Arista Networks Inc (ANET) 2025 Q3 法說會逐字稿

Welcome to the Third Quarter 2025 Arista Networks financial results earnings conference call. (Operator Instructions) As a reminder, this conference is being recorded and will be available for replay from the Investor Relations section on the Arista website following this call.

歡迎參加 Arista Networks 2025 年第三季財務業績財報電話會議。（操作員說明）提醒各位，本次會議正在錄音，會議結束後，您可以在 Arista 網站的投資者關係部分收聽錄音回放。

Mr. Rudolph Araujo, Arista's Head of Investor Advocacy. You may begin.

魯道夫·阿勞霍先生，阿里斯塔投資者權益倡導主管。你可以開始了。

Thank you, Christa. Good afternoon, everyone, and thank you for joining us. With me on today's call are Jayshree Ullal, Arista Networks' Chairperson and Chief Executive Officer; and Chantelle Breithaupt, Chief Financial Officer. This afternoon, Arista Networks issued a press release announcing the results for the fiscal third quarter ending September 30, 2025. If you want a copy of the release, you can access it online on our website.

謝謝你，克麗斯塔。各位下午好，感謝各位的參與。今天和我一起參加電話會議的有 Arista Networks 董事長兼執行長 Jayshree Ullal 和財務長 Chantelle Breithaupt。今天下午，Arista Networks 發布新聞稿，公佈了截至 2025 年 9 月 30 日的第三財季業績。如果您想要新聞稿副本，可以在我們的網站上線上取得。

During the course of this conference call, Arista Networks management will make forward-looking statements, including those relating to our financial outlook for the fourth quarter of the 2025 fiscal year longer-term business model and financial outlook for 2026 and beyond, our total addressable market and strategy for addressing these market opportunities including AI, customer demand trends, tariffs and trade restrictions, supply chain constraints, component costs, manufacturing output, inventory management and inflationary pressures on our business, lead times, product innovation, working capital optimization and the benefits of acquisitions which are subject to the risks and uncertainties that we discuss in detail in our documents filed with the SEC, specifically in our most recent Form 10-Q and Form 10-K and which could cause actual results to differ materially from those anticipated by these statements.

在本次電話會議期間，Arista Networks 管理層將發表前瞻性聲明，包括與我們對 2025 財年第四季度財務展望、2026 年及以後的長期業務模式和財務展望、我們的總潛在市場和應對這些市場機遇的戰略（包括人工智能）、客戶需求趨勢、關稅和貿易限制、供應鏈限制、組件成本、製造成本、公司成本、新成本這些聲明受到我們在提交給美國證券交易委員會 (SEC) 的文件中詳細討論的風險和不確定性的影響，尤其是在我們最新的 10-Q 表格和 10-K 表格中，這些風險和不確定性可能導致實際結果與這些聲明中預期的結果存在重大差異。

These forward-looking statements apply as of today and you should not rely on them as representing our views in the future. We undertake no obligation to update these statements after this call. This analysis of our Q3 results and our guidance for Q4 2025 is based on non-GAAP and excludes all non-cash stock-based compensation impacts, certain acquisition required charges and other non-recurring items. A full reconciliation of our selected GAAP to non-GAAP results is provided in our earnings release.

這些前瞻性陳述僅代表截至今日的觀點，您不應將其視為我們未來觀點的依據。我們不承擔在本次通話後更新這些聲明的義務。本次對我們第三季業績的分析以及 2025 年第四季的展望均基於非公認會計準則，不包括所有非現金股票選擇權補償影響、某些收購所需費用和其他非經常性項目。我們在獲利報告中提供了所選GAAP與非GAAP結果的完整調節表。

With that, I will turn the call over to Jayshree.

接下來，我將把電話交給傑伊什裡。

Thank you, everyone, for joining us this afternoon on our third quarter 2025 earnings call. Arista continues to drive its 19th consecutive record quarter of growth in this AI era. We achieved almost $2.31 billion this quarter with software and services attributing approximately 18.7% of revenue. Our non-GAAP gross margin of 65.2% was influenced by favorable mix and inventory benefits. Americas was strong at almost 80% and international at approximately 20%.

感謝各位今天下午參加我們2025年第三季財報電話會議。在人工智慧時代，Arista 繼續保持連續第 19 個季度創下成長紀錄。本季我們實現了近 23.1 億美元的收入，其中軟體和服務收入約佔總收入的 18.7%。我們的非GAAP毛利率為65.2%，這得益於有利的產品組合和庫存收益。美洲市場佔比接近 80%，國際市場佔比約 20%。

On September 11 at our Analyst Day, we showcased both networking for AI and AI for network working with our continued momentum across our data-driven network platforms. Unlike many others, our either length portfolio highlights our accelerated networking approach, bringing a single point of network control for zero-touch automation, trusted security traffic engineering and telemetry to dramatically improve compute and GPU utilization. Superior AI networks from Arista improves the performance of AI accelerators.

9 月 11 日，在我們的分析師日上，我們展示了人工智慧網路和人工智慧在網路工作中的應用，以及我們在數據驅動型網路平台上的持續發展勢頭。與許多其他產品不同，我們的任意長度產品組合突出了我們加速的網路方法，為零接觸自動化、可信任安全流量工程和遙測提供了單一的網路控制點，從而顯著提高了運算和 GPU 利用率。Arista 的卓越 AI 網路可提升 AI 加速器的效能。

Of course, we interoperate with NVIDIA, the worldwide market leader in GPUs, but we also recognize our responsibility to create a broad and open ecosystem, including AMD, Anthropic, ARMM, Broadcom, Open AI, Pure Storage and VAST Data to name a few, and build that modern AI stack of the 21st century. This stack includes the trial of compute memory storage and a solid network foundation to run training and inference models.

當然，我們與全球 GPU 市場領導者 NVIDIA 進行互通，但我們也意​​識到我們有責任創建一個廣泛開放的生態系統，其中包括 AMD、Anthropic、ARMM、Broadcom、Open AI、Pure Storage 和 VAST Data 等，並建立 21 世紀的現代 AI 技術堆疊。此堆疊包括計算記憶體儲存試驗和用於運行訓練和推理模型的堅實網路基礎。

Our stated goal of $1.5 billion AI aggregate for 2025, comprising of both back end and front end is well underway. We are now committed to $2.75 billion out of our new target of $10.65 billion in revenue, representing 20% revenue growth in 2026. We are experiencing momentum across cloud and AI titans, neo cloud providers and the campus enterprise. The demand and scale of AI buildouts is clearly unprecedented, as we look to move data faster across multiplanar networks.

我們設定的 2025 年人工智慧總投資額達到 15 億美元的目標（包括後端和前端）正在順利實現。我們現在致力於實現 2026 年營收目標 106.5 億美元中的 27.5 億美元，這意味著營收成長 20%。我們看到雲端運算和人工智慧巨頭、新興雲端供應商以及校園企業都迎來了強勁的發展勢頭。隨著我們尋求在多平面網路中更快地傳輸數據，人工智慧建設的需求和規模顯然是前所未有的。

People and leadership are key to our success. And to that end, we announced Todd Nightingale as our President and Chief Operating Officer last quarter. This time, we want to celebrate the promotion of Ken Duda, our President and Chief Technology Officer, not only of engineering, but our top AI and cloud segment of customers as well. Ken, as many of you know, has been a champion of architecture, innovation and culture since founding Arista over 20 years ago.

人才和領導力是我們成功的關鍵。為此，我們在上個季度宣布任命 Todd Nightingale 為公司總裁兼營運長。這次，我們要慶祝 Ken Duda 的晉升，他不僅是我們的工程總裁兼首席技術官，也是我們頂級人工智慧和雲端客戶群的負責人。正如你們許多人所知，自 20 多年前創立 Arista 以來，Ken 一直是建築、創新和文化的倡導者。

Ken, would you like to say a few words?

