosting environment.

隨著時間推進，我們將擴大內部團隊，持續開發 Statsig 的資料倉儲原生（warehouse native）與雲端面向。此外，隨著我們整合並最佳化託管環境，今年剩餘期間毛利率將面臨一些壓力。

Now, turning to our outlook.

接下來談談我們的展望。

As a reminder, the philosophy of how we set guidance is through the lens of execution. We are pleased with our progression on driving adoption of our core platform, our different AI technologies, and multi-product adoption. Our new pricing and packaging rollout is progressing very well, and in the first quarter, 25% of total ARR contracted, both new business and renewals, was under new pricing and packaging. We will continue to increase this percentage as we make it easier for our sellers to quote and make it easier for our customers to understand the path to platform adoption.

提醒一下，我們設定財測指引的理念是以執行為核心。我們對推動核心平台、各項 AI 技術以及多產品採用的進展感到滿意。我們新的定價與方案（packaging）推出進展非常順利；在第一季，無論是新業務或續約，簽約的總 ARR 中有 25% 採用新的定價與方案。隨著我們讓銷售更容易報價、也讓客戶更容易理解採用平台的路徑，我們將持續提高此比例。

We are already seeing early signs of willingness to test new features and products on the platform, given the easier on-ramp from a contract view. This will also lend itself to allowing easier adoption of our AI agents as we continue to iterate and ship.

由於從合約角度看更容易上手，我們已看到客戶願意在平台上測試新功能與新產品的早期跡象。這也將有助於在我們持續迭代與交付的過程中，更容易採用我們的 AI 代理（AI agents）。

So for the first -- for the second quarter of 2026, we expect revenue to be between $96.9 million and $99.1 million, representing an annual growth rate of 18% at the midpoint. We expect non-GAAP operating income to be between negative $3.6 million and negative $1.6 million. We expect non-GAAP net income per share to be between negative $0.02 and negative $0.01, assuming basic weighted average shares outstanding of approximately $134 million.

因此，對於 2026 年第二季，我們預期營收介於 9,690 萬至 9,910 萬美元之間，以中位數計算年增率為 18%。我們預期非 GAAP 營業利益介於負 360 萬至負 160 萬美元之間。我們預期非 GAAP 每股淨利介於負 0.02 至負 0.01 美元之間，假設基本加權平均流通股數約為 1.34 億股。

For the full year of 2026, we expect full year revenue to be between $397 million and $403 million, an annual growth rate of 17% at the midpoint. This assumes a $5 million to $7 million contribution from the Statsig business, taking into account the assumption of the customer contracts and the impacts to the deferred. We expect our full-year non-GAAP operating income to be between $2.5 million and $6.5 million. This reflects incremental investment we'll need to incorporate the stats in business. We expect non-GAAP net income per share to be between $0.03 and $0.06, assuming weighted average shares outstanding of approximately $145.1 million is measured on a fully diluted basis.

對於 2026 全年，我們預期全年營收介於 3.97 億至 4.03 億美元之間，以中位數計算年增率為 17%。此預估包含 Statsig 業務貢獻 500 萬至 700 萬美元，並考量承接客戶合約以及對遞延項目的影響。我們預期全年非 GAAP 營業利益介於 250 萬至 650 萬美元之間。這反映了我們為納入 Statsig 業務所需的額外投資。我們預期非 GAAP 每股淨利介於 0.03 至 0.06 美元之間，假設加權平均流通股數約為 1.451 億股，並以完全稀釋基礎衡量。

In closing, we are accelerating our pace of innovation, and we're growing the value that we can deliver to our customers. We have confidence in our ability to scale a durable and growing business while also bringing agentic analytics to the world.

最後，我們正在加速創新步伐，並提升我們能為客戶交付的價值。我們有信心在擴大一個具韌性且持續成長的業務的同時，也把代理式分析（agentic analytics）帶給全世界。

With that, we'll open it up for Q&A. Over to you, John.

接下來我們將開放問答。交給你了，John。

(Event Instructions)

（活動說明）

Taylor McGinnis, UBS followed by Rob Oliver, RW Baird.

UBS 的 Taylor McGinnis，接著是 RW Baird 的 Rob Oliver。

Maybe first, Spenser, for you, could you just maybe explain why OpenAI is foregoing the Statsig business and if there's any parts that OpenAI is retaining in that?

也許先請 Spenser 回答，你能否說明一下為什麼 OpenAI 會放棄 Statsig 業務，以及 OpenAI 是否會保留其中任何部分？

And then Andrew, maybe as a second one for you, the helpful color on breaking out some of the Statsig impact this year to the guide. If we strip that out, does that mean that you're taking down, I guess, the organic growth guide a little bit on, revenue this year? And maybe you could just unpack that and the margin impact.

然後 Andrew，第二個問題想請教你：你提供了很有幫助的補充說明，將今年指引中 Statsig 的影響拆分出來。如果把那部分剔除，是否表示你們今年在營收的有機成長指引上稍微下修？也請你進一步說明一下，以及對毛利率的影響。

So first just to answer the question on the Statsig side, I mean, Vijay and I have known each other for years. Amplitude and Statsig have been competitors and kind of pushing the bleeding edge in their respective niches. I'm extraordinarily excited that we get to kind of carry a bunch of that forward with both the customers, the technology as well as the brand. I think Vijay was looking for a home for the kind of continued support of the Statsig customer base. And after looking at a number of different places, him and I agreed that the best place that would be Amplitude.

先回答關於 Statsig 的問題。我是說，Vijay 和我認識多年。Amplitude 和 Statsig 一直是競爭對手，並在各自的利基領域推動技術前沿。我非常興奮我們能把其中很多延續下去，包括客戶、技術以及品牌。我認為 Vijay 在為 Statsig 客戶群尋找一個能持續提供支援的歸宿。在看過許多不同選項後，他和我一致認為最適合的地方就是 Amplitude。

We just executed that agreement on Friday, so we're kind of still just getting up to speed with all everything it entails and making sure this is trend, making sure those customers continue to be supported and then figuring out what the long-term plans for Amplitude and Statsig are together. But I'm just -- I'm very, very, very excited about it.

我們是在週五才剛執行那份協議，所以我們某種程度上還在加緊熟悉它所涵蓋的一切，並確保這項安排順利推進、確保那些客戶能持續獲得支援，然後再釐清 Amplitude 與 Statsig 長期要如何一起發展。不過我只是——我對此非常、非常、非常興奮。

OpenAI will be continuing to run the technology internally that they have from Statsig, and so they'll be continuing to use it, but that will obviously be supported by Vijay and the existing Statsig team at OpenAI.

OpenAI 將會繼續在內部運行他們從 Statsig 取得的技術，因此他們會持續使用它，而這顯然會由 Vijay 與 OpenAI 內現有的 Statsig 團隊來提供支援。

And Taylor, part of the guidance we have is incorporating the accounting associated agreement like this, where you have to take a fair value assessment on the revenue that's aligned to the annual occurring revenue mentioned. But by taking that fair value assessment, you actually take a haircut on the value, it actually reduces down. And so what you're seeing in the amount I'm indicating that comes from ARR and the lower revenue is really related to that fair value assessment. And I would tell you that we had a good quarter in Q1, we beat expectations. What our guidance was, and that's flowed through into our guidance for FY26. So for us, we think it's a huge opportunity for us to go build out a great product that a lot of customers may be very interested in.

Taylor，我們的指引其中一部分是要納入這類協議相關的會計處理：你必須就與所提到的年度經常性收入（ARR）對應的收入做公允價值評估。但做了公允價值評估之後，實際上會對價值打折，會把金額往下調。所以你看到我所指出、來自 ARR 的金額與較低的收入，主要就是跟那個公允價值評估有關。我也想告訴你，我們在第一季（Q1）表現不錯，超出預期。我們當時給出的指引也已反映到我們對 FY26 的指引中。對我們而言，我們認為這是一個巨大的機會，讓我們去打造一個很多客戶可能會非常感興趣的優秀產品。

Perfect. Thanks for that.

完美。謝謝你的說明。

And just a quick follow-up, if I may. So if I look at the net new ARR numbers, it looks like maybe it was flattish on a year-over-year basis. I know you mentioned that there were a number of changes that you guys made in the quarter from leadership changes to pricing and packaging. So did that at all have any impact in the quarter? And maybe you could just talk about what occurred and how you guys are thinking about that for the remainder of the year?

