Figma, Inc. (FIG) 2026 Q2 法說會逐字稿

Hello, everyone. Thank you for joining us, and welcome to the Figma second-quarter 2026 earnings call.(Operator Instructions)

各位好。感謝各位加入，歡迎參加 Figma 2026 年第二季財報電話會議。（接線員指示）

I will now hand the conference over to Kate DeLeo, Vice President of Investor Relations. Kate, please go ahead.

接下來我將把會議交給投資人關係副總裁 Kate DeLeo。Kate，請開始。

Good afternoon, and thank you for joining us on today's conference call to discuss Figma's results for the second-quarter of 2026. On the call, we have Dylan Field, Figma's Co-Founder and Chief Executive Officer; and Praveer Melwani, our Chief Financial Officer. During the course of today's call, we may make forward-looking statements, including, but not limited to, statements regarding our guidance and future financial performance, market demand, product development, growth prospects, business strategies and plans, partnerships, ability to attract and retain customers and ability to compete effectively.

各位午安，感謝各位參加今天的電話會議，一同討論 Figma 2026 年第二季的業績表現。今天與會者包括 Figma 共同創辦人兼執行長 Dylan Field，以及我們的財務長 Praveer Melwani。在今天的會議過程中，我們可能會做出前瞻性陳述，包括但不限於與我們的財測與未來財務表現、市場需求、產品開發、成長前景、商業策略與計畫、合作夥伴關係、吸引與留住客戶的能力，以及有效競爭的能力相關之陳述。

These forward-looking statements are based on management's current views and assumptions and should not be relied upon as of any subsequent date, and we disclaim any obligation to update any forward-looking statements. Actual results may vary materially from today's statements. Information concerning our risks, uncertainties and other factors that could cause results to differ from these forward-looking statements are included in our filings with the SEC, including our quarterly report on Form 10-Q for the quarter ended June 30, 2026.

這些前瞻性陳述係基於管理層目前的觀點與假設，且不應在任何後續日期加以依賴；我們亦不承擔更新任何前瞻性陳述之義務。實際結果可能與今日陳述有重大差異。關於我們的風險、不確定性及其他可能導致結果與這些前瞻性陳述不同之因素的資訊，已載於我們向美國證券交易委員會（SEC）提交的文件中，包括截至 2026 年 6 月 30 日止季度的 Form 10-Q 季度報告。

Our discussion today will include certain non-GAAP financial measures. These non-GAAP financial measures should be considered in addition to, not as a substitute or in isolation from GAAP measures. Our non-GAAP measures exclude the effect of our GAAP results of stock-based compensation and certain other items. Reconciliations of non-GAAP financial measures to comparable GAAP measures can be found in our press release accompanying this call, which is posted to the Investor Relations page on our website.

我們今天的討論將包含若干非 GAAP 財務衡量指標。這些非 GAAP 財務衡量指標應作為 GAAP 指標的補充，而非替代，亦不應單獨使用或孤立解讀。我們的非 GAAP 指標排除了 GAAP 結果中股權基礎薪酬及若干其他項目的影響。非 GAAP 財務衡量指標與可比 GAAP 指標之調節表，可見於本次電話會議隨附的新聞稿；該新聞稿已發布於我們網站的投資人關係頁面。

I would now like to turn the conference call over to Dylan.

接下來我想把電話會議交給 Dylan。

Hi, everyone, and thanks for joining. I'm excited to share the results of another strong quarter for Figma. In Q2, we delivered $370 million in revenue, representing a year-over-year growth rate of 48% and our third consecutive quarter of accelerated growth. Q2 was also our first full quarter of AI monetization. And what we're seeing follows a pattern that is familiar to Figma, a core group of users driving outsized usage, paving the way for broader adoption across the organization.

各位好，感謝加入。我很高興與大家分享 Figma 又一個強勁季度的成果。第二季我們實現營收 3.70 億美元，年增率 48%，並且是連續第三個季度成長加速。第二季也是我們 AI 變現的第一個完整季度。我們所看到的情況延續了 Figma 熟悉的模式：由一群核心使用者帶動顯著更高的使用量，進而為組織內更廣泛的採用鋪路。

This gives us confidence in the AI consumption opportunity ahead. On the P&L front, net dollar retention rate was 136%. Non-GAAP gross profit dollars grew by 40% year-over-year, an acceleration on the previous quarter. Non-GAAP operating margin was 10%, reflecting the typical seasonal impact of Config. Free cash flow margin was 14%, and we ended Q2 with $1.7 billion in cash, cash equivalents and marketable securities. These numbers reflect incredible execution from the Figma team.

這讓我們對未來 AI 用量（consumption）的機會更有信心。在損益表方面，淨美元留存率為 136%。非 GAAP 毛利金額年增 40%，較前一季加速。非 GAAP 營業利益率為 10%，反映 Config 一貫的季節性影響。自由現金流利益率為 14%，第二季末我們持有 17 億美元的現金、約當現金及有價證券。這些數字反映了 Figma 團隊卓越的執行力。

They also show that as companies reimagine how they build products with AI, they are doubling down on Figma. This is because what Figma offers is unique, a performant professional-grade canvas where humans as well as agents can work side by side, deep product context that makes agents actually useful and full creative control through a combination of AI and direct manipulation.

這也顯示，當企業重新想像如何用 AI 打造產品時，他們正加碼投入 Figma。原因在於 Figma 提供的是獨一無二的：一個高效能、專業級的畫布，讓人類與代理（agents）能夠並肩協作；深厚的產品情境脈絡，使代理真正有用；以及透過 AI 與直接操控相結合所帶來的完整創作控制力。

These differentiators are even more valuable in a world where code is a commodity and value is moving up the stack. We see a big opportunity for Figma as we make code a primitive on our platform and become the canvas for full stack creation. Let me explain. In June, at Config, we announced Code Layers, which we plan to roll out in early access soon.

在程式碼逐漸商品化、價值往更上層堆疊移動的世界裡，這些差異化優勢更顯珍貴。當我們把程式碼變成平台上的一種基礎元件（primitive），並成為全端創作的畫布時，我們看見 Figma 的巨大機會。我來解釋一下。6 月在 Config 大會上，我們宣布了 Code Layers，並計畫很快以搶先體驗（early access）方式推出。

AI has made individuals more productive than ever. Everyone is working with their own agent, exploring their own path in their own tool. Teams are moving faster, but they're often pointed in completely different directions. Collaboration takes a back seat as tunnel vision takes over. Individuals grow more attached to the direction they've explored and less open to ideas from their team.

AI 讓個人的生產力達到前所未有的水準。每個人都在與自己的代理協作，在自己的工具裡探索自己的路徑。團隊的速度更快，但往往指向完全不同的方向。當視野變得狹隘、只顧埋頭前進時，協作就退居次要位置。個人會更執著於自己探索出的方向，也更不願意接納團隊其他人的想法。

Code Layers is designed to address these challenges. With Code Layers, interactive code lives directly on the Figma canvas. Teams can edit designs and code or manipulate them visually. They can also compare different variations of code-backed prototypes on the canvas side by side. And this makes iteration fast and collaboration the default. Code Layers is built on the same technical foundations as Figma Make.

Code Layers 的設計就是為了解決這些挑戰。透過 Code Layers，互動式程式碼可直接存在於 Figma 畫布上。團隊可以編輯設計與程式碼，或以視覺方式進行操作。他們也能在畫布上並排比較不同版本、由程式碼驅動的原型。這讓迭代變得快速，並讓協作成為預設模式。Code Layers 建立在與 Figma Make 相同的技術基礎之上。

Simply put, it's Figma Make on the canvas. And Figma Make itself is getting more powerful. For example, in May, we started rolling out the ability for teams to start working directly in the production code base with Make. Teams can go from idea to ship product without leaving Figma. 1Password uses Figma from prototype all the way to code that ships to production.

簡單來說，就是把 Figma Make 放到畫布上。而 Figma Make 本身也正變得更強大。例如在 5 月，我們開始逐步推出讓團隊能透過 Make 直接在正式（production）程式碼庫中工作的能力。團隊可以不離開 Figma，就從想法一路到產品上線。1Password 使用 Figma，從原型一路到部署到正式環境的程式碼。

Their design systems team has built a full AI-assisted prototyping pipeline with Figma Make, complete with the custom MCP and AI-powered skills, and that lets them scaffold and publish new prototypes automatically. They've also opened Figma Make access to licensed engineers. And when that process starts outside of Figma, we've also made it easier to bring your work into Figma with our MCP server.

他們的設計系統團隊以 Figma Make 建立了一套完整的 AI 輔助原型製作流程，包含自訂 MCP 與 AI 驅動的技能（skills），使他們能自動搭建並發布新的原型。他們也已將 Figma Make 的使用權開放給具授權的工程師。而當流程是在 Figma 之外開始時，我們也透過 MCP 伺服器，讓你更容易把工作帶回 Figma。

At Clay, designer Alex Fortney was tasked with redesigning the company's tools panel, a surface undergoing its fourth redesign in the last five years. Using MCP, Alex was able to expedite what would have been a tedious process by pulling all the existing components out of legacy code base and into Figma. That way, she could audit the entire surface visually and altogether.

在 Clay，設計師 Alex Fortney 被指派重新設計公司的工具面板——這個介面在過去五年內已經是第四次改版。透過 MCP，Alex 能把原本繁瑣的流程加速：將既有元件從舊有程式碼庫中全部拉出並匯入 Figma。如此一來，她就能以視覺方式、一次性地完整檢視並盤點整個介面。

As she put it, the Figma MCP has saved me countless hours of manual labor on all the design system files, and it's made it a lot easier for Clay's engineers to translate design into production-ready code. MCP is what makes this possible at scale. Write-back support lets teams push work into Figma, not just pull from it. And in Q2, MCP write-to-Figma usage grew 75% quarter-over-quarter.