肯，你想說幾句話嗎？

Thanks, Jayshree. I would like to one of the best things about working at Arista is getting to build some of the most ambitious networks ever built, ultra low latency trading networks, global scale, our networks having most recently multi-benefit AI networks.

謝謝你，Jayshree。在 Arista 工作最棒的事情之一就是能夠建立一些有史以來最具雄心的網絡，超低延遲交易網絡，全球規模，我們最新的網絡還擁有多效益人工智慧網絡。

Our success in AI has many sources, the sheer power and performance of our hardware platforms, our innovations in fabric architecture, our AI-focused telemetry and provisioning automation are reputation for the highest quality software, found our leadership in the Ultra Ethernet Consortium the UVC and our work in Ethernet scale of networking or E.SUN, and most importantly, the way we partner with the world's largest AI companies, partnership has been key to our success over and over at Arista and the AI revolution is no exception.

我們在架構設計方面的成功源於多方面因素：我們硬體平台的強大功能和卓越性能、我們在架構設計方面的創新、我們以人工智慧為中心的遙測和配置自動化技術，以及我們高品質的軟體聲譽；我們在超以太網聯盟 (UVC) 中的領導地位，以及我們在以太網規模化網絡 (E.SUN) 關係方面的工作；合作關係一直是 Arista 一次又一次成功的關鍵，人工智慧革命也不例外。

In addition to being a lot of fun, these partnerships benefit our company, both through the sheer revenue opportunity but also in providing us with the opportunity to learn and innovate at the edge of what's possible. We can then apply what we've learned to bring solutions to the broader netting market, helping a much larger and more diverse customer base, build the most advanced and reliable infrastructure in the industry.

除了帶來許多樂趣之外，這些合作關係也使我們公司受益，不僅帶來了巨大的收入機會，而且還為我們提供了在可能性的前沿學習和創新的機會。然後，我們可以運用所學到的知識，為更廣泛的漁網市場帶來解決方案，幫助更大、更多樣化的客戶群體，建構業內最先進、最可靠的基礎設施。

For example, our Etherlink distributed switch to rock, power some of the largest AI fabrics in the world. It's also an excellent underway for data centers of whole sorts, providing a full line rate fabric with no hotspots at petabyte scale for all workloads, including AI. Etherlink speeds are going from 800 gigabits today to 1.6 terabits in the near future while leveraging our EOS operating system and our NDI diagnostics infrastructure for top hardware and software reliability.

例如，我們的 Etherlink 分散式交換器為世界上一些最大的 AI 架構提供動力。它也為各種類型的資料中心提供了極佳的開端，為包括人工智慧在內的所有工作負載提供 PB 級全線速架構，且無熱點。Etherlink 的速度將從目前的 800 吉比特提高到不久的將來的 1.6 太比特，同時利用我們的 EOS 作業系統和 NDI 診斷基礎設施來實現頂級的硬體和軟體可靠性。

Arista AVA or Autonomous Virtual Assist AI to help our customers design, build and operate their networks. AVA draws on both our internal knowledge base and also on the customers' data stored in NetDL, Arista's network data lake. Plus AVA has agentic capabilities to help troubleshoot proactively. Other recent innovations include Swab, a switch aggregation technology that provides the features of campus stacking along with fault containment and in-service software upgrades for maximum uptime.

Arista AVA（自主虛擬助理人工智慧）可以幫助我們的客戶設計、建構和營運他們的網路。AVA 既利用了我們的內部知識庫，也利用了儲存在 Arista 網路資料湖 NetDL 中的客戶資料。此外，AVA 還具備代理功能，可主動協助進行故障排除。近期其他創新包括 Swab，這是一種交換器聚合技術，它提供園區堆疊功能以及故障隔離和線上軟體升級功能，從而最大限度地延長正常運行時間。

By running a common EOS and common net DI platform across so many use cases, we are able to maintain alignment between our different market segments, leveraging central engineering investments efficiently as we pursue cloud, enterprise and AI markets simultaneously. I am so grateful for the opportunity to lead the Arista engineering and cloud teams in an era with so many exciting opportunities.

透過在眾多用例中運行通用的 EOS 和通用的網路 DI 平台，我們能夠保持不同市場區隔之間的一致性，在同時開拓雲端、企業和 AI 市場的同時，有效地利用中央工程投資。我非常榮幸有機會在這個充滿令人興奮的時代領導 Arista 的工程和雲端團隊。

Thank you, Ken, and congratulations on a fantastic 21-year career, a very well-deserved promotion at Arista. You have always built the always-on resilient leaf-spine architecture, both now for networking for AI workloads, and AVA to bring AI to networking. At Oracle AI world, Ken was invited to formally announce our operation with Oracle Exelon. This builds upon a decade of partnership with Oracle starting with our Exadata migration from InfiniBand to Ethernet for AI networks to Rocky RDMA over converged Ethernet and now multiplanar networking across cloud AI for on-time job completion and gigawatt scale AI data centers.

謝謝你，肯，恭喜你擁有了輝煌的 21 年職業生涯，在 Arista 獲得晉升實至名歸。您一直以來都建立了始終在線的彈性葉脊架構，現在既用於 AI 工作負載的網絡，也用於將 AI 引入網絡的 AVA。在 Oracle AI 世界大會上，Ken 受邀正式宣布我們與 Oracle Exelon 的合作。這建立在我們與 Oracle 長達十年的合作關係之上，從我們 Exadata 從 InfiniBand 遷移到以太網以實現 AI 網絡，到 Rocky RDMA 通過融合以太網，再到現在的跨雲 AI 的多平面網絡，以實現作業按時完成和千兆瓦級 AI 數據中心。

As part of our Leadership 2.0, we have built and focused a cloud and AI mission and organization, now led by industry veteran, Tyson Lamoreaux, reporting to Ken and Hugh. I am so delighted to formally welcome Tyson to Arista. Tyson, if you guys know him well, built the first cloud network for Amazon AWS in the 2000 era and pioneered the first AI network for a stealth sovereign AI company the last couple of years.

作為我們領導力 2.0 的一部分，我們建立了一個專注於雲端和人工智慧的使命和組織，現在由行業資深人士 Tyson Lamoreaux 領導，向 Ken 和 Hugh 匯報工作。我非常高興正式歡迎泰森加入 Arista。如果你們對泰森很了解的話，他曾在 2000 年代為亞馬遜 AWS 建立了第一個雲端網絡，並在過去幾年裡為一家隱密的自主人工智慧公司開創了第一個人工智慧網路。

Tyson, you had a busy few weeks here. Tell us more.

泰森，你這幾週過得挺忙的。請詳細說明。

Thank you, Jayshree, and thanks for the question. It's really incredible to join the team at a time where it is building so much momentum. Spending time with customers has been a top priority for me since coming on board and I've been so impressed with how strong these partnerships are, both with our long-standing titans and with our emerging customers. We're deeply engaged with them on next-gen architectures for their cloud networks, front end, back end, scale up, scale out and scale across I mean, really everywhere. It's translating to a ton of wins, and I got to say, it's a lot more than I anticipated before I got here.

謝謝你，Jayshree，也謝謝你的問題。能夠在團隊發展如此迅速的時候加入，真是太棒了。自從加入公司以來，與客戶共度時光一直是我的首要任務，我非常讚賞我們與長期合作夥伴和新興客戶之間建立的牢固關係。我們與他們就其雲端網路的下一代架構、前端、後端、向上擴展、向外擴展和跨域擴展進行了深入合作，我的意思是，真的在所有方面都是如此。這轉化為大量的勝利，我必須說，這比我來之前預想的要多得多。

I really love our continuing commitment to open standard innovation like Eson and UEC. And of course, the practical here now and always problems that we're addressing by building the hardware systems, software everything that delivers exceptional power efficiency, reliability, density, visibility and manageability for our customers.