如果可以的話，我再快速追問一下。所以如果我看淨新增 ARR 的數字，看起來可能年對年大致持平。我知道你提到你們在本季做了不少變動，從領導層變動到定價與方案包裝。因此這些是否對本季有任何影響？也許你可以談談發生了什麼，以及你們對今年剩餘時間的看法？

Whenever you make big changes in organizations or you're making changes in core processes, invariably, there's going to be an impact. I would tell you that we're pretty proud of the fact that given those changes that we made, we were still able to, one, exceed the guidance we had put out with respect to ARR and turned in a pretty good quarter, especially with respect to net double retention increasing, the number of [100 gig] customers we added. So yes, there's always going to be some impact, but I think we did a pretty good job of executing through it.

每當你在組織上做重大變更，或是在核心流程上做調整，無可避免都會帶來影響。我想說的是，我們對自己感到相當自豪：在做了這些變更的情況下，我們仍然能夠（第一）在 ARR 方面超出我們先前給出的指引，並交出一個相當不錯的季度表現，尤其是淨美元留存率（net dollar retention）提升、以及我們新增的 [100 gig] 客戶數量。所以是的，總會有一些影響，但我認為我們在執行上做得相當不錯。

Rob Oliver, RW Baird followed by Jackson Ader, KeyBanc.

接下來是 RW Baird 的 Rob Oliver，之後是 KeyBanc 的 Jackson Ader。

So I guess first question, Andrew, for you.

我想第一個問題是給 Andrew 的。

Really great progress on the new pricing model. I mean, it feels like just yesterday you guys were in pilot on that and now you're at 25% of ARR.

新的定價模型進展真的很棒。感覺就像昨天你們還在做試點（pilot），現在已經覆蓋到 25% 的 ARR 了。

I guess a couple of questions there to start. One, how should we think of the progression of that? I think you said it's key to the selling motive, but is that something as customers come up for renewal this year, we can expect that number to continue to move here? And any, recognizing it's very early, any early indications on what kind of pricing uplift or impact it's having on the contracts in terms of the combinations of usage, and then I had a quick follow-up as well.

我想先從這裡問兩個問題。第一，我們應該如何看待這個進展的節奏？我想你說過這是銷售動能的關鍵，但這是否意味著當客戶今年陸續到期續約時，我們可以預期這個比例會持續上升？另外，在承認現在還很早的前提下，有沒有任何早期跡象顯示它對合約的定價上調（pricing uplift）或影響為何——例如在使用量組合方面？我也還有一個快速追問。

Yeah, I'd say we're pretty pleased with our progress on the pricing package as well. The response from our sales team has been tremendous. They love the simplicity and the way they can actually express value back to clients. That proxy on value from a price perspective and the methodology is one that customers really understand. I can tell you there are a number of deals in Q1 that customers added more product associated with our platform because of the simplicity and the way they get cost predictability on that new strategy.

是的，我會說我們對定價與方案包裝的進展也相當滿意。我們的銷售團隊反應非常熱烈。他們喜歡它的簡潔，以及他們能夠把價值清楚地回饋給客戶。從價格角度來看，那個以價值作為代理指標（proxy on value）的方式與方法論，是客戶非常能理解的。我可以告訴你，在第一季有不少交易，客戶因為這個新策略的簡單性，以及能獲得成本可預測性，而在我們的平台上加購了更多產品。

So I do expect that the percentage of our ARR that's going to be on the new pricing and packaging will increase. We're not going to force customers through hard migrations. We're going to give them the carrot and show them the value, and we expect that customers are going to really want to adopt the new pricing and packaging.

所以我確實預期，採用新定價與方案包裝的 ARR 占比會提高。我們不會強迫客戶進行硬性遷移。我們會用「胡蘿蔔」的方式，向他們展示價值，而我們也預期客戶會真的想採用新的定價與方案包裝。

Great, helpful.

很好，很有幫助。

And then my follow-up, Spenser, in your prepared remarks, you made it clear that being an AI company right now is the most important thing. And I guess that creates a ton of exciting opportunity like around Statsig. It also creates a fair amount of uncertainty around both the gross and operating margin line.

接著我的追問是給 Spenser 的。在你準備好的發言中，你很清楚地表示，現在成為一家 AI 公司是最重要的事情。我想這帶來了很多令人興奮的機會，例如圍繞 Statsig；但同時也讓毛利率與營業利益率兩條線出現相當程度的不確定性。

Just wondering, I know we've got updated numbers for you guys and thoughts around that. But you recognized that you just closed Statsig on Friday. Can we expect at some point perhaps this year we'll get updated thoughts from you guys around costs to integrate, go-to-market and potential further impacts both on the gross cost to serve side as well as on the operating margin side.

我想問的是，我知道你們已經提供了更新的數字與一些看法。但你也提到你們是週五才剛完成 Statsig 的交割。今年之內我們是否可以期待在某個時間點，你們會更新對整合成本、go-to-market，以及對毛利端的服務成本（gross cost to serve）與營業利益率端可能進一步影響的看法？

Thanks, guys. Appreciate it.

謝謝各位，感謝你們。

For sure.

當然。

So, yeah, I mean, again, it's a few days old, so we did our best with the guidance that we put out, but we'll absolutely have a much better picture as we get into the next quarter and the subsequent quarters.

所以，是的，我的意思是，這件事畢竟才過了幾天，所以我們已經盡力在我們發布的指引中納入考量；但隨著我們進入下一季以及後續幾季，我們絕對會有更清楚的圖像。

Let me talk about Statsig specifically, and then I'll talk about more generally on the AI transformation with Amplitude. Statsig specifically, I think it'll be long-term very accretive to the business. A lot of Amplitude customers are very interested in their product. A lot of Amplitude customers are also Statsig customers, like we just talked to one yesterday, Atlassian, who's a big user on the Statsig experimentation side while being a big user on the Amplitude Analytics and they're actually really excited because now the data from the two products will talk to each other, and that'll drive a whole bunch more value and usage and good things for both Atlassian and Amplitude. And we expect to see similar things across the entire customer base.

我先具體談 Statsig，然後再更一般性地談 Amplitude 的 AI 轉型。就 Statsig 而言，我認為長期來看它會對業務非常具增益（accretive）。很多 Amplitude 客戶對他們的產品非常有興趣。也有很多 Amplitude 客戶同時也是 Statsig 客戶——就像我們剛才提到的，我們昨天才跟其中一家聊過：Atlassian。他們在 Statsig 的實驗（experimentation）方面是大型用戶，同時也是 Amplitude Analytics 的大型用戶；而他們其實非常興奮，因為現在兩個產品的資料可以互通，這將為 Atlassian 與 Amplitude 帶來更多價值、更多使用量，以及更多正向效益。我們也預期在整個客戶群中會看到類似的情況。

Now, in terms of exactly quantifying them, again, it's very rough in the air because it's only a few days old, but we'll have a much better picture into it come next earnings call.

至於要精確量化這些影響，再次強調，因為才過了幾天，所以目前仍非常粗略、不太明朗；但到下一次財報電話會議時，我們會有更清楚的掌握。

In terms of AI generally, it's absolutely going to be -- so I think on operating. In leverage, it absolutely will be accretive people get more efficient, we'll be able to get a lot more done with the same number of people, will be able to have two or three times the impact without having to grow the team, so I'm extraordinarily excited about that.

至於 AI 整體而言，它絕對會——我想在營運面、在槓桿效益上，它絕對會帶來增益：人員會更有效率，我們能用同樣的人數完成更多事情，不需要擴編團隊就能產生兩到三倍的影響力，所以我對此非常興奮。

The mistake I see a lot of companies making is they're like, we've just said, hey, just go and be aggressive on your internal spend. I see a lot of companies that's like, oh, let's only have like a $200 budget per person for AI spend. And it's like that's nowhere near unleashes the full capabilities. I mean, you see some of our top engineers that are shipping 5 times the amount of pull requests, but they're also spending thousands of dollars a month or more on tokens. So we're figuring out exactly how to budget and price it out, but we absolutely expect that will translate to operating leverage long-term.