她表示，Figma MCP 在所有設計系統檔案上為我省下了無數小時的手動勞動，也讓 Clay 的工程師更容易把設計轉換成可投入正式環境的程式碼。MCP 讓這一切得以在規模化下實現。回寫（write-back）支援讓團隊能把工作推送進 Figma，而不只是從 Figma 拉取。而在第二季，MCP 寫入 Figma 的使用量較前一季成長 75%。

Code Layers pushing into production with Figma Make and the Figma MCP server together will all drive more usage and credit consumption by expanding what you can do with code in Figma. As AI makes code easier to write, everything though is starting to look the same. These models are trained on what already exists.

Code Layers、透過 Figma Make 推進到正式環境，以及 Figma MCP 伺服器三者結合，將透過擴展你在 Figma 中可用程式碼完成的事情，帶動更多使用量與點數（credits）消耗。隨著 AI 讓程式碼更容易撰寫，然而一切開始看起來都差不多。這些模型是以既有內容進行訓練的。

So what you get back is in distribution. It's the expected answer. The teams that will stand out are the ones with a bold point of view that are willing to push past obvious solutions. At Config, we launched Motion and Shaders, two new expressive capabilities that used to require leaving Figma, but now are native to our canvas. And with Figma Motion, teams can build custom animations from scratch, layer those on to existing designs or ask the agent to even generate a starting point.

因此你得到的結果會落在常態分布之中。那是預期中的答案。真正能脫穎而出的團隊，是那些擁有大膽觀點、願意突破顯而易見解法的團隊。在 Config，我們推出了 Motion 與 Shaders 兩項新的表現力能力，過去需要離開 Figma 才能做到，但現在已原生於我們的畫布之上。透過 Figma Motion，團隊可以從零開始打造自訂動畫，把動畫疊加到既有設計上，甚至請代理先生成一個起始版本。

And this is all in the same canvas that their team already works in. Atlassian is a great example. The design systems team is building motion directly into the Atlassian design system, giving anyone on the team the ability to design with and generate motion themselves. This is important because Motion is something teams almost always want but rarely get because it's time-consuming and it's expensive to build.

而這一切都在同一個畫布中完成——也就是他們的團隊原本就已在其中工作的地方。Atlassian 是很好的例子。其設計系統團隊正把動效直接建入 Atlassian 設計系統，讓團隊中的任何人都能自行設計並生成動效。這很重要，因為動效幾乎是團隊總是想要、但很少真正取得的能力，原因在於它耗時且建置成本高。

As Senior Product Designer, Alexandra Pereira put it, Figma Motion turns animated illustrations from a specialist handoff into a system capability. Like Motion, Shaders make products feel live in new ways. With Shaders, anyone can describe a texture or effect like liquid glass and a Figma agent will build it for you. And then you can manipulate the output directly with fine-grain controls that give you ways to customize your results.

正如資深產品設計師 Alexandra Pereira 所說，Figma Motion 讓動畫插畫從需要專家交接的工作，變成系統本身的能力。如同 Motion，Shaders 也以全新方式讓產品更有「活著」的感覺。有了 Shaders，任何人都能描述像液態玻璃這樣的材質或效果，而 Figma 代理就會替你把它做出來。接著你還能用細緻的控制項直接操控輸出，讓你有更多方式自訂結果。

Figma Weave brings a similar approach to AI-generated media. Instead of stopping at the first prompt, you can sculpt generated outputs like Clay, connecting models and refining results until what's on the canvas is exactly what you had in your head. Weave Tools can now run inside Figma, which means teams can generate images and other visual assets without switching tools.

Figma Weave 也把類似的方法帶到 AI 生成媒體。你不必停在第一個提示詞，而是可以像捏陶土一樣雕塑生成結果，串接模型並反覆精煉，直到畫布上的內容完全符合你腦海中的想法。Weave Tools 現在也能在 Figma 內執行，這表示團隊可以在不切換工具的情況下生成圖片與其他視覺素材。

This matters more as visual assets become a bigger part of how software gets made. In addition, this also opens up Figma to new audiences we haven't served historically, in-house brand designers, creative agencies doing complex hands-on work. Taxi Studio, a UK-based brand design agency, set up a Weave workflow to generate 3D renders for design presentation with our client, Carlsberg, using three simple inputs, a beer glass, a hop leaf and also a background.

隨著視覺素材在軟體製作流程中占比愈來愈高，這點就更重要。此外，這也讓 Figma 觸及我們過去較少服務的新受眾：企業內部品牌設計師，以及從事複雜、需要大量手作的創意代理商。英國品牌設計代理商 Taxi Studio 建立了一套 Weave 工作流程，與我們的客戶 Carlsberg 合作，僅用三個簡單輸入——啤酒杯、啤酒花葉片以及背景——就能生成用於設計提案的 3D 渲染圖。

They create a starting point for brand imagery and then further refined lighting, camera angle and texture. As designer Jack Goozee put it, All this took a day, whereas it would have taken a 3D specialist weeks and tens of thousands of pounds to ideate through these elements. It is such a great way to elevate and add richness to the work while staying very much in control. And it's this control that sets Weave apart, the ability to express your creative vision exactly as you imagine it.

他們先建立品牌影像的起點，接著再進一步微調光線、鏡頭角度與材質。正如設計師 Jack Goozee 所說：這一切只花了一天；若由 3D 專家來針對這些元素進行發想，可能需要數週時間，並花費數萬英鎊。這是一種很棒的方式，能在仍然高度可控的前提下提升作品層次並增加豐富度。而正是這種「控制力」讓 Weave 與眾不同——能把你的創意願景精準表達成你所想像的樣子。

Together, Motion, Shaders and Weave give teams the tools to make work that's genuinely distinctive, not just AI generated. The market this can serve is significant. And overall, it points to a larger shift. The line between building software and making creative work is dissolving. This opens up new mediums, new possibilities and new audiences for Figma.

Motion、Shaders 與 Weave 結合在一起，讓團隊擁有工具去做出真正獨特的作品，而不只是「AI 生成」而已。它能服務的市場規模相當可觀。整體而言，這也指向一個更大的轉變。打造軟體與創作內容之間的界線正在消融。這為 Figma 開啟了新的媒介、新的可能性與新的受眾。

The third opportunity is agents. The Figma agent is built natively for design. It's fluent in Figma and increasingly powered by our own proprietary models. Because it works on the same canvas as your team, it has access to all the same tools that you do, including our new expressive capabilities like Motion and Shaders.

第三個機會是代理（agents）。Figma 代理是為設計原生打造的。它精通 Figma，並且愈來愈多由我們自有的專有模型提供能力。因為它與你的團隊在同一個畫布上運作，所以能使用你所使用的所有工具，包括我們像 Motion 與 Shaders 這些新的表現力能力。

One way to think of the agent is as a capable design intern you can hand work off to, everything from time-consuming tasks like documenting your design system to generating design variations that your team can build on. And because it works directly on the Figma canvas, you can have multiple agents running in parallel while you and your team focus on higher level work.

你可以把代理想像成一位能力很強的設計實習生，你可以把工作交給它處理——從耗時的任務（例如為你的設計系統撰寫文件）到產生設計變體，讓團隊在其基礎上繼續建構。而且因為它直接在 Figma 畫布上工作，你可以同時並行運行多個代理，讓你和團隊專注在更高層次的工作。

But the agent can do more than complete tasks. They can also build custom tools. And that's what generative plugins are for. You can describe what you need and the agent will build something that your entire team can reuse, like custom chart generators or plugins that pull in live data directly onto the canvas. The response to generative plugins has been strong.

但代理不只能完成任務。它們也能打造自訂工具。這正是生成式外掛（generative plugins）的用途。你只要描述需求，代理就能做出整個團隊都能重複使用的工具，例如自訂圖表產生器，或能把即時資料直接拉到畫布上的外掛。生成式外掛的反應相當熱烈。

As of July 31, weekly plugin creation was more than double what it was prior to the launch of generative plugins. Together, this represents a significant opportunity for our business. As the Figma agent takes on more work, AI consumption increases and the ceiling on what a team can create in Figma moves up. The Figma agent rolled out in open beta in June, and the early signs are promising.

截至 7 月 31 日，每週外掛建立數量已超過生成式外掛推出前的兩倍以上。綜合來看，這為我們的業務帶來一個重大的機會。隨著 Figma 代理承擔更多工作，AI 的使用量會增加，而團隊在 Figma 中能創造的上限也會提高。Figma 代理於 6 月以公開測試版推出，早期跡象令人鼓舞。

As of July 31, over 50% of paid customers with more than $10,000 in ARR were already using the Figma agent on a weekly basis. The agent is also expanding who uses AI in Figma, not just how much. As of July 31, more than 20% of weekly credit consuming users on paid plans were exclusively consuming credits using the Figma agent. The common thread across code, new creative capabilities and the Figma agent is that they increase the surface for AI consumption and the possibilities for what and who can create in Figma.

截至 7 月 31 日，年經常性收入（ARR）超過 10,000 美元的付費客戶中，已有超過 50% 每週都在使用 Figma 代理。代理也在擴大「誰」會在 Figma 中使用 AI，而不只是「用多少」。截至 7 月 31 日，付費方案中每週消耗點數（credits）的使用者裡，超過 20% 是只透過 Figma 代理來消耗點數。在程式碼、新的創意能力與 Figma 代理之間的共同主軸是：它們擴大了 AI 點數消耗的觸點，以及在 Figma 中「能創作什麼、由誰來創作」的可能性。

Together, they can grow our total addressable market in ways that we are only beginning to capture. More than 10 years ago, Figma introduced the Infinite Canvas as a shared space for teams to design software. The next evolution is a canvas for full stack creation, one place where anyone, anywhere can reach for whatever tool they need to build whatever they dream up, exactly as they imagine it.