我非常欣賞我們對 Eson 和 UEC 等開放標準創新的持續投入。當然，我們正在透過建立硬體系統、軟體等一切來解決當前和未來可能出現的實際問題，從而為我們的客戶提供卓越的電源效率、可靠性、密度、可視性和可管理性。

I think my background as a builder and operator are really well suited to helping the team anticipate customer needs and delivering the right products for them. I guess the last thing I'd highlight is the culture. I mean, it's just tremendous. The customer focus, commitment to quality, innovation and operational excellence are a top notch here and have made me feel right at home.

我認為我作為建造者和運營商的背景非常適合幫助團隊預測客戶需求並為他們提供合適的產品。最後我想強調的是文化方面。我的意思是，這簡直太棒了。這裡以客戶為中心，致力於品質、創新和卓越運營，這些都是一流的，讓我感覺賓至如歸。

Thanks, Jayshree. Back to you.

謝謝你，Jayshree。輪到你了。

Thank you, Tyson, and welcome home. With Tyson's credentials and a track record, Arista is really poised to address multiple facets of the cloud and AI innovation at a system-wide level converging silicon hardware, software, cables, optics and RACs as an overall platform. At the optical Compute Conference, OCP, Arista unveiled its first Ethernet for scale-up network, so E.SUN specification, along with important 12 industry experts. While we began with four co-founders, we are now supporting and increasing to more people so that we can build the right interoperable scale-up standard.

謝謝你，泰森，歡迎回家。憑藉 Tyson 的資歷和過往業績，Arista 完全有能力在系統層面解決雲端和人工智慧創新的多個方面，將矽硬體、軟體、電纜、光纖和 RAC 融合為一個整體平台。在光運算大會 (OCP) 上，Arista 發布了其首款面向擴展網絡的以太網，符合 E.SUN 規範，同時還有 12 位重要的行業專家出席。雖然我們最初只有四位聯合創始人，但我們現在正在支持和增加更多的人，以便我們能夠建立正確的、可互通的規模化標準。

While there's always right noise, Arista also continues to clarify our role in white box and how we will continue to coexist like we always have the past decade or more. The concept is clear. It's all about good, better and where in some simple use cases, a commodity white box is good enough. Yet in other cases, customers seek the value of better Arista blue boxes with state-of-the-art hardware with built-in for Signal Integrity, physical, passive, active component and troubleshooting management. The best is, of course, the Arista-branded US platforms for the ultimate superiority.

儘管總是會有一些合適的噪音，但 Arista 也一直在澄清我們在白盒音樂中的角色，以及我們將如何像過去十年或更長時間以來那樣繼續共存。這個概念很清晰。一切都關乎好、更好，以及在某些簡單的使用場景中，一個普通的白盒子就足夠好了。但在其他情況下，客戶追求的是更好的 Arista 藍色盒子，它配備了最先進的硬件，內建訊號完整性、物理、被動、主動組件和故障排除管理功能。當然，最好的還是 Arista 品牌的美國平台，才能達到極致的優越性。

We find ourselves amid an undeniable and explosive AI megatrend. As AI models and tokens grow in size and complexity, Arista's driving network scale of AI XPUs, handling the power and performance Basically, the tokens must translate to terawatts, teraflops and terabits. We are experiencing a golden era networking with an increasing TAM now of over $100 billion in forthcoming years. Our centers of data strategy, ranging from client to branch to campus to data and now cloud and AI centers a very consistent mission for the company. We will continue to invest in our customers, our leaders our partners and certainly, most of all, our innovative technology.

我們正身處一場不可否認且爆發式發展的AI大趨勢之中。隨著 AI 模型和代幣的規模和複雜性不斷增長，Arista 驅動 AI XPU 網路規模，處理算力和效能。基本上，這些代幣必須轉換為太瓦、太浮點運算和太比特。我們正經歷網路產業的黃金時代，未來幾年其潛在市場規模將超過 1,000 億美元。我們的資料策略中心涵蓋客戶、分公司、園區、資料中心以及現在的雲端和人工智慧中心，這始終是公司的重要使命。我們將繼續投資我們的客戶、我們的領導者、我們的合作夥伴，當然，最重要的是，我們的創新技術。

And with that, Chantelle, I'd like to hand it to you as our CFO for financial specifics.

好了，Chantelle，接下來就交給你這位財務長來處理財務細節了。

Thank you, Jayshree. It is great to see the broadening of the AI ecosystem, and I am excited for Arista to be an innovative unit. Turning now to Q3 performance. Total revenues were $2.3 billion, up 27.5% year over year, above our guidance of $2.5 billion. This was supported by strong growth across all of our product sectors.

謝謝你，Jayshree。很高興看到人工智慧生態系統的不斷擴展，也為 Arista 成為創新企業而感到興奮。現在來看第三季業績。總營收為 23 億美元，年增 27.5%，高於我們先前預期的 25 億美元。這得益於我們所有產品領域的強勁成長。

International revenues for the quarter came in at $468.3 million or 20.2% of total revenue, down from 21.8% in the prior quarter. The overall gross margin in Q3 was 65.2%, and above our guidance of 64%, down from 65.6% last quarter and up from 64.6% in the prior year quarter. The year-over-year gross margin improvement was primarily driven by strength in the enterprise segment.

本季國際營收為 4.683 億美元，佔總營收的 20.2%，低於上一季的 21.8%。第三季整體毛利率為 65.2%，高於我們先前 64% 的預期，低於上一季的 65.6%，高於去年同期的 64.6%。毛利率年增主要得益於企業業務的強勁表現。

Operating expenses for the quarter were $33.3 million or 16.6% of revenue, up from last quarter at $370.6 million. R&D spending came in at $251.4 million or 10.9% of revenue up from $243.3 million in the last quarter. Sales and marketing expense was $109.5 million or 4.7% of revenue compared to $105.3 million last quarter. Both quarter-over-quarter dollar increases were driven by additional headcount, inclusive of the VeloCloud acquisition.

本季營運支出為3,330萬美元，佔營收的16.6%，高於上季的3.706億美元。研發支出達 2.514 億美元，佔營收的 10.9%，高於上一季的 2.433 億美元。銷售和行銷費用為 1.095 億美元，佔營收的 4.7%，而上一季為 1.053 億美元。這兩個季度的美元成長都是由新增員工人數所推動的，其中包括對 VeloCloud 的收購。

Our G&A costs came in at $22.4 million or 1% of revenue, up from last quarter at $22 million. Our operating income for the quarter was $1.12 billion, landing at 48.6% of revenue. Other income and events for the quarter was a favorable $98.9 million, and our effective tax rate was 21.2%. This resulted in net income for the quarter of $962.3 million or 41.7% of revenue. Our diluted share number was 1.277 billion shares resulting in a diluted earnings per share number for the quarter of $0.75, up 25% from the prior year.

我們的一般及行政費用為 2,240 萬美元，佔營收的 1%，高於上一季的 2,200 萬美元。本季營業收入為 11.2 億美元，佔總營收的 48.6%。本季其他收入及事項為9,890萬美元，實際稅率為21.2%。該季度淨收入為 9.623 億美元，佔營收的 41.7%。本季稀釋後股數為 12.77 億股，稀釋後每股收益為 0.75 美元，比上年增長 25%。

Now on to the balance sheet. Cash, cash equivalents and investments ended the quarter at $10.1 billion. Of the $1.5 billion repurchase program approved in May 2025, $1.4 billion remains available for repurchase in future quarters. The actual timing and amount of future repurchases will be dependent on market and business conditions, stock price and other factors.