我看到很多公司犯的錯誤是，他們會說：嘿，就在內部支出上積極一點。我看到很多公司是：喔，我們每個人 AI 支出預算就只有 200 美元。這根本無法釋放全部能力。我的意思是，你會看到我們一些頂尖工程師交付的 pull request 數量是原來的 5 倍，但他們每月在 token 上也花了數千美元甚至更多。所以我們正在釐清到底要如何編列預算與定價，但我們絕對預期，長期來看這會轉化為營運槓桿效益。

The last thing to call out is, as part of using these products, there is inference spend. So, right, if you're using Global Agent and you're asking Amplitude, hey, find me what's the cause of this drop in this conversion funnel, like that's going to cost us a bunch of tokens and all of that. Now, for the here now, we've said, hey, let's just support it and we're going to bundle it in with our core stuff because that just means more Amplitude usage.

最後要補充的是，使用這些產品會有推論（inference）支出。也就是說，如果你在用 Global Agent，然後你問 Amplitude：嘿，幫我找出這個轉換漏斗下滑的原因——那會讓我們消耗一堆 token 等等。就目前當下，我們的做法是：先支援它，並把它跟我們的核心產品綁在一起，因為那代表 Amplitude 的使用量會增加。

And you can see that in a little bit with what Andrew shared with the gross margin numbers. And so that does short-term pressure but we again, expect that to be accretive long-term, most importantly, to revenue growth, but then also to operating margin as requires less people on our end to support more customers.

你也可以從 Andrew 分享的毛利率數字中稍微看到這一點。因此短期會有壓力，但我們再次預期長期會帶來增益——最重要的是對營收成長，其次也會對營業利益率有幫助，因為我們端需要更少的人力就能支援更多客戶。

Just one clarification to Rob, I'll tell you, we did take into consideration the operating expenses associated with the Statsig business into our guide.

再跟 Rob 澄清一點，我跟你說，我們在指引中確實已將與 Statsig 業務相關的營運費用納入考量。

Jackson Ader, KeyBanc Clark Wright, DA Davidson.

Jackson Ader，KeyBanc；Clark Wright，DA Davidson。

The first question I had Spenser on the command line interface and the MCP server that you're kind of turning live, making it frictionless, right to actually adopt and use Amplitude. But if I think about the other side of maybe it's enterprise customers where you're having forward deployed engineers, right, who are ostensibly, I would think trying to like make sure you don't go hand to hand with customers and make sure that they're adopting things.

我第一個問題是想問 Spenser，關於你們正在讓它上線的命令列介面（command line interface）和 MCP 伺服器，你們某種程度上是在把它做成即開即用、降低摩擦，讓大家能真正採用並使用 Amplitude。可是如果我從另一個角度想，可能是企業客戶那邊，你們會有前線駐點工程師（forward deployed engineers），對吧？我想他們表面上應該是在確保你們不需要跟客戶「手把手」纏鬥，同時也確保客戶在採用這些東西。

So those two things like the frictionless and the forward engineers just seem not at odds, but just like (multiple speakers) --

所以這兩件事——無摩擦（frictionless）以及前線工程師——看起來不是互相衝突，但就是有點像（多位講者）——

--(multiple speakers) you need a one-line thing. Why do you need someone to teach you?

——（多位講者）你只需要一行就好。為什麼還需要有人教你？

Okay, so I think a few things. One is that a lot of the -- by the way, I love the question. I love that you pay attention during the demos, because not everyone does. So thank you.

好，我想有幾點。第一是很多的——順帶一提，我很喜歡這個問題。我也很喜歡你在 demo 的時候有在注意，因為不是每個人都會。所以謝謝你。

So first, the frictionless, it's so much of the process before to get set up with any data system, including Amplitude was extraordinarily manual. You have to define your objectives, you have to define a taxonomy, you have to put in the SDK, you have to put one line of code wherever you do it. You have to then create a bunch of charts and dashboards and the basis of that. And so there's tremendous opportunities to automate that with AI and so that's what we did with Amplitude CLI with all of that installation process. It's actually pretty wild. I didn't -- three months ago, that wouldn't have been possible. And so it's really cool to see it possible today.

首先，所謂無摩擦，過去要把任何資料系統（包括 Amplitude）建置起來，流程都非常手動。你得定義目標、定義分類法（taxonomy）、導入 SDK、在你要放的地方加上一行程式碼。接著你還得建立一堆圖表和儀表板，並以此為基礎運作。所以這裡有非常大的機會可以用 AI 來自動化，而這就是我們用 Amplitude CLI 把整個安裝流程自動化所做的事。其實蠻不可思議的。我——三個月前這還不可能做到。能在今天看到它變成可能，真的很酷。

Now, the flip side is what we see with customers is every single one of them is looking for education on how do I adopt an AI. Arguably, you could say that all the growth in these AI natives, if you look at the private markets, comes from companies that are just doing a really good job of educating customers. It's actually no longer a technology bottleneck in that the models are getting so fast. Software is like, there's -- it's very easy to build. And so the customer is choosing on, okay, who do I trust to actually kind of get me through this? So if you're an AI native bleeding edge, hey, it's just like, give me the one line CLI and I'm off to the races.

那另一方面，我們從客戶那邊看到的是，每一個客戶都在尋求「我該如何採用 AI」的教育。你甚至可以說，如果你看私募市場，這些 AI 原生公司的所有成長，某種程度上都來自於他們非常擅長教育客戶。現在其實不再是技術瓶頸，因為模型變得非常快。軟體也就是——幾乎很容易就能做出來。所以客戶在選擇的是：好，那我該信任誰能帶我走過這段路？因此如果你是 AI 原生、走在最前沿的團隊，嘿，就像是：給我那一行 CLI，我就可以全速前進。

But if you're a traditional company, like say Fox Broadcasting or Walmart, you're going to want a lot of hands-on help from an Amplitude to make sure you educate your team. It's one thing just to have a bunch of software running, you can get that from anywhere, but it's another to say, hey, educate me on how to use analytics from a bleeding edge AI standpoint, what the future is going to be and help me rescale the hundreds or thousands of people I have with my organization. And that you just need a human touch to do.

但如果你是傳統公司，例如 Fox Broadcasting 或 Walmart，你會希望 Amplitude 提供大量的手把手協助，確保你能教育你的團隊。只是讓一堆軟體跑起來是一回事——那到處都買得到——但另一回事是：嘿，請教育我如何以最前沿 AI 的角度使用分析工具、未來會長什麼樣子，並幫我把組織內數百或數千人重新擴編／再訓練到位。這件事你就是需要人的觸感才能做到。

So it's the four deployed engineers are much like, yes, there is a -- okay, well, why do we even need that if we have the one line of code, but actually getting adoption of Amplitude or any software product within the enterprise is much less about like, do you have the, nth widget or the specific feature and it's much more about, hey are you going to be the best person out there to educate my organization on what the future of this technology is?

所以這些前線駐點工程師很像是：對，確實會有人問——好吧，如果我們有那一行程式碼，為什麼還需要這個？但其實在企業內推動 Amplitude 或任何軟體產品的採用，重點遠不在於你有第 n 個小工具或某個特定功能，而更多在於：嘿，你會不會是市場上最能教育我的組織、讓我們理解這項技術未來走向的人？

Okay. All right, that makes sense.

好。好，這樣就說得通了。

I'm just thinking like, you're shifting to an AI first company, right, which has come from a lot of personnel, or which has manifested itself in a lot of personnel changes, leadership changes, personnel, we're changing pricing and packaging, right? Now doing like an acquisition, right? Like there's integration of another of another product. So there's a lot going on. What is your plan to make sure that execution with so many balls there?

我在想的是，你們正在轉型成 AI first 的公司，對吧？這帶來了很多人員變動——或說這已經體現在很多人事調整、領導層變動、人員變動上——我們也在改變定價與包裝，對吧？現在又做了收購，對吧？等於還要整合另一個產品。所以同時有很多事情在進行。你們的計畫是什麼，來確保在這麼多球同時在空中時，仍然能把執行做好？

I mean, look, just to be very candid and direct, I think vast majority of SaaS companies are being way too conservative with the change. And I've taken the opposite approach where it's like, look, market spoken, but it's an opinion of what the future is going to look like. We know from talking to customers what they want. We see the innovations from technological standpoint. And so we want to run as fast as possible to where the puck is going on all of this stuff.