合在一起，它們能以我們才剛開始掌握的方式擴大我們的總可服務市場（TAM）。十多年前，Figma 推出了無限畫布（Infinite Canvas），作為團隊設計軟體的共享空間。下一次演進，是一個用於全端創作的畫布——一個地方，讓任何人在任何地方都能取用所需的任何工具，打造他們所夢想的一切，並且精準呈現成他們所想像的樣子。

Before I close, I want to share some updates on our leadership team. First, our Chief Technology Officer, Kris Rasmussen, will become Figma's Chief Architect. After almost 10 years at Figma, Kris sees an opportunity to scale his impact by working directly on Figma's most business-critical engineering challenges, starting with Figma agent. We are kicking off the search for a new CTO.

在我結束之前，我想分享一些關於我們領導團隊的更新。首先，我們的技術長（CTO）Kris Rasmussen 將成為 Figma 的首席架構師（Chief Architect）。在 Figma 將近 10 年後，Kris 看見一個機會：透過直接投入 Figma 最攸關業務的工程挑戰來擴大他的影響力，首先就是 Figma 代理。我們也將啟動尋找新任 CTO 的流程。

And in the meantime, the engineering teams responsible for our AI and editor efforts will report directly to me. Additionally, our long-time security leader, Dev Akhawe, will become Figma's Chief Security Officer. Dev and his team were early to adopt AI in Figma's cybersecurity efforts, and I'm excited for Dev to step into this role.

在此期間，負責我們 AI 與編輯器（editor）工作的工程團隊將直接向我匯報。另外，我們長期的資安領導者 Dev Akhawe 將成為 Figma 的首席資安長（Chief Security Officer）。Dev 與他的團隊很早就把 AI 導入 Figma 的資安工作中，我也很期待 Dev 接下這個角色。

We also have two internal leadership transitions. First, after 7 incredible years, our Chief Product Officer, Yuhki Yamashita, has decided to close the Figma chapter and take extended time off. Yuhki has helped shape Figma's product and culture, and I'm so grateful for his dedication and impact. Loredana Crisan, who joined Figma as our Chief Design Officer in 2025, will expand her scope and lead Figma's product function as well. Loredana joined us almost a year ago after spending nearly a decade leading design and product at Meta for Messenger and Meta's AI efforts.

我們也有兩項內部領導職務的轉換。首先，在 7 年精彩的旅程之後，我們的產品長（CPO）Yuhki Yamashita 決定為 Figma 這一章畫下句點，並休長假。Yuhki 協助塑造了 Figma 的產品與文化，我非常感謝他的投入與影響力。Loredana Crisan 於 2025 年加入 Figma 擔任首席設計長（Chief Design Officer），她將擴大職責範圍，同時領導 Figma 的產品職能。Loredana 在加入我們之前，曾在 Meta 近十年領導 Messenger 的設計與產品，以及 Meta 的 AI 相關工作。

Her remarkable design and product aptitude and a strong point of view that she brings have already made a tremendous impact on Figma. Lastly, Sheila Vashee, our Chief Marketing Officer, will be departing Figma at the end of August. Sheila has built strong marketing foundations for Figma as we have scaled our go-to-market efforts.

她卓越的設計與產品能力，以及她帶來的鮮明觀點，已經對 Figma 產生了巨大的影響。最後，我們的行銷長（CMO）Sheila Vashee 將於 8 月底離開 Figma。在我們擴大市場推進（go-to-market）工作的過程中，Sheila 為 Figma 建立了扎實的行銷基礎。

Nairi Hourdajian, Figma's long-time Chief Communications Officer, will be our new CMO. Nairi is a leader I trust deeply. Her creativity and sharp perspective have shaped so much of our strategy in marketing and across Figma's business. I know her bold approach will further strengthen Figma's marketing and differentiate our brand.

Figma 長期的首席傳播長（Chief Communications Officer）Nairi Hourdajian 將成為我們的新任 CMO。Nairi 是我非常信任的領導者。她的創意與敏銳視角，塑造了我們在行銷以及整個 Figma 業務中許多策略。我相信她大膽的作風將進一步強化 Figma 的行銷，並讓我們的品牌更具差異化。

A year into our journey as a public company, the opportunity is bigger than even we expected. The tools are changing, the creative possibilities are expanding and Figma finds itself at the forefront of this shift. We have so much work ahead and even more left to build for our customers. I'm super excited for it, and I know the team is too.

成為上市公司的一年後，機會甚至比我們預期的還要更大。工具正在改變，創意的可能性正在擴張，而 Figma 正站在這場轉變的最前線。我們前方還有大量工作要做，也還有更多要為客戶打造的東西。我對此非常興奮，我知道團隊也是。

And with that, I'll pass it off to Praveer.

接下來，我把時間交給 Praveer。

Thanks, Dylan. As Dylan shared, our vision of becoming the canvas for full stack creation is resonating with customers. We consistently hear from our customers that they want to build more, move faster and push the boundaries of what is possible, all while elevating craft and taste. We're proud that our financial results indicate that we are delivering the right products and features to our customers as AI transforms the way products are built.

謝謝你，Dylan。正如 Dylan 分享的，我們致力於成為全端創作的畫布之願景，正與客戶產生共鳴。我們持續聽到客戶表示，他們希望打造更多、加速前進並突破可能性的邊界，同時提升工藝與品味。我們很自豪，我們的財務結果顯示，隨著 AI 改變產品打造方式，我們正為客戶提供正確的產品與功能。

Q2 was another record quarter. Importantly, it was our first full quarter in which AI credit monetization contributed to results. Q2 revenue was $370 million, up 48% year-over-year and our third straight quarter of accelerated year-over-year revenue growth. Just as importantly, that growth was accretive to gross profit in Q2. Non-GAAP gross profit for Q2 grew 40% year-over-year with our gross profit dollar growth accelerating 9 percentage points quarter-over-quarter.

Q2 再次創下季度新高。重要的是，這是 AI 點數變現首次在完整季度中對業績做出貢獻。Q2 營收為 3.70 億美元，年增 48%，且連續第三個季度年增速加快。同樣重要的是，這項成長在 Q2 對毛利具有增益效果。Q2 非 GAAP 毛利年增 40%，毛利金額的成長率較前一季加速 9 個百分點。

Let's start with AI credit monetization. As a reminder, we embed AI credits in each of our seats. Starting in mid-March, we implemented credit limits on all seats. Today, teams are able to purchase additional credits beyond these limits. A key signal we wanted to understand was how credit utilization would trend as customers began paying for incremental usage.

先從 AI 點數變現談起。提醒一下，我們在每個席位中都內嵌 AI 點數。自 3 月中旬起，我們對所有席位實施點數上限。目前，團隊可以在上限之外購買額外點數。我們想了解的一個關鍵訊號是：當客戶開始為增量使用付費時，點數使用率將如何變化。

A full quarter in, we are encouraged by what we've observed. Sustained usage is translating into revenue in two ways. First, credits included in every seat make those seats more valuable, supporting upgrades, new team conversion and retention. As of the end of Q2, approximately 2/3 of paid customers with more than $10,000 in ARR added full seats compared to their prior renewal, which is consistent with prior quarters.

在完整一季之後，我們對觀察到的情況感到鼓舞。持續使用正以兩種方式轉化為營收。第一，每個席位所包含的點數提升了席位價值，支持升級、新團隊轉換與留存。截至 Q2 末，ARR 超過 10,000 美元的付費客戶中，約有 2/3 相較於上次續約新增了完整席位，與前幾季一致。

At the same time, gross retention rate was stable in the mid- to high 90s, highlighting the mission-critical nature of our platform for our customers. Second, as customers exceed the credits built into their seats, they can purchase additional credit add-on subscriptions or enable pay-as-you-go. We're encouraged by the paid conversion we're seeing and the continued broadening of adoption across our customer base.

同時，總留存率穩定維持在 90% 中段至高段，凸顯我們平台對客戶而言具備任務關鍵性。第二，當客戶超出其席位內建點數時，可購買額外點數加購訂閱，或啟用隨用隨付。我們對目前看到的付費轉換，以及在客戶群中持續擴大的採用情況感到鼓舞。

As of the end of Q2, over 80% of paid customers with more than $10,000 in ARR were consuming AI credits weekly. We are also continuing to learn how customers want to purchase, scale and govern AI usage. We're moving quickly to iterate on our pricing model based on customer feedback. Just this week, we began rolling out user level limits, giving admins more precise control over how AI credits are allocated across their organizations.

截至 Q2 末，ARR 超過 10,000 美元的付費客戶中，超過 80% 每週都在消耗 AI 點數。我們也持續學習客戶希望如何購買、擴展與治理 AI 使用。我們正根據客戶回饋快速迭代定價模型。就在本週，我們開始推出使用者層級的限制，讓管理員能更精準地控制 AI 點數在組織內的分配。

We're also working to make it easier for users to request additional credits when they need them. There is meaningful room ahead as adoption deepens and we introduce new features and products that consume AI credits. Our products and features that are in beta and are still rolling out to customers, including Figma agent, Figma Make on local code, Motion, generative plugins and Code Layers do not currently consume paid credits. Early usage of these new products and features is trending ahead of expectations.

我們也在努力讓使用者在需要時更容易申請額外點數。隨著採用加深，以及我們推出更多會消耗 AI 點數的新功能與產品，未來仍有可觀的成長空間。目前仍在 beta、並持續向客戶推出的產品與功能，包括 Figma agent、Figma Make on local code、Motion、生成式外掛與 Code Layers，現階段尚不會消耗付費點數。這些新產品與功能的早期使用情況正高於預期。

We're already seeing that the Figma agent is drawing new users to consume AI credits as well as driving existing users to deepen their usage. As of July 31, over 50% of paid customers with more than $10,000 in ARR were already using the Figma agent on a weekly basis. Now let's turn to our key metrics. In Q2, net dollar retention rate for paid customers with more than $10,000 in ARR remained strong at 136%, even as we begin to anniversary our pricing and packaging changes from March 2025.