接下來來看資產負債表。截至本季末，現金、現金等價物及投資總額為 101 億美元。在2025年5月核准的15億美元股票回購計畫中，未來幾季仍有14億美元可供回購。未來回購的實際時間和數量將取決於市場和商業狀況、股價和其他因素。

Now let's move next to operating cash performance for the third quarter. We generated approximately $1.3 billion of cash from operations in the period, reflecting a strong business model performance. DSOs came in at 59 days, down from 67 days in Q2, driven by billing linearity Inventory turns were 1.4 times, flat to last quarter. Inventory increased to $2.2 billion in the quarter, up from $2.1 billion in the prior period. Most of this increase is due to higher evaluation inventory, indicating uptake of our new products and new use cases.

接下來我們來看第三季的經營現金流表現。該期間，我們從經營活動中產生了約 13 億美元的現金，反映出我們強勁的商業模式表現。DSO 為 59 天，低於第二季的 67 天，這主要得益於計費的線性化。庫存週轉率為 1.4 次，與上季持平。本季庫存增至 22 億美元，高於上一季的 21 億美元。這一增長主要歸因於評估庫存的增加，表明我們的新產品和新用例得到了更廣泛的應用。

Our purchase commitments and inventory at the end of the quarter totaled $7 billion, up from $5.7 billion at the end of Q2. We will continue to have some variability in future quarters as a reflection of the combination of demand for our new plots and the lead times from our key suppliers. Our total deferred revenue balance was $4.7 billion, up from $4.1 billion in Q2.

本季末，我們的採購承諾和庫存總額為 70 億美元，高於第二季末的 57 億美元。由於新地塊的需求和主要供應商的交貨週期等因素的影響，未來幾季我們仍將面臨一些波動。我們的總遞延收入餘額為 47 億美元，高於第二季的 41 億美元。

As of Q3, the majority of the deferred revenue balance is product related. Our product deferred revenue increased approximately $625 million versus last quarter. We remain in a period of ramping our new products, winning new customers and expanding new use cases, including AI. These trends have resulted in increased customer-specific acceptance clauses and an increase in the volatility of our product deferred revenue balances.

截至第三季度，大部分遞延收入餘額與產品相關。我們的產品遞延收入比上一季增加了約 6.25 億美元。我們目前仍處於大力推動新產品開發、贏得新客戶和拓展新應用場景（包括人工智慧）的階段。這些趨勢導致客戶特定驗收條款增加，以及我們產品遞延收入餘額的波動性增加。

As mentioned in prior quarters, the deferred balance can move significantly on a quarterly basis, independent of underlying business drivers. Accounts payable days was 55 days, down from 65 days in Q2, reflecting the timing of inventory receipts and payments. Capital expenditures for the quarter were $30.1 million. In October 2024, we began our initial construction work to build expanded facilities in Santa Clara and we expect to incur approximately $100 million in CapEx during fiscal year 2025 this project.

如前幾季所述，遞延餘額可能會按季度發生顯著變化，而與潛在的業務驅動因素無關。應付帳款週轉天數為 55 天，低於第二季的 65 天，反映了庫存收貨和付款的時間安排。本季資本支出為 3,010 萬美元。2024 年 10 月，我們開始在聖克拉拉建造擴建設施的初步建設工作，預計在 2025 財年，該項目將投入約 1 億美元的資本支出。

Q3 delivered a strong performance, underscoring our strategic progress. This continues to give us confidence for the remainder of FY25 and through FY26. Let's first start with our outlook for Q4. Revenue of $2.3 billion to $2.4 billion with continued growth expected across our cloud, AI enterprise and providers markets; gross margin in the range of 62% to 63%, inclusive of possible known tariff scenarios. Operating margin of approximately 47% to 48%. Our effective tax rate is expected to be approximately 21.5% with approximately 1.281 billion diluted shares.

第三季業績表現強勁，凸顯了我們的策略進展。這讓我們對 2025 財年剩餘時間和 2026 財年充滿信心。我們先從第四季的展望開始。預計營收將達到 23 億美元至 24 億美元，雲端、人工智慧企業和供應商市場將持續成長；毛利率將在 62% 至 63% 之間，包括可能的已知資費方案。營業利益率約 47% 至 48%。預計我們的實際稅率約為 21.5%，稀釋後股份約為 12.81 億股。

Incorporating this Q4 outlook, our guidance for FY25 is as follows. Full year revenue growth of approximately 26% to 27% or $8.87 billion at the midpoint. We are on track to deliver between $750 million and $800 million for our campus segment and our AI center target of at least $1.5 billion. For gross margin, the outlook is approximately 64%, inclusive of possible known tariff scenarios. We anticipate operating margin of roughly 48%, demonstrating Arista's strong operational execution and scalable business model.

結合第四季的展望，我們對 2025 財年的展望如下。全年營收成長約 26% 至 27%，或以中間值計算為 88.7 億美元。我們可望實現園區建設工程7.5億至8億美元的目標，以及人工智慧中心至少15億美元的目標。毛利率預期約為 64%，其中包括已知的各種關稅情境。我們預計營業利潤率約為 48%，這表明 Arista 擁有強大的營運執行力和可擴展的商業模式。

Our outlook for FY26 presented at our September Analyst Day remains relatively unchanged. Full year revenue growth of approximately 20%, now at a higher dollar amount of $10.65 billion, inclusive of both a campus target of $1.25 billion and an AI center target of $2.75 billion. For gross margin, our range is expected of approximately 62% to 64%, driven by customer mix and for operating margin and outlook of approximately 43% to 45%, allowing for investments in relation to achieving the strategic goals of Arista.

我們在9月分析師日上提出的2026財年展望基本上維持不變。全年營收成長約 20%，達到 106.5 億美元，其中包括 12.5 億美元的園區目標和 27.5 億美元的人工智慧中心目標。毛利率預計在 62% 至 64% 之間，主要受客戶組合的影響；營業利潤率預計在 43% 至 45% 之間，這為實現 Arista 的策略目標提供了投資空間。

In closing, the momentum continues. The breadth and depth of our customer interactions have never been stronger or more exciting. In true Arista style, we remain pragmatic yet are aware of the potential over the next few years. I wish to extend a warm welcome to Tyson. We are thrilled that you have joined our team, and congratulations to Todd on the well-deserved promotion.

總之，這股勢頭仍在繼續。我們與客戶互動的廣度和深度從未如此強大、如此令人興奮。秉承 Arista 一貫的風格，我們保持務實，同時也意識到未來幾年的發展潛力。我謹代表泰森先生，對他表示熱烈歡迎。我們很高興你加入我們的團隊，也恭喜 Todd 獲得實至名歸的晉升。

I will now turn the call back to Rudy for Q&A.

現在我將把電話轉回給魯迪，讓他回答問題。

Thank you, Chantelle. We will now move into the Q&A portion of the Arista earnings call. To allow for greater participation, I'd like to request that everyone please limit themselves to a single fashion. Thank you for your understanding. Christa, please take it away.

謝謝你，香特爾。接下來我們將進入 Arista 財報電話會議的問答環節。為了讓更多人參與，我希望大家盡量只選擇一種服裝風格。感謝您的體諒。克麗斯塔，請把它拿走。

(Operator Instructions) Tal Liani, Bank of America.

（操作員說明）塔爾·利亞尼，美國銀行。

I want to ask about the sequential or the guidance, and it's a more fundamental question, but I'll back it up with numbers. If you look at last year, the growth was very consistent kind of the last three quarters of the year, the sequential growth rate is between 6.5%, 7.5%. When you look at this year, you started with 10% growth in and it goes to 5% and now only 1.6%. So there is a deceleration. And the question is, what is the underlying what are the underlying drivers for the deceleration? What do we need to take for it for next year? What does it mean? And should we be concerned about the growth going forward?

我想問的是順序或指導方面的問題，這是一個更根本的問題，但我會用數據來佐證。如果回顧去年，最後三個季度的成長非常穩定，環比成長率在 6.5% 到 7.5% 之間。回顧今年，年初成長率為 10%，後來降至 5%，現在只有 1.6%。所以速度有所減慢。問題是，導致減速的根本原因是什麼？明年我們需要準備什麼？這是什麼意思？我們是否應該對未來的成長感到擔憂？

Thanks, Tal. First of all, to answer your last line, there is no concern on our demand. I think the shipments and the revenue follows based on our supplies. So if we're able to make the shipments, then the than the revenue as you saw in Q2, when it due passed any of our guidance, right? However, there are times we can't ship everything despite the demand.