我的意思是，坦白且直接地說，我認為絕大多數 SaaS 公司在變革上都太保守了。而我採取相反的做法：你看，市場已經發聲了——當然那是對未來樣貌的一種看法。我們從與客戶的對話中知道他們想要什麼；我們也看得到技術面帶來的創新。所以我們想盡可能快地往冰球將要去的方向跑，把這些事情都做到位。

As part of that, acknowledging it is going to be bumpy and it is going to be chaotic. And there will be things that we don't expect or can't perfectly plan for in advance. It is much more important to get there with a lot of speed for a lot of different reasons than it is to say, hey, let's try to protect some existing thing we have the existing thing we have, frankly, isn't valued much. And so what's much more interesting to me is, can we generate billions of dollars in revenue in this new world?

在這個過程中，我們也承認它會顛簸、會混亂；也會有一些我們沒預期到、或無法事先完美規劃的事情。基於很多不同的原因，以很快的速度抵達那個目標，比起說「嘿，我們來保護一些既有的東西」要重要得多。老實說，我們現在擁有的既有東西，市場給的估值也不高。所以對我來說更有趣的是：我們能不能在這個新世界裡創造數十億美元的營收？

And so whether that is changes on the organization in terms of leadership, whether that's changes on functional roles. Whether it's changing on product, on pricing, on working with, partnering with OpenAI and Statsig for the future of Statsig, like we're just going to be really aggressive on making sure we reinvent the whole category.

因此，不管是組織面的領導層調整、職能角色的變更，或是產品、定價的改變，或是與 OpenAI 合作、以及與 Statsig 合作來打造 Statsig 的未來——我們都會非常積極，確保我們能重塑整個類別。

In my mind, the same thing that happened in the coding space over the last two years, where it just looks fundamentally different today than it did two years ago, that is going to happen in our category with analytics, experimentation, especially we play in the whole thing. And so it's a race to see who can be that the fastest. And so that's what I'm really focused on is not the close to $400 million in ARR that we have. I'm much more focused on the billions and potential in the future that are going to be created.

在我看來，過去兩年在程式開發（coding）領域發生的事——也就是今天看起來跟兩年前根本完全不同——同樣也會發生在我們的類別：分析、實驗（experimentation），尤其我們涵蓋整個版圖。所以這是一場比賽，看誰能最快做到。因此我真正專注的不是我們接近 4 億美元的 ARR，而是更專注於未來將被創造出的數十億美元規模與潛力。

Clark Wright, DA Davidson followed by Scott Berg, Needham.

Clark Wright，DA Davidson；接著是 Scott Berg，Needham。

Any update on the ramp of events and the pricing curve that you've implemented to help enterprises scale usage previously and ongoing?

關於你們為了協助企業擴大使用量而導入的事件（events）爬坡與定價曲線（pricing curve），不論是先前或目前進行中，有沒有最新進展？

Yeah, so one of the things that we were talking about, Clark, is that our new pricing and packaging that we'd rolled out, we did a lot of testing on. We had kind of a soft rollout this quarter. It was still one that was handheld because we hadn't implemented many of our systems related to doing the quoting or letting reps actually do the quoting themselves. That's all been now implemented and we're seeing great responses back from clients. I think that they're appreciative of the changes we've made. They see that as they add more events that they're getting a marginal incremental cost reduction from their perspective, but for us, it's always going to be increasing the ARR as events increase.

有的。Clark，我們之前談到的一點是：我們推出的新定價與包裝做了大量測試。本季算是有點「軟性上線」。當時仍然比較偏手把手，因為我們還沒把很多與報價（quoting）相關的系統建好，也還沒讓業務代表能自行報價。現在這些都已經完成導入，我們也看到客戶回饋非常好。我認為他們很感謝我們做的改變。他們看到的是：當他們增加更多 events 時，從他們的角度來看，邊際增量成本會下降；但對我們而言，隨著 events 增加，ARR 也會持續上升。

I think they like the simplicity of how they can quickly adopt the modules that are surrounding analytics. It's enterprises want cost predictability that they can align back to what their value propositions are. And as our sales team becomes more adept at showing and delivering what customers will get in value from Amplitude, I think that the pricing and packaging changes we've made will really reinforce their ability to move at pace.

我想他們也喜歡能夠快速採用分析周邊模組的那種簡潔性。企業想要的是成本可預測性，能夠回扣到他們的價值主張。隨著我們的銷售團隊越來越擅長展示並交付客戶能從 Amplitude 獲得的價值，我認為我們所做的定價與包裝調整，會大幅強化他們加速推進的能力。

That's helpful. Thank you.

這很有幫助。謝謝。

And then one of the other things that you noted during the prepared remarks was the TAM expansion with Statsig. Can you explain what budgets you're going after? I think the other piece that was consolidation that that's unlocked with this partnership. And what could you do with that? I think that you couldn't as a standalone entity.

另外，你在事先準備的談話中提到，透過 Statsig 帶來的 TAM 擴張。你能說明你們要切入哪些預算（budgets）嗎？我想另一個部分是，透過這個合作解鎖了「整合／整併（consolidation）」；你們可以用它做什麼？以及有哪些事情是你們若作為獨立個體時做不到的？

So the thing that, I mean, it's all kind of, there's overlapping, so it's not like we didn't have any of it, but they've done two things extraordinarily well. One, experimentation, the bleeding edge teams in AI are using them for experimentation. Like, I don't know if you ever use ChatGPT, but if you ever get those like, hey, do you prefer prompt A or prompt B? That is Statsig internally at OpenAI powering that. And so we're really excited that we get the opportunity to offer that out there just broadly to everyone.

我想關鍵是——其實這些都有些重疊，所以不是說我們以前完全沒有，但他們有兩件事做得特別好。第一是實驗（experimentation）：在 AI 的最前沿團隊用他們來做實驗。比如說，我不知道你有沒有用過 ChatGPT，但如果你曾經看到那種「嘿，你比較喜歡提示詞 A 還是提示詞 B？」那就是 OpenAI 內部用 Statsig 在支援這件事。所以我們非常興奮，因為我們有機會把這樣的能力更廣泛地提供給所有人。

The other thing that they've done extraordinarily well is work with the data leader and specifically their data warehouse architecture. While we obviously, we've done that a bunch at Amplitude, I mean, they're definitely the bleeding edge on that, where they actively both allow you to query on warehouses directly as well as run experiments and a whole bunch of other infrastructure. And so that's really exciting for us because especially at some of these larger -- at the largest customers, when you start getting into the multi-million dollar range, we often see this category of functionality owned by the data leader. And as part of that, we're excited to get much closer to them and unlock a lot of data warehouse and data warehouse adjacent budget.

他們另外一件做得非常出色的事情，是與資料負責人合作，特別是在他們的資料倉儲架構上。當然，我們在 Amplitude 也做過很多類似的事，但他們絕對是在這方面的最前沿：他們不僅主動讓你可以直接在資料倉儲上查詢，還能執行實驗，以及一整套其他基礎設施。因此這對我們來說非常令人振奮，因為特別是在一些更大型——在最大的客戶那裡，當你開始進入數百萬美元的規模時，我們常常看到這類功能是由資料負責人所掌握。也因此，我們很期待能與他們更緊密合作，並解鎖更多資料倉儲以及與資料倉儲相鄰的預算。

Scott Berg, Needham followed by Nick Altmann, BTIG.

Needham 的 Scott Berg，接著是 BTIG 的 Nick Altmann。

Spenser, I want to talk a kind of an architectural type question is with the pressure on gross margins, how have you thought about things like open source models or some small language models being used within the broader Amplitude platform versus maybe some of the frontier models that you're using inference and such today. There's a very large private software company that kind of announced that a large, we'll call open-source model and their new platform. It was really intriguing in terms of what they're doing with it. Love to hear what you have all considered through that process.

Spenser，我想談一個偏架構類型的問題：在毛利率承壓的情況下，你們如何看待在更廣泛的 Amplitude 平台中使用開源模型或一些小型語言模型，相較於你們今天在推論等方面使用的一些前沿模型？有一家非常大型的私人軟體公司剛宣布在他們的新平台上採用一個大型的——我們稱之為開源——模型。他們的做法很耐人尋味。很想聽聽你們在這個過程中考量了哪些事情。

Yeah, we're early on this to be clear. Inference spend is growing quite a bit. You see that reflected in a few places, both the operating margin guide and the gross margin guide. Now, ultimately, what we see from customers is, in most cases, they want the bleeding edge thing. So they want the latest sonic release or the latest Codex release from OpenAI or Anthropic or one of the other providers.