我們已經看到，Figma agent 正吸引新使用者來消耗 AI 點數，同時也帶動既有使用者加深使用。截至 7 月 31 日，ARR 超過 10,000 美元的付費客戶中，已有超過 50% 每週使用 Figma agent。接著來看我們的關鍵指標。Q2 中，ARR 超過 10,000 美元的付費客戶之淨美元留存率仍維持強勁的 136%，即便我們開始進入 2025 年 3 月定價與方案調整的周年期。

We are continuing to both add new customers and go deeper with our existing customers. In Q2, paid customers with more than $10,000 in ARR grew 34% year-over-year and paid customers with more than $100,000 in ARR grew 46% year-over-year. We continue to invest in our go-to-market teams to support our global customer base.

我們持續一方面新增客戶，另一方面在既有客戶中加深滲透。Q2 中，ARR 超過 10,000 美元的付費客戶年增 34%，ARR 超過 100,000 美元的付費客戶年增 46%。我們持續投資於我們的 go-to-market 團隊，以支援全球客戶群。

Last quarter, we expanded our footprint with a new office in Sao Paulo to better serve our customers in one of the fastest-growing markets. We also now offer local data hosting in Brazil after introducing data localization in Australia and India earlier this year. All of these efforts have continued to help drive our global business and international revenue, which grew 50% year-over-year in Q2.

上季，我們在聖保羅設立新辦公室以擴大據點，更好地服務這個成長最快的市場之一的客戶。在今年稍早於澳洲與印度導入資料在地化之後，我們現在也在巴西提供本地資料託管。這些努力持續推動我們的全球業務與國際營收；Q2 國際營收年增 50%。

Now let me highlight a few of our customer wins from the quarter. One of the world's largest technology companies expanded its enterprise contract with Figma in Q2, purchasing an AI credit add-on covering more than 25,000 total paid seats with the company now having more paid seats held by engineers than designers, a strong signal of developer adoption at scale.

接下來我強調幾個本季的客戶贏單。全球最大的科技公司之一在 Q2 擴大了與 Figma 的企業合約，購買了一項 AI 點數加購方案，涵蓋超過 25,000 個付費席位；該公司目前由工程師持有的付費席位數已多於設計師，這是開發者大規模採用的強烈訊號。

A technology infrastructure company expanded its AI credit add-ons multiple times within a single quarter, increasing its purchase credit commitment by five times from its first add-on as part of a company-wide push for AI native workflows across its product development org. During a month-long trial of Figma Make, a global financial institution held an internal Hackathon across its product, design and engineering teams, reducing prototype development time from a full quarter to a matter of days.

一家科技基礎設施公司在單一季度內多次擴增其 AI 點數加購，作為在產品研發組織推動 AI 原生工作流程的公司級行動之一，其點數承諾購買量相較首次加購提高了五倍。在為期一個月的 Figma Make 試用期間，一家全球金融機構在其產品、設計與工程團隊間舉辦內部黑客松，將原型開發時間從整整一季縮短到短短數天。

New admin tools for managing user credits also unblocked a path to broader adoption by giving the company greater visibility and control over AI usage across teams. Together, these productivity gains and stronger governance capabilities led the company to purchase a significant enterprise AI credit add-on subscription, while AI credit consumption increased 2.5 times quarter-over-quarter.

用於管理使用者點數的新管理工具，也透過讓公司對跨團隊 AI 使用具備更高的可視性與控制力，打通了更廣泛採用的路徑。綜合這些生產力提升與更強的治理能力，公司因此購買了一項具規模的企業級 AI 點數加購訂閱，同時 AI 點數消耗量較前一季成長 2.5 倍。

Building on a successful 5-year partnership with one of Europe's largest software providers, we identified opportunities to further accelerate adoption of our AI products. With targeted training for power users, we drove organization-wide adoption. The impact was immediate. AI credit consumption grew 2.5 times month-over-month in the first month following the engagement and a new contract.

在與歐洲最大軟體供應商之一成功合作 5 年的基礎上，我們找到了進一步加速 AI 產品採用的機會。透過針對進階使用者的定向訓練，我們推動了全組織的採用。成效立竿見影。在該次合作與新合約簽訂後的第一個月，AI 點數消耗量月增 2.5 倍。

Turning to the income statement. Unless noted, all metrics are non-GAAP. We have provided a reconciliation of GAAP to non-GAAP financials in our earnings release, which is posted to our website. Gross profit grew and gross margin improved in Q2. Gross profit was $314 million, up 40% year-over-year, and gross margin was 85%, up 2.5 percentage points quarter-over-quarter.

接著看損益表。除非另有說明，所有指標皆為非 GAAP。我們已在發布於官網的財報新聞稿中提供 GAAP 與非 GAAP 財務數據的調節表。Q2 毛利成長且毛利率改善。毛利為 3.14 億美元，年增 40%；毛利率為 85%，較前一季提升 2.5 個百分點。

The acceleration in gross profit growth and the improvement in gross margin this quarter is the result of our first full quarter of AI credit monetization. Looking ahead, we have a clear set of tools to manage inference costs as adoption scales. We route across models based on task complexity, optimize across providers through our model-agnostic architecture and are beginning to bring first-party models trained on Figma's design corpus into specific design tasks inside the Figma agent.

本季毛利成長加速與毛利率改善，源自 AI 點數變現的首個完整季度。展望未來，隨著採用規模擴大，我們有一套清晰的工具來管理推理成本。我們會依任務複雜度在不同模型間進行路由，透過模型無關的架構在供應商間最佳化，並開始將以 Figma 設計語料訓練的第一方模型導入 Figma agent 內的特定設計任務。

For the right tasks, we believe those models can deliver comparable quality at lower cost and latency and will continue to represent a key area of investment and development in the second half of the year. With the new products and features rolling out in beta post-Config, the goal remains to drive usage, retention and growth while improving quality, latency and cost ahead of GA.

我們相信，對於合適的任務，這些模型能以更低的成本與延遲提供可比的品質，並將在今年下半年持續成為投資與開發的重點領域。隨著 Config 之後以 beta 形式推出的新產品與功能陸續上線，我們的目標仍是推動使用、留存與成長，同時在正式 GA 之前持續改善品質、延遲與成本。

We do not charge our customers for their usage of products that are currently in beta, and we bear the cost of inference without offsetting consumption revenue. As a result, gross margin will vary from quarter-to-quarter in the near term. Over the long term, we expect additional usage to drive revenue and gross profit dollar growth.

我們不會向客戶收取目前仍處於 Beta 階段產品的使用費用，且我們承擔推理成本，而不以消費收入加以抵銷。因此，短期內毛利率將在各季度之間有所波動。長期而言，我們預期額外的使用量將帶動營收與毛利金額成長。

Q2 operating income was $36 million, a 10% operating margin. We hosted over 10,000 members of our community in San Francisco in Q2 for Config, our annual user conference. We view Config as an investment in our community and customers. This investment impacts our Q2 operating income and free cash flow. Inclusive of Config, our Q2 operating expenses grew at a slower rate than our topline.

第二季營業利益為 3,600 萬美元，營業利益率為 10%。第二季我們在舊金山為 Config（我們的年度用戶大會）接待了超過 10,000 名社群成員。我們將 Config 視為對社群與客戶的投資。這項投資影響了我們第二季的營業利益與自由現金流。包含 Config 在內，我們第二季營業費用的成長速度低於營收成長。

Q2 free cash flow was $53 million, a margin of 14%. The largest single driver for the year-over-year variance was costs related to our increased inference spend. The Config expenses also impacted our free cash flow. We ended the quarter with $1.7 billion in cash, cash equivalents and marketable securities, and we remain confident in the long-term cash-generating profile of the business.

第二季自由現金流為 5,300 萬美元，利潤率為 14%。年對年差異的最大單一驅動因素是我們推理支出增加所帶來的成本。Config 的費用也影響了我們的自由現金流。本季末我們持有 17 億美元的現金、約當現金與有價證券，我們仍對公司長期的現金創造能力充滿信心。

Now let's close it out with our outlook. A reminder on our guidance philosophy. We provide a snapshot of our current view based on recent trends, including what we have high confidence in and where visibility is more limited, we observe sustained trends before fully incorporating them. For the third-quarter, we expect revenue of $373 million to $375 million or 36% growth at the midpoint of the range.

接下來以我們的展望作結。先提醒一下我們的指引理念。我們會根據近期趨勢提供對當前觀點的快照，包含我們高度有信心的部分，以及能見度較有限之處；在將趨勢完全納入之前，我們會先觀察其是否具持續性。針對第三季，我們預期營收為 3.73 億至 3.75 億美元，以區間中點計算成長 36%。

For the full year, we are raising our outlook to $1.463 billion to $1.467 billion, implying 39% growth at the midpoint, a raise of $40 million. The raise is reflective of strength in AI credit consumption for products being monetized today, positive early signal on the back of our new launches, strong conversion and continued expansion.

針對全年，我們將展望上調至 14.63 億至 14.67 億美元，以中點計算意味著 39% 的成長，上調 4,000 萬美元。此次上調反映了目前已開始變現之產品的 AI 點數消耗強勁、新品推出後的正向早期訊號、強勁的轉換率以及持續的擴張。

We are maintaining our full year non-GAAP operating income guidance of $125 million to $135 million, a 9% operating margin at the midpoint. This is the right moment to lean into investment given the strong signals we see. The question we ask ourselves is whether investment in product and go-to-market increases the likelihood that Figma builds a durable advantage over the long term, even at the temporary cost of near-term margin.

我們維持全年非 GAAP 營業利益指引為 1.25 億至 1.35 億美元，以中點計算營業利益率為 9%。鑑於我們看到的強勁訊號，現在正是加大投資的適當時機。我們自問的問題是：即便短期內以暫時犧牲近期利潤率為代價，對產品與市場推進（go-to-market）的投資是否能提高 Figma 長期建立持久競爭優勢的可能性。

Over the long term, we remain focused on innovating while driving durable revenue growth and maximizing operating profit dollars. To close, Q2 was another strong quarter, our first full quarter of AI credit monetization, a deepening of our product portfolio, a broadening of our AI workflows, an acceleration of growth in revenue and gross profit dollars and continued strength in retention and expansion.