謝謝，塔爾。首先，回答您最後的問題，我們對此要求沒有任何顧慮。我認為出貨量和收入都取決於我們的供應情況。所以如果我們能夠完成發貨，那麼收入就會像你在第二季度看到的那樣，超過我們所有的預期，對吧？然而，有時即使需求旺盛，我們也無法滿足所有出貨需求。

And so you're accordingly seeing that. I wouldn't read too much into the quarterly variances. But I would say we feel we have never felt more strongly about the demand aspect of this. reflected in the continued commitment to 20% growth, even though the number keeps increasing from 8.75% to now 8.87. So no change in demand, some variation in shipments.

所以你才會看到這種情況。我不認為應該過度閱讀季度差異。但我想說的是，我們從未像現在這樣強烈感受到需求面的成長。這體現在我們持續致力於實現20%的成長目標，儘管實質成長率從8.75%上升到目前的8.87%。因此，需求沒有變化，只是出貨量略有波動。

Aaron Rakers, Wells Fargo.

Aaron Rakers，富國銀行。

I'll stick to kind of the model as well. I'm curious when I look at the gross margin guidance for this quarter, I think it was 62% to 63%. I guess if we were to assume that your services gross margin stays consistent, at 81%, 82%. It would seem to imply that product gross margin falls below 60%. So I guess the question is, can you unpack the gross margin drivers in this quarter in terms of the guidance? How much is tariff weighted? Or is there other dynamics to consider? And does that change kind of the expectation as we look forward?

我也會堅持這種模式。我看了本季的毛利率預期，好像是 62% 到 63%，這讓我很好奇。我想，如果我們假設您的服務毛利率保持不變，在 81% 或 82% 左右。這似乎意味著產品毛利率低於 60%。所以我想問的是，您能否根據業績指引，分析本季毛利率的驅動因素？關稅權重是多少？或者有其他因素需要考慮嗎？展望未來，這是否會改變我們的預期？

Okay. Sure, Aaron. First of all, I think you're overestimating our services and software margins, but be that as it may. We do have a mix of product margin where it's significantly below 60% with our cloud and AI titans driving the volume and higher obviously, for the enterprise customers. the average of which, together with services is yielding numbers.

好的。當然可以，亞倫。首先，我認為你高估了我們的服務和軟體利潤率，但不管怎樣。我們的產品利潤率分佈較為分散，其中雲端運算和人工智慧領域的巨頭產品銷售佔比遠低於60%，而企業客戶的利潤率則明顯更高。這些產品的平均利潤率，加上服務收入，最終構成了整體績效。

So when the mix tilts heavily towards the cloud and AI, you can expect some pressure on our gross margins. But overall, I think we managed it very well the manufacturing team, now led by Todd does a fantastic job here. So again, the discipline and mix plays well together, but I don't think it's any change from prior years where when we have a heavy mix of AI and cloud, we feel it in our gross margin.

因此，當業務組合嚴重偏向雲端運算和人工智慧時，我們的毛利率可能會面臨一些壓力。但總的來說，我認為我們做得非常好，製造團隊，現在由托德領導，在這裡做得非常出色。所以，再次證明，技術和組合能夠很好地結合起來，但我認為這與往年的情況並無不同，當我們大量使用人工智慧和雲端運算時，我們的毛利率就會受到影響。

Yes. The only thing I would add to that is I wouldn't assume to the last part of your question that it insinuates or offers a new model going into next year. This is a normal part of our mix conversation and well within the guide that you've seen us perform at these levels before.

是的。我唯一要補充的是，我不會假設你問題的最後一部分暗示或預示著明年將推出新車型。這是我們混音對話中很正常的一部分，也完全符合我們先前在這些水準上的表現。

Michael Ng, Goldman Sachs.

Michael Ng，高盛集團。

Thank you for the question. I was wondering if you could just talk about Arista's positioning as we move into more full rack solutions. Is this going to be more of a partnership model? How do you think about addressing this, I say, growing convergence between compute and networking?

謝謝你的提問。我想請您談談隨著我們向更完整的機架解決方案轉型，Arista 的市場定位是什麼。這會更偏向合作模式嗎？您認為應該如何應對運算和網路之間日益增長的融合趨勢？

Michael, that's a very good question. First of all, as you heard in Analyst Day, Andy Bechtolsheim is personally driving along with the hardware team a significant number of these racks. I think at any given time, we have five to seven projects with different accelerator options Obviously, NVIDIA is the gold standard today, but we can see four or five accelerators emerging in the next couple of years.

邁克爾，你問得好。首先，正如你在分析師日上聽到的那樣，Andy Bechtolsheim 正在親自帶領硬體團隊運送大量這樣的機架。我認為在任何時候，我們都有五到七個項目，每個項目都有不同的加速器選項。顯然，NVIDIA 目前是黃金標準，但我們可以看到在未來幾年內會出現四到五種新的加速器。

Arista is being sought to bring all aspects, the cabling, the co-packaging, the power, the cooling as well as the connection to different XPU cartridges, if you may, as the network platform of choice in many -- so we are involved in a lot of early designs.

Arista 正在尋求在各個方面（包括佈線、聯合包裝、電源、散熱以及與不同 XPU 卡帶的連接）提供解決方案，使其成為許多企業首選的網路平台——因此我們參與了許多早期設計。

I think a lot of these lines will materialize as the standards, Ethernet are getting stronger and stronger. We now have a UEC spec, you heard me talk about the scale-up Ethernet stacks, ESUN, where we can bring different work streams onto the same ethernet heaters, transport heaters, datalink layer, et cetera. I think a lot of this will be underway in 2026 and really emerged in 2027 as scale up Ethernet becomes a more important part of that.

我認為隨著乙太網路等標準的日益完善，這些設想中的許多都會成為現實。我們現在有了 UEC 規範，你們都聽我說過可擴展乙太網路協定棧 ESUN，我們可以將不同的工作流放到同一個乙太網路加熱器、傳輸加熱器、資料鏈結層等等。我認為很多這樣的事情將在 2026 年開始，並在 2027 年真正出現，因為乙太網路的規模化應用將成為其中更重要的一部分。

In terms of how we will gain more recognition of revenue, some of this will be not the classic OEM model. They may be more the Blue Box JDM model. where we work with them on IP and have reference designs and offer them capabilities well beyond the network but many of them will also entail selling the network as is in these racks.

至於我們如何獲得更多收入認可，其中一些將不是傳統的 OEM 模式。它們可能更接近藍盒JDM模式，即我們與他們合作開發IP，提供參考設計，並向他們提供遠超網絡本身的功能，但其中許多也包括將這些機架中的網絡按原樣出售。

Atif Malik, Citi.

阿提夫‧馬利克，花旗銀行。

Jayshree, in your prepared remarks, you mentioned large language model providers like OpenAI and topic, and they have announced partnerships with your cloud tighten. Can you share with us who is driving the decision-making on networking hardware on these announcements? And just your commentary on your share being stable within the circle of your Titan?

Jayshree，你在事先準備好的演講稿中提到了像 OpenAI 和 Topic 這樣的大型語言模型供應商，他們已經宣布與你們的雲端平台建立合作關係。能否告知我們，在這些公告中，網路硬體方面的決策是由誰主導的？你對你在泰坦星圈內保持穩定的份額有什麼看法？

Yes. So to answer your last question, first, I think our share is strong. We always, as you know, coexist with two other types of competitors. One is the bundling strategy with NVIDIA and the other is the white box. So we have not seen any significant changes in share up or down at the moment, it's stable. Having said that, it's also a massive market. And we think rising tide rises all boats and this boat is feeling pretty good.