是的，先說清楚，我們在這方面還在早期階段。推論支出正在相當明顯地成長，你可以在幾個地方看到反映，包括營業利益率指引以及毛利率指引。最終我們從客戶那裡看到的是，多數情況下他們想要最前沿的東西。所以他們想要 OpenAI 或 Anthropic 或其他供應商最新的 sonic 版本，或最新的 Codex 版本。

And that also leads to higher scores on a benchmark. Like the last quarter, we published a benchmark where we got a 76% accuracy rate. And we could downgrade that in some places, but you'd probably be looking at maybe a 40% accuracy rate and that's a pretty significant difference.

而這也會帶來在基準測試上更高的分數。像上一季我們發布了一個基準測試，達到 76% 的準確率。我們當然可以在某些地方降級，但你可能就只剩下大約 40% 的準確率，這是非常顯著的差異。

Now, over the long-term, we'll obviously find places to cheaper, more effective models where it makes sense. But in general, and we'll sort out the path on gross margin as part of it. In general, right now, we're just in the, hey, like let's make sure we win the market first and foremost, and then there's optimization down the line that comes with that. So we do expect our ability either with the open source models or some of the cheaper models, like if you look at Anthropic's Haiku model, that's a really great one that for actually a good chunk of cases actually works decently well.

不過從長期來看，我們顯然會在合適的地方，找到更便宜、更有效的模型。但整體而言——而且我們也會在這個過程中把毛利率的路徑理順——目前我們基本上還是在「先確保贏得市場」的階段，之後才會有隨之而來的最佳化。所以我們確實預期，無論是透過開源模型或一些更便宜的模型都能提升我們的能力；例如你看 Anthropic 的 Haiku 模型，那就是一個很棒的選擇，對於相當一部分情境其實表現得相當不錯。

But again, when you're doing some of the complex data querying, especially with Amplitude, we see a lot of customer demand for high accuracy, and that means the most bleeding edge ones. And there's always a new one released. Now, the good news in all of this is the curve from this is crazy. I mean, you're seeing factor of 10 or more improvements in the price performance of these models year on year. And so it's hard to say exactly what it's going to be in 12 months from now, but I know, where are -- it's going to be a lot better and that means we can choose, okay, exactly where it makes sense in the price performance, so we have reasonable gross margins.

但再說一次，當你在做一些複雜的資料查詢，尤其是在 Amplitude 的情境下，我們看到客戶對高準確率有很強的需求，而這就意味著要用最前沿的模型。而且總是會有新的模型發布。好消息是，這條曲線真的很瘋狂：你看到這些模型的性價比（price performance）幾乎每年都有 10 倍或更多的提升。所以很難精確說 12 個月後會是什麼樣子，但我知道——它一定會好很多；這也代表我們可以選擇在性價比上「到底在哪裡最合理」，以便我們能維持合理的毛利率。

Understood. Thank you. Very helpful.

了解。謝謝你。非常有幫助。

And then Andrew, I want to get into the StatSig a little bit more. The good OpenAI acquired that business (multiple speakers) (technical difficulty) they were about $4 million for a year. Are you, I guess, saying or implying that the balance of the 16 that you're bringing over versus that 40 is effectively staying with OpenAI?

然後 Andrew，我想再更深入談一下 Statsig。OpenAI 收購那個業務時（多位講者）（技術問題）他們一年大約是 400 萬美元。你是不是在說或暗示，你們帶過來的 16（的部分）相對於那 40（的部分），等於是有效地留在 OpenAI 那邊？

And then, I guess, did you happen to pay for any part of this business? Did you know if there's a purchase price, cash or style, subscript allocation that's committed to this? (multiple speakers) (technical difficulty)

另外，我想問你們是否有為這個業務的任何部分付費？你是否知道是否有購買價格、現金或其他形式、以及是否有承諾的認購分攤（subscript allocation）？（多位講者）（技術問題）

Now, a couple of things you need to understand about Statsig former business prior to OpenAI acquiring them. OpenAI was a fairly large customer for them. And that was a substantial portion of their ARR. So then your question is, what is OpenAI's intention with Statsig? It's back to what Spenser mentioned earlier. They're intending to use it for internal non-commercial uses. And they're continuing to use it to support their core products.

你需要先理解幾件事：在 OpenAI 收購 Statsig 之前，Statsig 原本的業務中，OpenAI 是他們相當大的客戶，而那也占了他們 ARR 的很大一部分。所以你的問題就變成：OpenAI 對 Statsig 的意圖是什麼？這回到 Spenser 先前提到的：他們打算將其用於內部、非商業用途，並且會持續用它來支援他們的核心產品。

So as we go forward, what we've taken on is the customer contracts, all of them, and we're taking on all of the brand assets, and we're increasingly going to be developing on the product itself. So delivering great solutions for those clients and future clients, frankly, of Amplitude.

因此往後看，我們接手的是所有客戶合約，並且接手所有品牌資產，同時我們也會愈來愈多地投入產品本身的開發。也就是說，為這些客戶以及坦白說為 Amplitude 未來的客戶，交付很棒的解決方案。

Well, one minor technical point we're also talking about as a partnership, not an acquisition. So it's nuanced, but yeah, it's an important one as well.

另外補充一個小的技術點：我們談的是合作夥伴關係，而不是收購。所以這裡有些細微差別，但沒錯，這也是一個重要的點。

Nick Altmann, BTIG followed by Arjun Bhatia, William Blair.

BTIG 的 Nick Altmann，接著是 William Blair 的 Arjun Bhatia。

Andrew, we appreciate the color on the Statsig contribution, but you guys, you've kind of talked about this before, but there's overlapping customers, there's overlapping product sets. At the same time, you guys have kind of also made an effort to consolidate your customers onto more of the Amplitude products.

Andrew，我們很感謝你對 Statsig 貢獻的補充說明。不過你們之前也談過：有重疊的客戶，也有重疊的產品組合。同時，你們似乎也一直在努力把客戶整合到更多 Amplitude 的產品上。

And so in terms of that revenue contribution framework that you outlined, what does that sort of imply for those customers where there is overlap working with both of you guys in Statsig and on the product side and on the customer side? Just any other details you can kind of unpack in the assumptions would be helpful.

所以就你剛才概述的那個營收貢獻框架而言，對於那些同時與你們和 Statsig 合作、在產品端與客戶端都有重疊的客戶，這代表什麼含義？如果你能再拆解一下假設中的其他細節，會很有幫助。

We don't think -- we don't actually see that there's huge overlap in the products that customers are using from Statsig. As Spenser mentioned, they were really good on experimentation. And they may, you may find customers where we did have overlaps, but they were using analytics from Amplitude, but experimentation from Statsig. So there really isn't like overlapping revenue. It's an opportunity for us to actually add more and more to a consolidated platform. So if they didn't replay or guides and surveys. In fact, we see a huge opportunity for us to go sell into the customer base that is overlapping. Not to mention, there's a whole new group of customers that Amplitude now has access to.

我們不認為——我們其實沒有看到客戶在使用 Statsig 的產品上有很大的重疊。正如 Spenser 提到的，他們在實驗（experimentation）方面做得非常好。你可能會找到一些我們確實有重疊的客戶，但他們是用 Amplitude 做分析（analytics），用 Statsig 做實驗。所以並不存在所謂的「重疊營收」。對我們而言，這反而是一個機會，讓我們能在一個整合的平台上加入越來越多的能力。比如如果他們沒有使用 replay 或 guides 和 surveys。事實上，我們看到一個很大的機會，可以向這個有重疊的客戶群進一步銷售。更不用說，Amplitude 現在也能接觸到一整群新的客戶。

Yeah, it's not -- Nick, I'll say it's not the cleanest, like, okay, yeah, theoretically, we both have experimentation. It's been developed in a little bit of a different way. So it's a little bit of different customer set. So there's a lot of great opportunity across both customer bases. Again, we're kind of three days into this thing. So we're still sizing that and we'll have more detail when we go through on the Q2 earnings call.