長期而言，我們仍專注於持續創新，同時推動可持續的營收成長並最大化營業利益金額。總結來說，第二季再度表現強勁，這是我們 AI 點數變現的第一個完整季度；我們的產品組合更為深化、AI 工作流程更為擴展，營收與毛利金額成長加速，留存與擴張也持續強勁。

More importantly, we are confident in how this all will compound in the quarters ahead. AI is changing how teams build software and how creative work is done, and that makes Figma more important. We are pairing product velocity with a monetization model that will evolve alongside usage and that gives us confidence in durable, profitable growth from here.

更重要的是，我們對這一切在未來幾季如何複利成長充滿信心。AI 正在改變團隊打造軟體的方式以及創意工作的完成方式，這讓 Figma 變得更重要。我們正以產品迭代速度搭配一套會隨使用量共同演進的變現模式，這使我們對未來可持續且具獲利性的成長充滿信心。

With that, I'll hand it over to the operator for Q&A.

接下來我把時間交給主持人進行問答。

(Operator Instructions) Alex Zukin, Wolfe Research.

（主持人指示）Alex Zukin，Wolfe Research。

Two quick ones for me. Dylan, given the amount of products out there right now in the marketplace from cloud design to Vercel to the vibe coding platforms, maybe just level set and remind us what are you seeing on both top of funnel and upsell dynamics and cross-sell? And why Figma seems to be winning even more in what looks like a more crowded field and the confidence that you have that, that differentiation should continue.

我這邊兩個簡短問題。Dylan，考量目前市場上從雲端設計、到 Vercel、到氛圍式（vibe）寫程式平台等各式產品的數量，能否請你先幫我們校準一下：在漏斗上端（top of funnel）、加購（upsell）動能以及交叉銷售（cross-sell）方面，你看到的是什麼？以及為什麼在看似更擁擠的競爭環境中，Figma 反而更勝一籌；你對這種差異化能持續下去的信心來源是什麼？

Absolutely. Thank you for the question. So I think that what we've seen overall is as teams explore how to integrate AI into their workflows, they go through a journey on that. And what we've continued to see is teams doubling down on Figma as they come out of that process. Now to your question, as you mentioned, there's many tools, they're solving different problems.

當然。謝謝你的提問。我認為我們整體看到的是：當團隊探索如何把 AI 整合進工作流程時，他們會經歷一段旅程。而我們持續看到的是，團隊在走完這個過程後，會更堅定地加碼使用 Figma。回到你的問題，如你所提，市面上有很多工具，它們在解決不同的問題。

Some of them are primarily coding tools. Others are more optimized for individuals or small teams. And I think it's really important to remind everyone that building software at scale is different. And the challenge isn't just generating code or making sort of design assets that anyone in the organization might want to create or bathroom signs or greeting cards.

其中一些主要是寫程式工具。另一些則更針對個人或小型團隊做最佳化。我認為很重要的一點是提醒大家：大規模打造軟體是不同的。挑戰不只是產生程式碼，或製作某個組織裡任何人都可能想做的設計素材，例如浴室標示牌或賀卡。

I think what's really important is to understand that our offering is really optimized for professional designers and it's unique. And the way we get there is a performant professional-grade canvas, one where humans and agents can work side by side, deep product context that makes those agents actually useful and very importantly, full creative control through a combination of AI, but also direct manipulation and a ton of long-tail functionality and workflow features we've really built into a professional design environment that our customers still have plenty of requests for us to improve on.

我認為真正重要的是理解：我們的產品是為專業設計師做了高度最佳化，而且具獨特性。我們之所以能做到，是因為我們提供高效能、專業級的畫布，讓人類與代理（agents）能並肩協作；提供深度的產品情境，使這些代理真正有用；更重要的是，透過 AI 與直接操作相結合，並搭配大量長尾功能與工作流程特性，讓使用者擁有完整的創作控制權——我們確實把這些能力建構在專業設計環境中，而客戶也仍有許多希望我們持續改進的需求。

And overall, you bring those capabilities together and you layer the ecosystem that we've built on top with MCP, with Code Layers coming and with Figma Make continuing to improve. And what we see when those all come together is that these tools are very good for different parts of the process, and teams continue to come back and double down on Figma to build products.

整體而言，當你把這些能力整合起來，再疊加我們所建立的生態系——包含 MCP、即將推出的 Code Layers，以及持續改進的 Figma Make——我們看到的是：這些工具在流程的不同環節各有所長，而團隊最終仍會回到並加碼使用 Figma 來打造產品。

I'll pass it over to Praveer for anything he wants to add.

我把時間交給 Praveer，看看他是否要補充。

No, the only thing I'd add, Alex, are some of the key indicators that I'm staring at. We're looking at the number of customers that are expanding at time of renewal within our 10,000-plus customers, and you saw about 2/3 of those customers adding full seats at time of renewal. Our NDR rate of 136% remains healthy and strong, and it's built on the back of continued and steady expansion, strength in gross dollar retention and us now being able to overlay our AI credit model on top. So there's a lot of tailwinds that we get the benefit of, but it starts with product and it ends with product. And I think Dylan kind of did a fantastic job there describing it.

Alex，我唯一想補充的是我正在關注的一些關鍵指標。我們在看的是：在我們超過 10,000 家客戶中，有多少客戶在續約時擴張；你也看到大約有 2/3 的客戶在續約時增加完整席次（full seats）。我們 136% 的 NDR（淨美元留存率）仍然健康且強勁，這建立在持續且穩定的擴張、強勁的總美元留存（gross dollar retention），以及我們現在能在其上疊加 AI 點數模型之上。因此我們受惠於許多順風因素，但起點是產品，終點也是產品。我認為 Dylan 剛才的描述非常到位。

Gabriela Borges, Goldman Sachs.

Gabriela Borges，高盛。

I wanted to pick both of your brain on two questions that we're getting with the stock being down in the aftermarket. Praveer, the first question is on the sequential that you're guiding to for 3Q. Maybe just give us a little bit more color. It looks much more conservative than what you guided to in 2Q and then much smaller in absolute dollars than what you guided to in 2Q. And then the flip side of this is on the gross margin.

我想就兩個問題請教你們兩位，因為盤後股價下跌，我們收到不少相關提問。Praveer，第一個問題是關於你們對第三季所給的季增（sequential）指引。能否再多給一些說明？看起來這比你們第二季給的指引保守得多，而且以絕對金額來看也比第二季指引小很多。另一方面則是關於毛利率。

I know you have a number of cohorts, a number of new products layering in, all of which will be monetized when the time is right. But maybe just level set us on how to think about what the cadence of gross margin looks like over the medium term and any kind of balance you can put around that for us?

我知道你們有多個客群（cohorts）、多項新產品持續疊加，而這些都會在適當時機進行變現。但能否請你幫我們校準一下：中期來看，毛利率的變化節奏大概會是什麼樣子？以及你能否就此提供任何可供我們掌握的區間或框架？

Yes. Thanks, Gabriela. Really good to hear from you. So our guidance philosophy here remains consistent in that we provide a snapshot of our current view based on recent trends, including what we have a high degree of confidence in and where we have -- visibility is a little bit more limited. We wait until we observe sustained trends there before fully incorporating them.

好的。謝謝你，Gabriela。很高興聽到你的提問。我們在這裡的指引理念仍然一致：我們會根據近期趨勢提供對當前觀點的快照，包含我們高度有信心的部分，以及在某些方面——能見度稍微有限之處。在將其完全納入之前，我們會等到觀察到趨勢具持續性後再行反映。

So both in Q1 and Q2, this provided us more telemetry in how AI consumption will translate into revenue, which we've been able to flow through for the back half of the year. However, as we continue to create new surfaces that are going to draw even more credits over a longer period of time in agent, in Code Layers, in Make and local code.

因此在第一季與第二季，這讓我們在 AI 使用量如何轉化為營收方面獲得了更多遙測數據，而我們也已能將其反映到今年下半年的預期中。不過，隨著我們持續打造新的介面（surfaces），在 agent、Code Layers、Make 以及本地程式碼等領域，這些介面將在更長的時間內消耗更多點數。

Today, those products are sitting in beta or early access programs and are not drawing down paid credits. And so what we've given ourselves an opportunity to do is to continue to invest to learn how those products actually mature to ensure that we're improving latency, reducing cost and improving quality before we transition those into generally available products.

目前，這些產品仍處於 beta 或搶先體驗計畫中，尚未開始扣抵付費點數。因此，我們給自己創造了一個機會，持續投入以學習這些產品實際成熟的方式，確保在轉為正式全面推出（GA）之前，我們能改善延遲、降低成本並提升品質。

And then as we make those transitions, similar to the investment trajectory that we've taken with our AI products thus far, we'll observe our ability to monetize them on the other side before fully incorporating them. And so the sort of philosophy actually also translates to our operating income in that we have an opportunity here to both drive deeper investment into a number of these newer surfaces, which is exactly what we've started to do and observe. And so that's why you're seeing deepening investment into the back half of this year is because we're really excited about the initial metrics and components there.

接著，當我們進行這些轉換時，會類似於我們迄今對 AI 產品所採取的投資軌跡：先觀察在另一端的變現能力，再把它們完整納入。因此，這樣的理念其實也會反映到我們的營業利益上：我們在這裡同時有機會加深對多個較新介面的投資——這正是我們已開始做並持續觀察的。所以你會看到今年下半年投資加深，是因為我們對那裡的初期指標與組成要素感到非常興奮。

Michael Turrin, Wells Fargo.