是的。所以要回答你最後一個問題，首先，我認為我們的市佔率很強。如您所知，我們始終與另外兩種類型的競爭對手共存。一種是與 NVIDIA 的捆綁銷售策略，另一種是白盒銷售策略。所以目前我們還沒有看到股價有任何顯著的上漲或下跌，股價保持穩定。話雖如此，它也是一個巨大的市場。我們認為水漲船高，這艘船現在感覺相當不錯。

Now specific to who makes the decision, it's really a combination. We intimately work with the software and LLM players because they certainly guide the design but we also work with the cloud titans, and it's a shared responsibility between both of them and where the responsibility for procuring the large data centers and the power and the location and the cooling is clearly done by our cloud Titans, but the specifications are exactly what's required on the scale up, scale out network is done by the partners like OpenAI and Anthropic. So it's really a joint decision.

至於由誰來做決定，這其實是多種因素共同作用的結果。我們與軟體和LLM廠商密切合作，因為他們當然會指導設計，但我們也與雲端巨頭合作，這是雙方共同的責任。購買大型資料中心、電力、位置和冷卻的責任顯然由我們的雲端巨頭承擔，但擴展、橫向擴展網路的具體要求則由OpenAI和Anthropic等合作夥伴負責。所以這其實是一個共同的決定。

Samik Chatterjee, JPMorgan.

Samik Chatterjee，摩根大通。

Jayshree, maybe just going back to your earlier response to another query, you mentioned some variability in shipments at a customer level, maybe driving some of the lumpiness quarter to quarter. Just curious if you had talked about previously the Tier 1 customers you're engaged with progressing to their cluster sizes, 100,000 and more. Like how -- has there been any change relative to those plants that are driving the duration here in terms of the fourth quarter guide? Or what is sort of behind the variability that you're seeing in terms of shipment? Is it supply driven at all? Any color on that front?

Jayshree，或許可以回到你先前對另一個問題的回答，你提到客戶層面的出貨量有些波動，這可能是造成季度間波動的原因之一。我只是好奇您之前是否談到過您正在接觸的一級客戶正在發展到集群規模達到 10 萬甚至更多。例如，就第四季業績指引而言，那些影響業績持續時間的工廠是否有任何變化？或者說，你所看到的出貨量波動背後究竟是什麼原因造成的？這完全是供給驅動的嗎？那方面有什麼進展嗎？

Yes. Yes. Samik, I would say it's largely supply driven. As you know, all four are doing well on the 100,000 mark. three have already crossed it. The fourth one, I don't know if they'll cross it by end of the year or next year, but they're getting there. So we're feeling pretty good on our large GPU deployments. At the same time, the variability I was stating is demand is greater than our ability to ship.

是的。是的。薩米克，我認為這主要是由供應驅動的。如你所知，這四款遊戲都進展順利，銷售量都接近10萬。其中三款已經突破了10萬大關。第四條線，我不知道他們是否能在今年年底或明年跨越，但他們正在朝著目標前進。所以我們對我們的大規模 GPU 部署感覺相當不錯。同時，我之前提到的波動性是，需求大於我們的出貨能力。

Lead times on many of our components, including standard memory and chips and merchant silicon and everything, nothing like 2022, but they have very long lead times ranging from 38 to 52 weeks. So we are coping with that. And you can see Chantelle is leaning in and making greater and greater purchase commitments, we wouldn't do that without demand.

我們許多組件的交貨週期，包括標準記憶體、晶片、商用矽片等等，雖然不像 2022 年那樣，但它們的交貨週期非常長，從 38 週到 52 週不等。所以我們正在應對這種情況。你可以看到 Chantelle 正在加大投入，做出越來越大的採購承諾，如果沒有需求，我們是不會這樣做的。

Amit Daryanani, Evercore.

Amit Daryanani，Evercore。

I guess my question, I think folks are kind of trying to ask around this a bit is the growth rates you've had in the last three, four quarters of 5%, 5%, call it, you sort of implying that will decelerate, as, not just in December, but also in '26, I think at this point, right, high 20% growth goes to low 20%. Maybe just talk about what is driving that kind of deceleration? Because certainly, if you look at things like your purchase commitment and your deferred product growth, it would almost imply things can accelerate not decelerate in the out quarter. So just what's driving the deceleration in the out quarters would be helpful to understand.

我想問的問題是，大家似乎都在試圖了解過去三、四個季度的成長率是 5%，5%，你是不是暗示這個成長率會放緩，不僅在 12 月份，而且在 2026 年也會放緩，我認為目前來看，20% 以上的成長率會降至 20% 左右。或許可以討論一下導致這種減速的原因是什麼？因為如果你看看你的採購承諾和你的遞延產品成長等因素，這幾乎可以肯定地表明，在接下來的一個季度裡，情況可能會加速而不是減速。因此，了解外側車輪減速的原因將很有幫助。

Okay, Amit, but I don't like the word deceleration. We're talking about big, big numbers here, guys. And I'm committing to double-digit 20% and above percentage, so call it deceleration, call it, variability across quarters, and demand is great. I just don't know whether it will land in '26 or '27.

好吧，阿米特，但我不太喜歡「減速」這個詞。夥計們，我們說的可是非常非常大的數字。我承諾成長率將達到兩位數（20% 及以上），所以你可以稱之為增速放緩，也可以稱之為季度間的波動，但需求仍然旺盛。我不知道它會在 2026 年還是 2027 年登陸。

Yes. The only other thing I'm going to add to this just generally is a topic is that when you think about that the large AI use cases, our acceptance clauses, it really comes down to that coming together and the timing of that. That doesn't follow a seasonality model. part of what we're working with.

是的。我唯一要補充的是，總的來說，當你思考大型人工智慧用例、我們的驗收條款時，你會發現，這實際上取決於這些用例的整合以及時機。這不符合季節性法則。這是我們目前面臨的部分問題。

Good point. It lands when it lands. That is a very good point that Chantelle is making that in the cloud, we started having predictability of how they landed and how they got constructed. In AI, it's taking longer.

說得對。它什麼時候落地就什麼時候落地。Chantelle 提出的觀點非常好，在雲端，我們開始能夠預測它們的著陸方式和建構方式。在人工智慧領域，這需要更長時間。

David Vogt, UBS.

David Vogt，瑞銀集團。

I'm going to ask this question at the risk of Jayshree yelling at me. When we think about your '26 outlook that you just raised, which we didn't expect you to raise this early in the cycle, if I just take what you're doing with regards to the AI-centric opportunity, Campus and Bello, it doesn't leave a lot of room for growth in the core business outside of AI and campus. Maybe can you speak to what you're seeing in that particular market and how we should think about that progressing through 2026?

我冒著被傑伊什裡吼的風險也要問這個問題。當我們考慮您剛剛提出的 2026 年展望時（我們沒想到您會在周期這麼早的時候提出這樣的展望），如果我只看您在以人工智能為中心的 Campus 和 Bello 領域所做的工作，那麼在人工智能和 Campus 之外的核心業務方面，並沒有留下太多的增長空間。您能否談談您在該特定市場觀察到的情況，以及我們應該如何看待該市場在 2026 年的發展？

Okay. Well, just -- I'll say in its Jayshree's to figure out which tone she wants to answer this. But I think the part that I would take a look at again, I go back to kind of how we started at early '25, maybe even '24, part of our style is to not assume 100% everything hits to get to a number, and we'd like to leave ourselves with some optionality.

好的。嗯，就——我得用 Jayshree 的語氣來判斷她想用哪種語氣回答這個問題。但我認為，我會再次審視一下，回到我們2025年初，甚至可能是2024年的時候，我們風格的一部分是不會假設所有事情都能100%成功來達到某個數字，我們希望給自己留有一些選擇餘地。

And so we're putting some goals for ourselves with the AI, we're going goals for ourselves with campus. It doesn't mean we're not focused on the rest. But I don't think it's the right approach to assume everything is going to be 100% and leave ourselves exposed and we'll continue to update as we see it. Right, Jayshree.