是的——Nick，我會說這不是最乾淨的那種情況：理論上我們雙方都有實驗功能，但它是以稍微不同的方式開發的，所以客戶群也有些不同。因此在兩邊的客戶基礎上都有很多很好的機會。再說一次，我們才剛進行這件事大概三天，所以我們還在評估規模；等到 Q2 財報電話會時，我們會提供更多細節。

Understood. And then the NRR continues to accelerate. The ARR growth remained at 17%. So Andrew, can you just maybe unpack why we're seeing those two metrics disconnect? Is it something on the new logo ACV side of the equation? Is a gross retention dynamic? Maybe just talk to us about the disconnect between those two metrics?

了解。然後 NRR 持續加速，但 ARR 成長仍維持在 17%。Andrew，你能否拆解一下為什麼我們會看到這兩個指標出現脫鉤？是新客（new logo）ACV 端的因素嗎？還是總留存（gross retention）的動態？能否談談這兩個指標之間脫鉤的原因？

Yeah, so in any given quarter, Nick, you have an amount of new ARR we're adding in new logo versus expansion. In fact, a year ago, you probably remember, we talked about having a really strong new logo quarter and not making as much progression on that dollar retention. I would say in Q1 this quarter, you saw the opposite effect. You saw really good expansions happen. That may have been partially due to the really strong new logo quarter we saw in Q4. And in Q1, that shifted more to an expansion quarter than we anticipated.

好的，Nick，在任何一個季度，你新增的 ARR 會由新客（new logo）與擴張（expansion）兩部分構成。其實一年前你可能還記得，我們談過那時有一個非常強的新客季度，但在美元留存（dollar retention）方面沒有太多進展。我會說在今年 Q1 這一季，你看到的是相反的效果：你看到非常好的擴張發生。這可能部分是因為我們在 Q4 看到非常強的新客季度。而到了 Q1，情況就比我們預期更偏向一個擴張季度。

And every quarter, we try to take a look. Look at what is the real balance of our pipeline between new logo and expansions and do our best to estimate what the impact is going to be. It wouldn't surprise me, given what we've seen in Q1, that you might see a shift into Q2, where there is more new logo versus expansion. So that's why I was commenting that in some cases, we're seeing quarter-to-quarter progressions on our long-term plan. In other quarters, you may see it not improve as much, and it's really related to that balance. Is it overweighted in new logos in any given quarter or is it overweighted in expansion? But the long-term view is that we're continuing to ship new products and add to the value that customers can purchase from Amplitude, which sets us up for additional expansions.

而且每一季，我們都會嘗試回頭檢視一下。看看我們的銷售管線在新客（new logo）與擴張（expansion）之間的實際平衡狀況，並盡力估算其可能帶來的影響。以我們在第一季看到的情況來看，如果你在第二季看到結構出現轉移——也就是新客多於擴張——我也不會感到意外。所以我才會提到，在某些情況下，我們會看到相較上一季在長期計畫上的推進；但在其他季度，你可能會看到改善幅度沒那麼大，而這其實就與那個平衡有關：某一季是新客占比偏重，還是擴張占比偏重？但從長期來看，我們會持續推出新產品，並增加客戶可向 Amplitude 購買的價值，這也讓我們能夠帶來更多後續擴張。

Arjun Bhatia, William Blair followed by Billy Fitzsimmons, Piper Sandler.

接下來是 William Blair 的 Arjun Bhatia，之後是 Piper Sandler 的 Billy Fitzsimmons。

I want to go back to the sort of the inferencing costs increasing. And Spenser, I'm just curious where the AI usage is coming in strong. And, you've made a lot of sort of enhancements to your MCP server, and I'm curious if that's also driving a meaningful change in how your customers are using the Amplitude platform.

我想回到推論（inferencing）成本上升這件事。Spenser，我想請教目前 AI 的使用量主要是在哪些地方特別強勁？另外，你們對 MCP 伺服器做了很多強化，我也想了解這是否也在實質上改變客戶使用 Amplitude 平台的方式。

Yeah, a lot of customers are using MCP. We're seeing huge uses of that, huge uses of global agent, a good chunk on specialized agents. That's kind of the bulk. Those all kind of hook up to the same underlying services. So you can say, hey, find the root cause of an issue in this chart. And that can either come in through MCP, come in through global agent, which is the chat interface or it's come through specialized agents, which have purposely designed all of those combined or what is the vast majority of the usage of those is the vast majority of what the inference costs are.

是的，很多客戶都在使用 MCP。我們看到它的使用量非常大，global agent 的使用量也非常大，另外也有相當一部分來自 specialized agents。大致上就是這些。它們其實都連到同一套底層服務。你可以說：嘿，幫我找出這張圖表中某個問題的根因。這個請求可以透過 MCP 進來，也可以透過 global agent（也就是聊天介面）進來，或是透過 specialized agents 進來；而這些加總起來——或者說其中絕大多數的使用——就是推論成本的主要來源。

Got you.

了解。

And then just your, I mean, your comment, and I think we see it broadly as well, that the software development life cycle is changing very quickly. Obviously, it's starting in just code gen and code writing. And what is your, just, what is your perspective on the steps that will be required before you start to see it in your category, like, does the fundamental composition of the software team or the ops team need to change? And is there more change management ahead of us, I guess, before we see sort of this hockey stick in analytics and monitoring and experimentation, like, it's just more of a philosophical question, but I'm curious where we are in this cycle in your mind?

另外，你的評論——我想我們也普遍看到——軟體開發生命週期正在非常快速地改變。很明顯，現在是從程式碼生成（code gen）與寫程式開始。那麼你的觀點是，在你們這個類別開始明顯感受到之前，還需要哪些步驟？例如，軟體團隊或營運（ops）團隊的基本組成是否需要改變？以及在我們看到分析、監控與實驗出現那種「曲棍球棒式」成長之前，是否還會有更多變革管理要做？這比較偏哲學性的問題，但我很好奇你認為我們在這個循環中目前處於哪個階段？

It's early, yet to be clear, so it's hard to prognosticate on exactly when. I'll tell you the vast, so I'd say, let's talk about tech and then we'll talk about non-tech. Within tech, they're already embracing this in terms of the automated instrumentation, in terms of automated analysis, automating the product development lifecycle. You might have seen blog posts or Twitter posts about how companies are having an agent Harness and then automatically building software.

目前還很早期，先說清楚，所以很難精準預測到底會在什麼時點發生。我會說——我們先談技術公司，再談非技術公司。在技術公司這邊，他們已經在擁抱這件事了，包括自動化埋點（automated instrumentation）、自動化分析，以及把產品開發生命週期自動化。你可能看過一些部落格文章或 Twitter 貼文，提到公司用一個 agent Harness，然後自動把軟體建出來。

And so that's awesome. They're kind of already living in this future and a lot of the companies work with Granola is a great example I called out, are pushing us on that, being like, hey, here's what we want to be relevant. Agent analytics was one of the outputs of that. And we're in the early days with it. We just announced it, I think six weeks ago, something like that. We had the first post about it six weeks ago. And so we're early days in the adoption, but very exciting. A lot of these companies were living the future and now it's on us to like, okay, let's make sure we go capture it.

所以這很棒。他們某種程度上已經活在那個未來裡。我之前點名的 Granola 是個很好的例子——我們合作的很多公司都在推動我們，像是：嘿，這就是我們希望你們能保持相關性的方向。Agent analytics 就是其中一個產出。我們也還在非常早期的階段。我們大概在六週前才宣布它，差不多那個時間點；我們第一次發文也是六週前。所以採用還在早期，但非常令人興奮。很多公司已經在過未來的生活，現在就輪到我們：好，那我們要確保把這個機會抓住。

On the non-tech company side, it's much, much earlier. They're looking to get educated on the basics and, hey, show me a maturity model. Okay, if I'm at step one, how do I get to step two? And then, okay, yeah, I get like eventually you guys will help me get to step four on having an agent Harness and automatically building the software, but that might be a few years away. And instead, they're just saying, okay, well, at least let me take some of my existing workflows and speed them up a whole bunch so that I don't need to do all this instrumentation. I can just use your CLI Wizard or I can, use the chat inner global agent in the chat interface and have it generate a dashboard for me. And now, I'm saving time. So they're more focused on making a bunch of existing stuff more efficient than going straight to the bleeding edge.