Michael Turrin，富國銀行。

I just want to ask a two-parter on Make, if I may. Dylan, I'm curious, given the rise in focus around open source and open weight models, if there's anything you see there that could potentially help further improve the overall position of Make as you're talking to customers. And for Praveer, we actually saw gross margin expansion this quarter. I think the 85% is a bit better than what we were expecting. So just curious if you have an updated view on whether we're reaching a local bottom in any way, given you're now monetizing credit consumption or just how to think about the improvement in gross margin there?

如果可以的話，我想就 Make 問兩個問題。Dylan，我很好奇，鑑於市場對開源與開放權重模型的關注度上升，你是否看到其中有任何可能，能在你與客戶溝通時進一步強化 Make 的整體定位。另外給 Praveer，我們這一季確實看到毛利率擴張。我認為 85% 比我們原先預期的要好一些。所以想請教你是否對於我們是否在某種程度上已觸及局部低點有更新看法——考量你們現在正對點數消耗進行變現——或是該如何看待毛利率改善的原因？

Thank you. I'll start with an answer around model and what we can see on the horizon, what we think might be possible. So I would orient everyone around three variables. One is latency, the second is quality, the third is cost. And as we work on first-party model development, which could include post trains of open source models, especially US-based open source models, you will likely see us be able to further discretize certain tasks, which could help with latency, it can help with quality and it can help with cost.

謝謝。我先從模型以及我們在地平線上看到的、我們認為可能發生的事情來回答。我會用三個變數來讓大家建立框架。第一是延遲（latency），第二是品質，第三是成本。當我們推進第一方模型開發（其中也可能包含對開源模型的後訓練，特別是以美國為基礎的開源模型）時，你們很可能會看到我們能把某些任務進一步離散化（discretize），這有助於延遲、也有助於品質，並且有助於成本。

And we think that's applicable not just to Make, it's also applicable to agent and to plenty of other surfaces across the Figma platform. And with that, I'll pass to Praveer for the second part.

我們認為這不僅適用於 Make，也適用於 agent，以及 Figma 平台上許多其他介面。接著我把第二部分交給 Praveer。

Really good to hear from you again. So the thing that we're seeing over here is actually a consistent behavior that we've observed as we've had different product launches over time. The initial -- while we're seeing widespread use of our AI products, so about 80% of our 10,000-plus customers are using -- are drawing down credits on a weekly basis.

很高興再次聽到你的聲音。我們在這裡看到的，其實是我們隨著不同產品在不同時間推出時，一直觀察到的一種一致行為。一開始——雖然我們看到 AI 產品被廣泛使用，大約有 80% 的 10,000 多家客戶每週都在使用——也就是每週都在扣抵點數。

What we are seeing is that we have an opportunity here to grow usage within those accounts, finding those initial champions and then expanding it over time. In addition, as I was mentioning with the prior question, a number of the newer products that we just rolled out either in beta or early access programs are not drawing down paid credits today.

我們看到的是：我們有機會在這些帳戶內進一步提升使用量，先找到最初的推動者（champions），再隨時間擴大。此外，如同我在上一題提到的，我們剛推出的一些較新產品目前仍在 beta 或搶先體驗計畫中，今天並不會扣抵付費點數。

And that then is an investment that we make in driving ubiquity of these newer surfaces. So the trajectory that we've taken over the past year now where we have these periods of investment that then serves as a headwind to gross margin. And then in a period where we flip on monetization, we can then start to see the acceleration in gross profit dollar growth is one that I expect in subsequent periods.

而這就成為我們在推動這些新介面普及化（ubiquity）上的一項投資。因此，我們過去一年所採取的軌跡是：會有一些投資期，這些投資期會對毛利率形成逆風。然後在我們開啟變現的期間，就會開始看到毛利（gross profit）金額成長加速——我預期在後續期間也會如此。

So I think we're now, again, in an investment period, an investment cycle as we continue to drive ubiquity of these newer products. And then my expectation over the medium to long term there is then we'll then start to be able to translate that into gross profit dollars and more durable long-term growth as well.

所以我認為我們現在再次處於一個投資期、一個投資循環，因為我們持續推動這些新產品的普及化。而我對中長期的預期是，我們將能把這轉化為毛利金額，並帶來更具韌性的長期成長。

Arjun Bhatia, William Blair.

Arjun Bhatia，William Blair。

I was -- I wanted to ask maybe a little bit just on the credit consumption of the products that are sort of drawing down these paid credits. I imagine that's mostly Make, maybe agent here to come. But as we're looking at the back half guidance, should the sequential consumption of paid credits for those surfaces continue to increase?

我想——我想問一下，關於那些正在扣抵這些付費點數的產品，其點數消耗情況。我想主要是 Make，可能之後也會有 agent。但當我們看下半年指引時，這些介面的付費點數消耗是否應該會持續呈現逐季增加？

I'm curious what you -- I know it's just one quarter in, but I'm curious what you saw in 2Q and then as we're going through July, if that trend -- if those trends accelerated or if there's any change there that you've baked into the guide?

我很好奇你們——我知道這才第一季，但我想了解你們在第二季看到了什麼，以及進入 7 月後，這個趨勢是否加速，或是否有任何變化已被你們納入指引之中？

Yes. Thanks for the question, Arjun. I think we've seen a number of different trajectories that customers take. Some were ready to make scaled purchases as soon as our AI credits came out of -- as soon as we began implementing those AI credit limits. Others started on pay-as-you-go offerings, and then they exceeded their limits there before purchasing more scaled add-ons.

是的。謝謝你的問題，Arjun。我認為我們看到客戶採取了幾種不同的軌跡。有些客戶在我們的 AI 點數從——在我們開始實施 AI 點數上限後——幾乎立刻就準備好進行規模化採購。另一些則先從隨用隨付（pay-as-you-go）的方案開始，之後在那裡超過上限，再購買更大規模的加購套件（scaled add-ons）。

And there's also a set of customers that required us to have more direct partnership with where we went into those accounts and really drove enablement side by side with the champions there and saw credit consumption multiply over those same periods prior to them actually going and purchasing these scaled add-ons.

另外也有一部分客戶需要我們更直接的合作：我們進入那些帳戶，與其中的推動者並肩推動導入與賦能（enablement），並看到在他們實際去購買這些規模化加購之前，點數消耗在同一段期間內成倍成長。

What's really interesting here is our customers are looking to us for strategic thought and direction. They're advocating for a partner. They want to be -- they want us to be a part of the conversation on how folks should be building product in this AI age. And as that starts to pick up, we fully believe that we should continue to have the opportunity to increase our credit consumption within these customer accounts as well. This is in addition to the newer surfaces that just came out in beta and early access program.

這裡非常有意思的是，我們的客戶正在向我們尋求策略性的思考與方向。他們在尋求一個合作夥伴。他們希望——希望我們能成為關於在 AI 時代人們應該如何打造產品這個對話的一部分。而隨著這開始升溫，我們完全相信，我們也將持續有機會在這些客戶帳戶內提高點數消耗。此外，還包括那些剛在 beta 與搶先體驗計畫中推出的新介面。

And you're exactly right that while they sit in beta and early access programs, they are not drawing down paid credits, but they will as they transition over into our -- into GAs over time, which then serves as a tailwind for us in the latter part of the year, early into next as well.

你說得完全沒錯：當它們仍在 beta 與搶先體驗計畫中時，並不會扣抵付費點數；但隨著時間推進、它們逐步轉為我們的——轉為正式全面推出（GA）後，就會開始扣抵，這也會在今年後段、以及明年初為我們帶來順風。

Billy Fitzsimmons, Piper Sandler.

Billy Fitzsimmons，Piper Sandler。

For Dylan, can you just contextualize how Figma's MCP usage has trended year-to-date? You mentioned it was up 75% year-over-year, but just help us think about how that tracked sequentially. Last quarter, there were some questions kind of about the general competitive environment. But talking to Figma customers, many are pulling in external AI workflows into Figma. Would be curious to the extent you can kind of quantify the magnitude of that occurring.

想請 Dylan 協助說明一下，Figma 的 MCP 使用量今年以來的趨勢如何？你提到年增 75%，但也請幫我們理解一下按季（sequentially）來看是如何變化的。上季有一些問題大概在討論整體競爭環境。但和一些 Figma 客戶交流後，很多人正把外部的 AI 工作流程拉進 Figma。如果你能的話，我也想了解這件事發生的量級大概有多大。

And if I could just ask one for Praveer. On the OpEx side, how is Figma's hiring needs kind of trended year-to-date versus maybe your initial expectations going into the year?

另外我也想問 Praveer 一題。在營業費用（OpEx）方面，Figma 今年以來的招募需求，相較於你們年初的預期，趨勢如何？

Yes, I can start, and thank you for the question. First off, I'll just quickly correct what you said there in terms of the 75% growth. It's actually quarter-over-quarter on the write MCP. So that's people that are using our MCP to get work into Figma. And so we've been thrilled to see the overall MCP growth as well as the specific growth on the write MCP and that use Figma tool call.

是的，我可以先開始，謝謝你的提問。首先，我先快速更正你剛才提到的 75% 成長的說法。實際上那是針對 write MCP 的季對季成長。也就是使用我們的 MCP 把工作匯入 Figma 的那些人。因此，我們很高興看到整體 MCP 的成長，以及 write MCP 的特定成長，還有使用 Figma 工具呼叫的成長。

Now when we look at overall the picture at MCP, we see a bunch of stuff happening. We see people pulling work from Figma to go build it elsewhere. We see people pushing work into Figma. Overall, our point of view is that we really want to make sure that wherever you start, whether it's a coding agent and then you realize, wow, there's a lot of opportunity to make the design better here or it's in Figma and you're starting with a very design forward view, that Figma is adding value to the process overall.