因此，我們為自己設定了一些人工智慧目標，也為自己設定了一些校園發展目標。但這並不意味著我們沒有關注其他方面。但我認為假設一切都會100%完美，讓自己處於危險之中，這種做法是不正確的，我們會根據情況繼續更新。沒錯，Jayshree。

Yes, absolutely. And so yelling isn't the tone I'd like to be attributed to excitement, maybe our enthusiasm is the one I'd like you to think about, which is clearly AI and campus is going to grow and do great guns for us, as it should because they are two very large TAMs whether it is Ken and Tyson driving the AI and cloud TAM or whether it's original driving the campus and these two are going to grow substantially in double digits, right?

是的，絕對的。所以，我不希望人們用大喊大叫來形容我的興奮，也許我們應該用熱情來表達，那就是人工智能和校園業務顯然會發展壯大，為我們帶來巨大的收益，這是理所當然的，因為它們是兩個非常大的潛在市場，無論是 Ken 和 Tyson 推動人工智能和雲計算的潛在市場，還是 Original 推動校園業務，這兩個業務都將以兩位數對校園的速度大幅增長？

So to your point, it doesn't leave the core business with a lot of opportunity. But that's not to say it may be flattish. It may be grow. It's to say that our customers are putting more attention there. and that the existing business, which is already on very large numbers will have lesser growth. We don't yet know if it's flattish or single-digit or whether more will go to AI, we frankly can't predict the mix this early in the game on 2026. But we think we're in for a great ride in 2026.

所以正如你所說，這並沒有給核心業務留下太多機會。但這並不意味著它可能是平坦的。它可能會生長。也就是說，我們的客戶將更多注意力放在了那裡，而現有業務（該業務規模已經非常龐大）的成長速度將會放緩。我們還不知道成長是平穩還是個位數，也不知道人工智慧是否會獲得更多份額，坦白說，在2026年到來之際，我們無法預測最終結果。但我們認為2026年將會是一段精彩的旅程。

Ben Reitzes, Melius Research.

本‧雷茨，梅利烏斯研究公司。

I appreciate you clarifying some of those earlier questions more in the long term tree. I think there was an earlier question on OpenAI and Anthropic. And just some of these larger builds with the private companies that obviously are becoming hyperscalers. Maybe without naming names or whatnot.

感謝您從長遠角度澄清了之前的一些問題。我認為之前有人問過關於 OpenAI 和 Anthropico 的問題。還有一些規模較大的項目，是由一些私人公司開發的，這些公司顯然正在轉型為超大規模資料中心。或許不會指名道姓之類的。

I just wanted to hear about your confidence I'm able to participate in some of these builds that are affiliated with some of your cloud titans. And do you think you'll get a lot of this growth? Is there anything that's changing or evolving that gives you more or less confidence as we end the year here in '25?

我只是想了解一下您是否有信心讓我參與一些與您的一些雲端巨頭相關的專案。你認為你會獲得很大的成長嗎？在即將結束的2025年，有沒有什麼改變或發展讓你更有信心或更沒信心？

Yes, that's a really good question, Ben, and thank you for that thoughtful question. Until now, majority of how we've measured our AI success through our cloud and AI titans has been a number of GPUs and how much are they installing and can we verify that the Ethernet network works. The majority of it to date has been scaled out.

是的，本，你問得真好，謝謝你提出這麼有見地的問題。到目前為止，我們衡量透過雲端和人工智慧巨頭實現人工智慧成功的主要方式是 GPU 的數量以及它們的安裝量，以及我們能否驗證乙太網路是否正常運作。迄今為止，大部分已經規模化生產。

First, I want to reflect that there are three big use cases sitting in front of us, scale up, scale out and scale across. Arista's participation to date has largely been in scale-out. So we got two major use cases in addition, augmenting this, and that's what makes the Etherlink portfolio that Ken described so eloquently so beautiful.

首先，我想指出我們面前有三個主要的用例：向上擴展、向外擴展和橫向擴展。迄今為止，Arista 的參與主要集中在規模擴張方面。因此，我們又增加了兩個主要用例，進一步增強了這一點，而這正是 Ken 所描述的 Etherlink 產品組合如此美妙的原因。

Now how are these being built? Clearly, they're being driven by large language models, tokens transformers, inference use cases, you name it all. So the influence is clearly coming from these players you named but the way they are driving the infrastructure, and I can't give track of the giga myself, it's 10 gigawatts here, 10 there, 30 there. It's adding up to a lot but I can just tell you, no matter what it is, Arista is being looked at as a very important and relevant participant, especially right now in the scale out and scale across we will participate in the scale up.

這些建築是如何建造的？顯然，它們是由大型語言模型、詞元轉換器、推理用例等等所驅動的。所以，這種影響顯然來自你提到的這些參與者，但他們推動基礎設施建設的方式，我自己也無法追蹤具體的千兆瓦數，這裡 10 吉瓦，那裡 10 吉瓦，那裡 30 吉瓦。事情很多，但我可以告訴你，無論是什麼，Arista 都被視為非常重要和相關的參與者，尤其是在目前的規模擴張和橫向擴張中，我們將參與規模擴張。

It will take a little longer. Today, there's largely a set of proprietary technologies like NVLink or PCIe. And I think that will happen more in '27. So that to say that as we get now confident about exceeding our $10 billion goal next year, we're looking at our next goal of $15 billion in the next few years. And I think AI will be a very large part of it and will be the companies you mentioned.

還需要一些時間。如今，市面上大多是一些專有技術，例如 NVLink 或 PCIe。我認為這種情況在 2027 年會更加普遍。所以，也就是說，隨著我們對明年實現 100 億美元的目標充滿信心，我們正在展望未來幾年內實現 150 億美元的下一個目標。我認為人工智慧將在其中扮演非常重要的角色，而你提到的那些公司也將是其中的關鍵一環。

Tim Long, Barclays.

提姆朗，巴克萊銀行。

Jayshree, maybe if we could just dig a little bit. You mentioned Blue Box a few times here kind of in that middle portion of the good, better, best, two-parter. One, could you talk a little bit about kind of the economic model margins or anything like that, that we should expect as Blue Box becomes a bigger part of the mix over time?

Jayshree，或許我們可以稍微挖掘一下。你在這裡幾次提到了 Blue Box，大概是在「好、更好、最好」這個兩部分內容的中間部分。第一，您能否談談隨著 Blue Box 業務組合的不斷擴大，我們應該預期在經濟模型利潤率或其他方面會有哪些變化？

And second, can you talk a little bit about where we would expect to see these types of deployments? I'm assuming something like scale up might be, as you described, a little bit more simple to the full EOS, but from either a customer or a use case standpoint, where would you expect Arista to be most successful with Blue Box deployment?

其次，您能否談談我們預計會在哪些地方看到這類部署？我猜想，正如你所描述的，像擴展這樣的操作可能比完整的 EOS 更簡單一些，但從客戶或用例的角度來看，你認為 Arista 在 Blue Box 部署方面最有可能取得成功的地方是什麼？

Yes. Thank you. Those are a good set of questions. I think I mentioned at the Analyst Day already quickly seeing success. I'll give you one example where they were just not getting their white box to work. These are AI mission-critical workloads. And we're seeing a neo-cloud come right in with, in this case, non-NVIDIA GPUs, in fact, where they're looking to deploy Arista with its excellent hardware.