在非技術公司那邊，則是早得多。他們還在學習基本概念，像是：嘿，給我看一個成熟度模型。好，如果我在第一階段，要怎麼到第二階段？然後，好的，我理解最終你們會幫我走到第四階段——擁有一個 agent Harness 並自動建置軟體——但那可能還要好幾年。相反地，他們現在只是說：好吧，至少先讓我把一些既有流程大幅加速，這樣我就不需要做那麼多埋點。我可以直接用你們的 CLI Wizard，或是在聊天介面用 global agent，讓它幫我生成一個儀表板。這樣我就省時間了。所以他們更聚焦在把既有工作變得更有效率，而不是直接衝到最前沿。

So anyway, early days, I do expect in the next two years, it will look substantially different but hard to say exactly, which quarter and when.

總之，還在早期。我確實預期在接下來兩年內，整體看起來會有顯著不同，但很難說會是在哪一個季度、什麼時點。

Billy Fitzsimmons, Piper Sandler followed by Koji Ikeda, Bank of America.

接下來是 Piper Sandler 的 Billy Fitzsimmons，之後是美國銀行（Bank of America）的 Koji Ikeda。

Hey, guys. I'll keep to one because I know we're getting close to the end of the call.

嗨，各位。我只問一題，因為我知道我們快到電話會議尾聲了。

Spenser, for you, appreciated the commentary and the prepared remarks on some of the new members of the C-suite. With Nate coming in as the Chief Commercial Officer, I want to double-click there because it seems like you're seeing some solid go-to-market progress in some of the initiatives you've already been doing, multi-product adoption. Any additional color you can provide on his plan or top priorities going in, expected changes to sales motions, sales inside apps, partner strategies, and just general changes relative to what you've kind of already done?

Spenser，謝謝你在評論與事先準備的講稿中提到一些新的 C-level 成員。Nate 以首席商務官（Chief Commercial Officer）加入後，我想深入追問一下，因為看起來你們在一些既有的市場推進（go-to-market）計畫上已經看到不錯的進展，例如多產品採用（multi-product adoption）。你能否多分享一些他上任後的計畫或最高優先事項？例如預期的銷售動作（sales motions）變化、應用內銷售（sales inside apps）、夥伴策略，以及相對於你們目前已做的事情，整體上會有哪些一般性的改變？

So yeah, let me talk about the transition, and I'll talk about going forward. So Thomas, our prior President, ran all of go-to-market, did a phenomenal job over the last 3.5 years, really upgraded us to an enterprise company (technical difficulty) (multiple speakers) before, we weren't even engaging with executives consistently, now we are, and we have all the services across the board to support them, and that's been fantastic. And you've seen that show up in terms of 100,000-plus customers, the platform adoption, a whole bunch of other metrics.

好的，我先談交接，再談未來方向。Thomas，我們之前的總裁，負責所有 go-to-market，在過去 3.5 年做得非常出色，真的把我們升級成一家企業級公司（technical difficulty）（multiple speakers）。以前我們甚至沒有持續性地與高階主管互動，現在有了；而且我們也具備全方位的服務來支援他們，這非常棒。你也看到這反映在 10 萬美元以上客戶、平台採用，以及一大堆其他指標上。

Nate actually as our Chief Commercial Officer was previously our Chief Revenue Officer (technical difficulty) (multiple speakers) and was in place reporting to Thomas for the last three years. So he's not actually, so that that part is not actually new and he was running sales. The change in his role now is he's now running the post sales motion as well as revenue operations and enablement. And so those we're thinking about, okay, how do we streamline to make sure that is seamless between the AEs and then the technical success managers and all the parts of the post-sales motion.

Nate 其實在擔任首席商務官之前，是我們的首席營收官（Chief Revenue Officer）（technical difficulty）（multiple speakers），過去三年一直在位並向 Thomas 匯報。所以這部分其實不算全新，他原本就負責銷售。他現在角色的改變在於：他也開始負責售後（post-sales）動作，以及營收營運（revenue operations）與賦能（enablement）。因此我們在思考的是：要如何精簡流程，確保從 AE 到技術成功經理（technical success managers）以及售後流程的各個環節之間能無縫衔接。

In terms of going forward, I had a slide on this briefly, but so a few different things. One, just more technical talent in post-sales generally. We've renamed it from customer success managers, gotten rid of a bunch of the extraneous roles and just call them technical success managers. We expect them to be able to educate prospects on the Amplitude product and platform, how to implement it. How to integrate it with their code base and be the go-to on AI when it comes to technical queries. We're also added for deployed engineers that can do a lot of the actual coding work. Now, it's earning the function. We just spun up that group about a month ago.

至於未來方向，我之前簡單放過一張投影片，主要有幾點。第一，整體而言在售後端增加更多技術人才。我們把名稱從 customer success managers 改掉，移除一些多餘的角色，直接稱為 technical success managers。我們期望他們能教育潛在客戶 Amplitude 的產品與平台、如何導入、如何與其程式碼基礎（code base）整合，並在技術問題上成為 AI 的主要對口。我們也新增了 deployed engineers，可以做很多實際的程式撰寫工作。這是新增的職能，我們大概一個月前才剛成立這個團隊。

But that can do like actually code and hook up like a lot of the problem with AI adoption internally within companies is not just turn on the software, it's actually hook it up to the right system. So make sure it's plugged into the back end, make sure the SDK is in, make sure the events are instrumented, make sure that the right people are getting reports. We now introduce the MCP clients, so make sure it can hook up the Slack and, Linear or Atlassian or what have you or Jira or what have you. And so that's what the four deployed engineers is actually do the coding work on behalf or alongside the customer.

因為他們可以真的去寫程式並把東西串起來——公司內部導入 AI 的很多問題不只是把軟體打開而已，而是要把它接到正確的系統上。要確保它接上後端、SDK 已經裝好、事件（events）已完成埋點、正確的人能收到報表。我們現在也導入 MCP clients，所以要確保它能串接 Slack、Linear 或 Atlassian，或 Jira 等等。因此這四位 deployed engineers 實際上就是代表客戶、或與客戶一起完成這些程式撰寫與串接工作。

So that change has been receiving. It's early, but that change has been received very well by customers. Really what they're looking for is expertise on how AI is going to change analytics as well as the associated functions within product management data and everything related.

因此，這項變更一直在推進中。現在還算早期，但這項變更已經獲得客戶非常好的回饋。他們真正想要的是專業見解：AI 將如何改變分析（analytics），以及產品管理、資料與所有相關職能將如何隨之演進。

And so, yeah, like they're really to learn. Actually, one of the funny ones is like, we've even gotten a lot of questions from customers about our AI week. It's like, oh, I want to do the same thing. How do I do it? And so, even if it's not necessarily one-to-one Amplitude related, just having people there that are able to speak about, here is how you can upskill yourself with AI, transform your organization, that's the value that they see a lot of value in it. And so we want to make sure to provide that in the post-sales motion.

所以，是的，他們真的很想學。其實有個有趣的例子是，我們甚至收到很多客戶詢問我們的 AI 週（AI week）。他們會說：「喔，我也想做一樣的活動。我該怎麼做？」因此，即使不一定是與 Amplitude 一對一直接相關，只要有人能夠分享「你可以如何用 AI 進行技能提升、如何轉型你的組織」，這就是他們認為非常有價值的地方。我們也希望確保在售後推進（post-sales motion）中提供這些內容。

Koji Ikeda, Bank of America, followed by Elizabeth Porter, Morgan Stanley.

Koji Ikeda，美國銀行（Bank of America），接著是 Elizabeth Porter，摩根士丹利（Morgan Stanley）。

Hi, this is George McGreehan, on for Koji Akita from BofA. Appreciate you guys taking our questions.

嗨，我是 George McGreehan，代表美國銀行（BofA）的 Koji Akita 發言。感謝各位回答我們的問題。

I kind of wanted to ask as a follow-up to that last question on go-to-market changes. And forgive me if I didn't hear it. Are there any sort of tweaks being made as it relates to sales incentive comp?