現在，當我們從整體來看 MCP 的全貌時，我們看到很多事情正在發生。我們看到有人把工作從 Figma 拉出去，到其他地方去建置。我們也看到有人把工作推進 Figma。整體而言，我們的觀點是，我們真的想確保無論你從哪裡開始——不管是從 coding agent 開始，然後你意識到，哇，這裡有很多機會可以把設計做得更好；或是你在 Figma 裡，以非常設計導向的視角起步——Figma 都能在整個流程中持續創造價值。

And what we want as well is to offer people ways to complete that whole workflow in Figma. But overall, yes, we're thrilled to see the growth of MCP. We'll be curious to watch the trade-off potentially between MCP and agent in the future. Definitely, as you're able to complete more of the workflow in Figma, we'll want to see if there's any change that creates in MCP, and that's something that we're watching as well.

我們也希望提供大家在 Figma 內完成整個工作流程的方法。但總體來說，是的，我們很高興看到 MCP 的成長。未來我們也會好奇觀察 MCP 與 agent 之間可能的取捨。當你能在 Figma 內完成更多工作流程時，我們會想看看這是否會對 MCP 帶來任何變化，這也是我們正在觀察的事情。

And then to your question on hiring, I think it's -- we continue to invest and build the team. But to your point, we are hiring fewer people today than we originally had planned, and that's because we've been able to augment the team that we have with AI and tools and have seen modernization of processes across the board.

接著關於你問到的招募，我認為——我們仍持續投資並擴建團隊。但如你所說，我們現在招募的人數比原先規劃的更少，原因是我們已能透過 AI 與工具來強化現有團隊，並且看到各項流程全面現代化。

So I think we've been really thoughtful now as folks are both coming in and transitioning out to make sure that the process by which that someone is coming in and operating in is the right one for the way that people should be building companies in this new age. And so even as our customers are going through this process of retooling as are we internally, and I think we're really excited about some of the levers that we've been able to find.

所以我認為，我們現在非常審慎地思考，當人員加入或轉出時，要確保他們進來後所運作的流程，是符合這個新時代打造公司的正確方式。因此，即便我們的客戶正在經歷這樣的重新工具化過程，我們內部也同樣如此；而且我想，我們對於已經找到的一些槓桿點感到非常興奮。

Rishi Jaluria, RBC.

Rishi Jaluria，RBC。

Just one for me. I'll keep it to one. As you think about some of the success that you're having with some of the newer SKUs, not just Figma Make, obviously, you've shared a lot around that, but even some of the exciting products that we saw at Config. Can you maybe walk us through not only what does the success look like in driving usage among the existing customer base, but are you seeing situations where you're actually landing net new customer logos as a result of having these additional products and what that kind of expansion motion could look like?

我這邊只有一題。我就問一題。當你們思考一些較新 SKU 的成功——不只是 Figma Make，顯然你們已分享很多——還有我們在 Config 看到的一些令人振奮的產品。你能否帶我們了解一下：這些成功不僅在於推動既有客戶群的使用量，另外你們是否也看到因為這些新增產品而拿下全新客戶 Logo 的情況？以及那種擴張動能可能會是什麼樣子？

Rishi, I really appreciate the question. I think largely, what we found is a lot of the new products here give us an opportunity to grow the number of folks that can sit on -- can sit and hold paid seats within existing paid plans. That's more within our larger customers as we go deeper and that will be evidenced and has been evidenced in our net dollar retention rate.

Rishi，我非常感謝這個問題。我想大致上，我們發現這裡許多新產品，讓我們有機會在既有的付費方案中，增加能夠——能夠坐在——能夠使用並持有付費席位的人數。這更多發生在我們較大型的客戶身上，當我們往更深處滲透時，這會、也已經反映在我們的淨美元留存率上。

I do think that there has been moments on the lower end parts of the business and the overall number of customers that are on platform, and we've disclosed this in prior quarters, we've seen an acceleration in our ability to go and acquire customers over there as well. And so we've both been able to grow the overall number of folks sitting on paid plans while also then deepening within existing.

我確實認為，在業務較低端的部分以及平台上的整體客戶數方面——我們在前幾季也揭露過——我們看到我們在那邊獲取客戶的能力有所加速。因此，我們一方面能夠增加使用付費方案的整體人數，同時也能在既有客戶內部持續加深滲透。

On agent specifically, what we found is -- and Dylan kind of -- Dylan did disclose this in our prepared remarks, that we've actually increased the overall number of weekly active credit consuming users on paid plans. And that today, folks that are actually drawing down paid credits or rather consuming credits on paid plans via agent represents about 20% of the overall. So that in and of itself is broadening who can now hold a paid seat and lowering the floor over time as well.

就 agent 而言，我們發現——而 Dylan 也在我們的事先準備講稿中提到——我們其實已提高了在付費方案中、每週活躍且會消耗點數（credit）的使用者總數。而目前，透過 agent 在付費方案中實際扣用付費點數、或者說消耗點數的使用者，約占整體的 20%。因此，這本身就在擴大現在能夠持有付費席位的人群，並且也在隨時間推移降低門檻。

Elizabeth Porter, Morgan Stanley.

Elizabeth Porter，Morgan Stanley。

I was hoping to get an update on the unit economics of AI credit revenue just after you've gone through this full quarter of monetization. So what are some of the levers, whether it's the task-based model routing, the optimization across providers or first-party models that are already starting to lower inference costs? And how we should think about those efficiencies affecting the incremental gross margin profile as usage scales?

我想在你們完成這一整季的變現之後，請你們更新一下 AI 點數收入的單位經濟。也就是說，有哪些槓桿——不論是以任務為基礎的模型路由、跨供應商的最佳化，或是第一方模型——已經開始降低推論成本？以及當使用量擴大時，我們應該如何看待這些效率對增量毛利率輪廓的影響？

Yes, we've taken a model -- sorry, a model-agnostic approach to the way that we're serving inference to our customers. And so that will continue to be a growing place of investment for us, especially as we deepen our investments on the first-party side. So today, we're able to serve an increasing share of requests that come in on agent via first-party models.

是的，我們在為客戶提供推論服務的方式上採取了——抱歉——模型不可知（model-agnostic）的做法。因此，這將持續是我們投資成長的一個領域，特別是當我們加深在第一方方面的投資時。所以今天，我們已能透過第一方模型來處理 agent 進來的請求中越來越高的占比。

And again, you then have the constant back and forth and trade-off that we're making around what is -- are we able to serve that customer and that query with lower latency, lower cost and higher quality alongside of it. And so we'll continue to make good decisions there and pull the right levers at the right moments in time. And as we've demonstrated over the course of the quarter, we have the ability here to accelerate our gross profit dollar growth as we transition more and more of our credit consuming products to being paid.

而且同樣地，我們會持續在取捨之間來回權衡：我們是否能以更低延遲、更低成本、同時更高品質來服務該客戶與該查詢。因此，我們會持續在適當的時間點做出正確決策，並拉動正確的槓桿。而且如同我們在本季所展現的，當我們把越來越多消耗點數的產品轉為付費時，我們有能力加速毛利美元的成長。

I'll just add, I think that there's so much we've done and so much more we can do when it comes to efficiency here. But again, you won't see us redo that at the expense of quality or at the expense of lower latency. We think those are also ways to drive revenue up when it comes to consumption.

我再補充一下，我認為在效率方面，我們已經做了很多，而且還有更多可以做。但同樣地，你不會看到我們為了效率而犧牲品質，或是以更高延遲為代價。我們認為，當談到用量型消費時，品質與更低延遲也同樣是推動營收上升的方法。

Samik Chatterjee, JPMorgan.

Samik Chatterjee，JPMorgan。

And maybe this is more for Praveer. Just the $40 million raise for the outlook for the year, I'm wondering how much of that is driven by the incremental credit usage that you're seeing relative to maybe other things coming in better like higher seats, et cetera? And are you now post the beta release of some of these AI products that you did at Config, are you now -- any change in thoughts in terms of how much they contribute as you're going through that beta release at this point?

也許這題比較適合 Praveer。關於全年展望上調 4,000 萬美元，我想了解其中有多少是由你們看到的額外點數使用量所驅動，相較於其他可能表現更好的因素，例如席位數更高等等？另外，在你們於 Config 釋出其中一些 AI 產品的 beta 版本之後，你們現在——在目前進行 beta 的階段——對於它們貢獻度的看法是否有任何改變？

Yes, I'll be explicit here. I think right now, at this moment, we are not taking credit for the products that are in early access programs or beta in our full year revenue outlook. I think that represents upside as we transition from these periods where the credits are not drawing down paid credits, and we transition those to GA products. And then as a result, we'll be able to take more credit for it in our revenue outlook.

是的，我在這裡說清楚。我認為就目前此刻而言，我們在全年營收展望中，並沒有把仍在早期存取計畫或 beta 的產品納入（不為其記功）。我認為，當我們從這些期間——點數尚未扣用付費點數——轉換，並把它們轉為 GA 產品時，這代表上行空間。因此，我們也就能在營收展望中為其納入更多貢獻。

Our philosophy here has always been to give you a transparent view of the things that we know and have a high degree of confidence over -- and in areas where we're still learning and/or products haven't fully transitioned to that -- to being paid, we give ourselves some opportunity to learn and share that with you when we have a higher degree of confidence.

我們在這裡的理念一直是，向各位提供我們已知且具高度信心事項的透明觀點——而在我們仍在學習、以及／或產品尚未完全轉換為付費的領域，我們會保留一些學習空間，並在我們有更高信心時再與各位分享。

Nick Altmann, BTIG.

Nick Altmann，BTIG。

I wanted to circle back on the proprietary model. And Dylan, you touched on it a little bit, but just which use cases and surfaces do you feel like the first-party model will take priority with your users? And then just given it sounds like the initial usage of your design agent is going really well, like how much of that would you attribute to your proprietary model? And do you anticipate the first-party model maybe accelerating AI usage and engagement in the near term as you expand that beyond the design agent?