我想就上一個關於 go-to-market 變更的問題追問一下。若我剛剛漏聽了也請見諒。請問在銷售激勵薪酬（sales incentive comp）方面，有沒有做任何調整？

And then kind of as an unrelated follow-up, I'd be interested to hear if there's any update. I believe last quarter 25% of queries on the platform were coming from AI agents. And if there's any ups to that number, I'm sure it's growing healthily. Yeah, just kind of like what's kind of the outlook and trends there?

另外一個不太相關的追問，我想了解是否有任何更新。我記得上季平台上有 25% 的查詢（queries）是由 AI 代理（AI agents）發起的。不知道這個比例是否有上升？我相信應該健康地成長中。想請教一下那邊的展望與趨勢大概如何？

Thank you.

謝謝。

For sure. In terms of sales incentive comp, we've had incentives in place both for platform adoption as well as for multi-year, and those are continuing. I mean, we'll look at those, tweak those every quarter based on what it is we're trying to do with the business.

當然。就銷售激勵薪酬而言，我們一直都有針對平台採用（platform adoption）以及多年期（multi-year）的激勵機制，這些都會持續。我們也會根據公司當季的業務重點，每一季檢視並微調。

Right now like right this second, a lot of the priority is making sure that Statsig customers coming over are very successful and we're able to continue to serve them and help them, so we're made the broader team are very focused on that and a few specific incentives around that, but we're always tweaking that stuff, it's not like there's a major massive shift in strategy there.

就現在此時此刻而言，很多優先事項是確保轉移過來的 Statsig 客戶非常成功，我們能持續服務並協助他們。因此，我們以及更大的團隊都非常聚焦在這件事上，也有一些針對此的特定激勵；但我們一直都在調整這些機制，並不是策略上有什麼重大、巨幅的轉變。

And then in terms of the agent adoption, it continues to grow and there has been a significant increase. We're not sharing numbers on this particular call, but we will have an update next quarter on agent adoption relative to human usage of.

至於代理（agent）的採用，仍在持續成長，而且確實有顯著增加。我們在這通電話上不會分享具體數字，但我們會在下個季度更新代理採用相對於人類使用（human usage）的情況。

Elizabeth Porter, Morgan Stanley followed by YC Wong from Citi.

Elizabeth Porter，摩根士丹利（Morgan Stanley），接著是花旗（Citi）的 YC Wong。

Go ahead, Lucas.

Lucas，請開始。

I'm on for Elizabeth Porter. Thanks for taking my questions.

我代表 Elizabeth Porter。謝謝你們回答我的問題。

Spenser, you talked about seeing into the PUC as fast as possible early in the call. It's just one of the aim where your incremental AI latest investment here, whether it's infrastructure, experimenting goals, like where's the PUC going for you guys?

Spenser，你在電話會議一開始提到要盡快看到 PUC。想請教你們這次增量的 AI 最新投資（無論是基礎設施、實驗目標等）主要是為了什麼？對你們而言，PUC 的走向是什麼？

Specifically on expenses, the big one is inference costs. To be able to support global agent, MCP, specialized agents as well as some of the other AI products, we've been spending quite a bit there. Those show up under cost of goods sold. So it's growing a lot, we'll monitor it, and we'll figure out what the right long-term place for that to be and how do we value capture and get paid for it too. But right now, we just want to drive adoption as the priority.

在費用方面，最大的一項是推論成本（inference costs）。為了支援全球代理（global agent）、MCP、專用代理（specialized agents）以及其他一些 AI 產品，我們在這方面投入相當多。這些會反映在銷貨成本（cost of goods sold）之下。因此它成長很快，我們會持續監控，並找出長期而言合理的水位，以及我們要如何擷取價值並對此收費。不過目前我們的優先事項是推動採用（adoption）。

The other big place is internal tooling for the team. The big one, we're using autonomous Claude. So that was what we standard on an AI week. I think Anthropic's done an amazing job with Claude and both the chat co-work as well as Claude code in terms of integrating it with existing business systems. So we actually you hook it up to Slack, e-mail, calendar. We actually built Salesforce hasn't gotten there yet. Together yet, I'm building MCP connectors. We built their own, built a few others. And so now you can access all Amplitude data through that interface. And so that's a significant spend internally. And then we also spend some on the on the team on cursor as well.

另一個大項是團隊的內部工具（internal tooling）。最大的部分是我們在使用 autonomous Claude。那是我們在 AI 週所標準化採用的工具。我認為 Anthropic 在 Claude 上做得非常出色，無論是聊天協作（chat co-work）或是 Claude code，把它與既有的企業系統整合方面都很強。所以我們實際上把它接到 Slack、電子郵件、行事曆。我們也做了 Salesforce——Salesforce 還沒做到那一步——把它們串在一起；我正在建 MCP 連接器（MCP connectors）。我們自己做了一些，也做了幾個其他的。現在你可以透過那個介面存取所有 Amplitude 資料。因此這在內部是一筆不小的支出。另外我們也在團隊的 Cursor 上有一些投入。

There's a tail of them. Obviously, Granola is another one I mentioned on the call that is both a customer of ours and vice versa. But yeah, the inference cost is the biggest one and then, Claude for the team and then Cursor.

還有一長串工具。當然，Granola 是我在電話會議中提到的另一個，它既是我們的客戶，我們也同時是他們的客戶。不過總之，推論成本是最大項，其次是團隊使用的 Claude，然後是 Cursor。

YC Wong, Citi.

YC Wong，花旗（Citi）。

Just a quick one for Andrew, like off we just close off with AI. Last quarter, we talked about 25 AI customers that crossing over the 100 ARR mark, like if you're improving deep integration with the foundational models, the models Cursor, including your positioning Amplitude to capture a larger share of the AI market.

給 Andrew 一個簡短問題，延續我們剛剛以 AI 作結的話題。上季我們談到有 25 家 AI 客戶的 ARR 已跨過 10 萬美元門檻；若你們正在加深與基礎模型（foundational models）的深度整合、以及像 Cursor 這類模型/工具，你們如何定位 Amplitude，以在 AI 市場中取得更大的份額？

Are this AI customer exhibiting like structurally different net retention rate consumption pattern that you're seeing compared to your traditional enterprise SaaS customer? And how do you view the opportunity longer-term here? Thanks.

這些 AI 客戶是否呈現出結構性不同的淨留存率（net retention rate）或消費/用量（consumption）模式，相較於你們傳統的企業級 SaaS 客戶？你們如何看待這個長期機會？謝謝。

So certainly we believe that AI companies, as they standardize on Amplitude, will continue to see greater and greater value from using Amplitude as they embed and use it to drive marketing purchases, build better products, get better insights on how users are acting. And we're seeing, in some cases, some of the larger AI customers we have increased their data ingestion into our platform.

我們當然相信，AI 公司一旦把 Amplitude 作為標準平台，隨著他們把 Amplitude 內嵌並用來推動行銷採購、打造更好的產品、取得更好的使用者行為洞察，他們將持續看到越來越大的價值。我們也看到，在某些情況下，我們的一些較大型 AI 客戶已增加其匯入我們平台的資料量（data ingestion）。

And one of the things we talked about on the gross margin headwinds was related to greater data ingestion. I think there was a confluence of necessarily, that's all related to the AI agents, well. There's also very classic cases where customers are using more of our capabilities and they're expanding their data usage, and some of the AI companies are certainly exhibiting that. And one of the things that Statsig had in their customer base was a strong AI customer component. So we look forward to updating you all in Q2 on what that looks like with the combined products and customers.

而我們在談到毛利率（gross margin）的逆風時，其中一項就與更高的資料匯入量有關。我認為這裡有多種因素同時發生，未必全部都與 AI 代理相關；也有一些非常典型的情境是，客戶使用了更多我們的能力，並擴大其資料使用量，而部分 AI 公司確實呈現這種情況。另外，Statsig 的客戶群中也有很強的 AI 客戶組成。因此我們期待在第二季（Q2）向各位更新，當產品與客戶整合後，整體會呈現什麼樣貌。

And that will conclude our first quarter earnings call. Thank you for your time and interest. We look forward to seeing you on the road this quarter as we attend conferences hosted by Needham, Jefferies, Bank of America, and D.A. Davidson.

以上為我們第一季財報電話會議的全部內容。感謝各位的時間與關注。我們期待本季在路演行程中與各位見面，屆時我們將參加由 Needham、Jefferies、美國銀行（Bank of America）以及 D.A. Davidson 主辦的會議。