我想再回到自研模型這個話題。Dylan，你剛才也稍微提到了一點，但就使用情境與觸點而言，你覺得第一方模型會優先在哪些地方為你的使用者服務？另外，既然聽起來你們的設計代理初期使用進展非常順利，你會把其中多少成效歸因於你們的自研模型？以及，當你們把它從設計代理擴展到更多場景時，你是否預期第一方模型在短期內可能會加速 AI 的使用與互動參與？

Yes. Thank you for the question. I would say that first-party models, plural, will have use cases and ways they show up across our platform. Right now, a lot of it you can think about as how do you work better with design, how do you work better with Figma and that is where we're seeing the most use cases show up. Over time, we think that will expand.

是的。謝謝你的提問。我會說第一方模型（複數）會在我們的平台上以不同的使用情境與方式呈現。目前，很多都可以理解為：如何更好地進行設計工作、如何更好地使用 Figma，而我們看到最多使用情境也正是在這裡出現。隨著時間推進，我們認為這會擴大。

At the same time, we're also still very much working with Frontier Labs, especially where customers want us to, like actually taking the design and building it and implementing it fully. That is something that I expect we will continue to work with Frontier Labs on in the immediate future, and we're grateful for those partnerships as well.

同時，我們也仍然非常積極地與 Frontier Labs 合作，特別是在客戶希望我們這麼做的地方，例如把設計真正建出來並完整落地實作。我預期在近期我們會持續在這方面與 Frontier Labs 合作，我們也很感謝這些夥伴關係。

And in general, I think that the more that we can make it so that first-party models combined with Frontier models end up at that -- the right place in that Frontier I mentioned between quality, latency and cost, the more that we can get to the right place there, the more we'll see acceleration of usage in general.

整體而言，我認為我們越能讓第一方模型與 Frontier 模型的組合，落在我提到的那個在品質、延遲與成本之間的「前沿」的最佳位置，我們越能達到那個最佳點，整體使用量就越會加速成長。

Tyler Radke, Citibank.

Tyler Radke，花旗銀行。

Maybe this one is for Praveer. So we're getting some questions. Obviously, really strong Q2 results. But as we look at the sequential guide into the Q3, can you just remind us some of the sets of assumptions you're making? It looks like sequentially kind of some of the smallest growth that you've guided to. I know there's some moving pieces with price.

這題可能是問 Praveer 的。我們收到一些問題。顯然，第二季（Q2）表現非常強勁。但當我們看第三季（Q3）的逐季指引時，你能否提醒我們你們採用的是哪些假設？看起來逐季而言，這可能是你們指引中較小的成長幅度之一。我知道價格方面有一些變動因素。

But I guess more specifically, as you think about AI becoming a larger piece of the business, obviously, usage is more volatile than subscription seats. So how are you just incorporating that mix dynamic into your guidance philosophy?

但更具體地說，當你們思考 AI 在業務中占比變得更大時，顯然使用量比訂閱席位更具波動性。那你們在指引理念上，是如何把這種組合變化納入考量的？

Yes. No, I appreciate the question, Tyler. I think maybe I'll start on the products that are and aren't included. I won't belabor it because I think I've spoken about this a couple of times during the call. The products that are in early access programs or betas that are not drawing down paid credits. We're going to wait until those transition over to GA products where we can observe how they actually monetize before taking credit for it in the guide.

是的。Tyler，我很感謝這個問題。我想我先從哪些產品被納入、哪些未被納入開始說。我不會贅述，因為我在這通電話中已經提過幾次。目前仍在早期試用（early access）或 beta 的產品，尚未開始扣抵付費點數（paid credits）。我們會等到這些產品轉為正式上市（GA）後，能觀察它們實際如何變現，再把它們的貢獻納入指引。

I think more broadly than that, the back half of this year, we start to come up against the anniversarying of the pricing changes that we made last year. So there's a couple of tougher comps there that we start to run into in Q3 and Q4. But largely, the key health indicators of the business and the places that I'm spending time and attention staring at, it's like how -- what is our ability to go and drive AI consumption across our customers. 80% of our 10,000-plus customers are consuming credits on -- 10,000-plus customers are consuming credits on a weekly basis.

更廣泛來看，今年下半年我們會遇到去年所做價格調整的周年比較基期。因此在 Q3 和 Q4 會開始面臨幾個較具挑戰的比較基期。但整體而言，我關注的業務關鍵健康指標、以及我投入時間與注意力緊盯的地方，是：我們推動客戶 AI 消費（consumption）的能力如何。超過 10,000 家客戶中，有 80% 每週都在消耗點數；超過 10,000 家客戶每週都在消耗點數。

What is our ability to go and add full seats at time of renewal? Within our 10,000 plus customers, 2/3 of our customers there added full seats at time of renewal, which is consistent with what we've observed in prior quarters. We're both being able to go deeper within these customers, go broader and then you then have the overlay of what can come in a number of the products that will roll out over time.

我們在續約時增加完整席位（full seats）的能力如何？在我們超過 10,000 家客戶中，有 2/3 的客戶在續約時增加了完整席位，這與我們在前幾季觀察到的情況一致。我們既能在這些客戶內部做得更深、也能做得更廣，此外還有一層因素是：隨著時間推移，會有多項產品陸續推出所帶來的貢獻。

And we do have some upside to the plan that should we execute, we can achieve. But today, we want to give you guys a true view of what we know. And then lastly, just to give you a flavoring of how our AI consumption revenue is actually translating or rather is being contracted. The majority of the structures today are via add-ons that are coterminous subscriptions with an individual subscription.

而且我們的計畫也有一些上行空間，只要我們執行到位，就有機會達成。但就今天而言，我們希望讓各位看到我們所掌握資訊的真實樣貌。最後，補充一下我們的 AI 消費收入實際如何轉化，或者更準確地說，是如何簽約。目前大多數的結構是透過加購（add-ons），並且是與個別訂閱同到期（coterminous）的訂閱。

So we have a fair amount of visibility in how those will translate over time. But you're exactly right that if more and more of that were to move to pay-as-you-go or if we end up with more extended contracting structures, we could see more variability and when that gets recognized.

因此我們對它們隨時間推移如何轉化有相當程度的能見度。但你說得完全正確：如果越來越多轉向隨用隨付（pay-as-you-go），或如果我們採用更長期/更延伸的合約結構，我們可能會看到更大的波動，以及收入認列時間點的差異。

Jack McShane, Stifel.

Jack McShane，Stifel。

This is Jack McShane on for Parker. Praveer, you mentioned in your prepared remarks and in this last question, customers looking to -- or are you guys looking to iterate your pricing model based on customer feedback. Not sure you'll provide much on future pricing plans, but can you provide any color on what feedback around pricing has been from your customers? Are customers looking for more certainty around AI costs? Or are they becoming increasingly comfortable with paying consumption?

我是代替 Parker 提問的 Jack McShane。Praveer，你在事先準備的講稿以及上一題中提到，客戶希望——或你們希望——根據客戶回饋來迭代你們的定價模型。我不確定你是否會透露太多未來的定價計畫，但你能否分享一些客戶對定價的回饋？客戶是否希望對 AI 成本有更高的確定性？或者他們是否越來越能接受按用量付費（consumption）？

Yes. I think at the very basic level, and I see this even as a purchaser of a number of different AI tools is your customers want control and choice. They want to understand that their investment here is driving return, and we need to be able to both provide our customers with that same level of choice and control while also then being able to be really clear with the ROI that they're seeing on the other side of it.

是的。我認為在最基本的層面——我自己作為多種 AI 工具的採購者也有同樣感受——客戶想要的是控制權與選擇權。他們希望理解他們在這裡的投入能帶來回報，而我們需要在提供同等程度的選擇與控制的同時，也能非常清楚地呈現他們在另一端看到的投資報酬率（ROI）。

I think that from a true ROI perspective, we're actually seeing some really interesting early indicators. Folks that are relying on Code Connect and then also using our MCP are seeing that the overall number of tokens consumed on the other side of it as they're translating it to a code it will be that much more efficient. We're finding that folks like the ability to both lead with AI in certain places, but then have the full creative control on the canvas. And so not each action needs to be credit consuming as a result.

從真正的 ROI 角度來看，我們其實看到一些非常有意思的早期指標。依賴 Code Connect、同時也使用我們 MCP 的使用者，在把內容轉換成程式碼的過程中，另一端所消耗的 token 總量會更有效率。我們發現大家喜歡在某些環節以 AI 先行，但在畫布（canvas）上仍保有完整的創作控制權。因此，不是每一個動作都必然需要消耗點數。

There are things that we're rolling out on the actual pricing and packaging side that give folks the ability to set user level controls, the ability to set restrictions in certain places. I think we want -- we hear from customers that they want to have the ability to draw down credits over longer periods of time. And as we think about the right ones and the right changes to make to our model, again, to provide customers with the control that they want and choice that they want, my expectation is that, that will start to break down even more barriers and make the sales process that much more efficient over time.

我們正在定價與方案包裝（packaging）面推出一些功能，讓使用者能設定使用者層級的控制、也能在某些地方設定限制。我認為我們從客戶那裡聽到的是：他們希望能在更長的期間內扣抵/使用點數。當我們思考要對模型做出哪些正確的調整與變更，以提供客戶想要的控制與選擇時，我預期這將進一步打破更多障礙，並讓銷售流程隨時間推移變得更有效率。

I'll just add before we end here that we are learning a lot and so is the market generally as it comes to purchasing software on different ways that AI credits and consumption should be purchased. And so as we continue to learn, we'll continue to improve. And I think that will be something that continues to happen over the long term because I think we're still in the early days of what people want to see here and how they'll express that.

在結束前我再補充一點：在以不同方式採購軟體、以及 AI 點數與用量應該如何購買這件事上，我們正在學習，市場整體也在學習。因此，隨著我們持續學習，我們也會持續改進。我認為這會是一件長期持續發生的事，因為我認為大家仍處於早期階段，還在摸索他們希望看到什麼，以及他們會如何表達這些需求。

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

目前沒有其他問題。今天的電話會議到此結束。感謝各位參與。您現在可以掛線。