是的。謝謝。這些都是很好的問題。我想我在分析師日上已經提到過，我們已經很快看到了成功。我舉個例子，他們就是搞不定那個白盒子。這些都是人工智慧的關鍵任務工作負載。我們看到一種新型雲端技術正在興起，在這種情況下，它使用的是非NVIDIA GPU，實際上，他們正在尋求部署擁有出色硬體的Arista。

And at first, they wanted to do an Open NOS, but now they are adopting a hybrid strategy where it's not only an Open NOS, but Ken's EOS is coming to shine in its full glory in this use case. So in this case, I think it's a Blue Box to start with quickly going into a hybrid state of blue and branded EOS box. The economics on that is not too different from cloud and AI titans, generally speaking, although there will be scenarios like you rightly mentioned, hasn't yet come to play.

起初，他們想做一個開放式網路作業系統 (Open NOS)，但現在他們正在採用混合策略，不僅要做一個開放式網路作業系統，還要讓 Ken 的 EOS 在這個用例中充分發揮其優勢。所以，我認為在這種情況下，它最初是一個藍色盒子，然後迅速過渡到藍色和品牌 EOS 盒子的混合狀態。總的來說，這方面的經濟效益與雲端運算和人工智慧巨頭並沒有太大區別，儘管如您所提到的，某些情況尚未發生。

But as we go to significant scale up volume, we expect more margin and economic capability coming together. In other words, the volume of these things will be larger, the pressure on margins will be greater. So -- but we will carefully have a mix of scale up, scale out and scale across to not affect the overall margins but definitely take our fair share in that. So hopefully, I answered your question on both Tim.

但隨著產量大幅提升，我們預期利潤率和經濟效益將會提高。換句話說，這些事物的數量會更大，利潤空間面臨的壓力也會更大。所以——但我們會謹慎地結合規模擴大、規模擴張和規模橫向擴展，以免影響整體利潤率，但肯定能從中獲得我們應得的份額。希望我已經回答了你關於蒂姆的問題。

Meta Marshall, Morgan Stanley.

Meta Marshall，摩根士丹利。

Great. Maybe a question for you, Jayshree. Just on I know you guys aren't breaking out the front end and back end anymore. But just as more inference kind of use cases are getting built out, just what are you seeing in terms of just like how the front-end network upgrades are happening maybe versus where your expectations were a year ago?

偉大的。傑伊什裡，或許我想問你一個問題。我知道你們不再使用前端和後端了。但隨著更多推理類別用例的構建，您認為前端網路升級的進展與您一年前的預期相比有何不同？

Yes. Thank you. I think a year or maybe even two years ago, it I may have told you this, we were literally outside looking in at all these back-end networks that were largely being constructed by with InfiniBand. We've seen a sea change, particularly this year, where obviously more and more times, we're being invited to construct their 800-gig. Last year, it was more 400 gig.

是的。謝謝。我想，一年前甚至兩年前，我可能跟你說過，我們當時實際上是在外面看著所有這些後端網絡，而這些網絡大多是由 InfiniBand 構建的。我們已經看到了巨大的變化，尤其是今年，顯然越來越多的人邀請我們去建造他們的 800G。去年，這個數字超過了 400 GB。

And I think next year will be a combination of 8 and 1.60 on the back end. The back end is putting pressure on the front end, which is why it's getting more difficult for us to say, okay, what's the back-end member that natively connect connects to GPUs and what is the front end.

我認為明年後端的數值會是 8 和 1.60 的組合。後端給前端帶來了壓力，所以我們越來越難說，好的，哪個後端成員原生連接到 GPU，哪個是前端。

But we know of concrete cases and our cloud titans, where not only is putting pressure on the AI number, but they're having to go and upgrade their cloud infrastructure to deal with it. That part is not happening in a small sort of way. But what's happening in a big sort of way is the back and front are co-listing and converging more and it's really becoming hard to tell, and it's probably 60 1.5 a dozen of the other.

但我們知道一些具體案例，例如雲端運算巨頭，人工智慧不僅給它們帶來了壓力，而且它們還不得不升級雲端基礎設施來應對這種情況。那部分事情的發生並非小事。但從某種程度上來說，現在發生的是，後台和前台正在共同上市，並且越來越趨於融合，真的很難分辨，兩者的比例可能是 60 比 1.5，或者其他十幾個。

I'd just like to point out that we're seeing that Arista, I think, is the only successful vendor outside of China selling both front end and back end. And this is where our engineering alignment is so important because we can offer the customer a consistent solution across their entire infrastructure. I think this is a unique differentiator that will really help us succeed as these networks become more and more mainstream.

我想指出的是，我們看到 Arista 似乎是中國以外唯一同時銷售前端和後端產品的成功供應商。正因如此，我們的工程協同才顯得尤為重要，因為我們可以為客戶提供覆蓋其整個基礎設施的一致解決方案。我認為這是一個獨特的差異化優勢，隨著這些網路越來越主流化，這將真正幫助我們取得成功。

Karl Ackerman, BNP Paribas.

卡爾‧阿克曼，法國巴黎銀行。

How should we think about your market opportunity between disaggregated scheduled fabrics versus non-scaled fabrics which appear to be used in the largest AI accelerator clusters at one of your largest customers. I mean, do you, in fact, happen to be the only networking switch better who offers both networking topologies? And I'm curious if other data center operators seek to adopt your DS architecture, given the congestion-free advantages it offers?

我們該如何看待您在分散式調度架構和非規模化架構之間的市場機會？後者似乎已被您最大的客戶之一的大型 AI 加速器集群所採用。我的意思是，你們是不是真的是唯一一家提供兩種網路拓樸結構的更優秀的網路交換器廠商？我很好奇其他資料中心營運商是否會考慮採用你們的DS架構，因為它具有無擁塞的優勢？

Well, I think you hit on it, and Ken hit on it, too, so I'd like him to answer part of the question. But look, we're not religious. We jointly developed the DSF architect with one of our leading cloud tightness meter. And we've been selling the not scheduled fabric for a very long time. So we've never been religious about this. And both are doing very, very well at our cloud titans and specifically the one we codeveloped with.

嗯，我覺得你說的有道理，肯也說得有道理，所以我想讓他回答一部分問題。但是，我們又不是信教的人。我們與一家領先的雲密性測量公司共同開發了 DSF 架構。我們銷售未列入計劃的布料已經很長時間了。所以我們在這方面從來沒有宗教信仰。這兩款產品在我們的雲端平台中都表現得非常出色，尤其是我們共同開發的那款。

That's exactly right. We've had both architectures in massive production scale for, I think, 15 years now. And we will continue to offer this range of choice to our customers, offering them their choice between the highest value fabric with deep buffers, no hotspots, congestion-free loss-free or an unscheduled fabric, which is maybe lower cost, but also can be more difficult to operate. And they both run the same software. So it gives the customer a range of options and a consistent operating model.

完全正確。這兩種架構我們已經大規模應用於生產環境約 15 年了。我們將繼續為客戶提供這種選擇範圍，讓他們可以選擇具有深緩衝層、無熱點、無擁堵無損耗的最高價值織物，或者選擇非計劃織物，這種織物可能成本較低，但操作也可能更困難。而且它們都運行相同的軟體。因此，它為客戶提供了一系列選擇和一致的營運模式。

Simon Leopold, Raymond James.

西蒙·利奧波德，雷蒙德·詹姆斯。

I wanted to come back to the topic around the blue box, which you've talked about quite a bit at the analyst meeting. So I appreciate it's not new. But what I don't quite think I understand how it may be evolving or changing in that it sounds like there's a broader base of customers that may be employing it and that this is a factor that's in your 2026 margin guidance. Could you elaborate on what you're assuming Blue Box trends are in 2026?

我想回到關於藍色方框的話題，您在分析師會議上已經多次談到這個話題。所以我知道這並不是什麼新鮮事。但我不太明白它可能會如何發展或變化，因為聽起來似乎有更廣泛的客戶群正在使用它，而這正是你們 2026 年利潤率預期中的一個因素。能否詳細說明您對 2026 年藍盒趨勢的預測？

I think the Blue Box trends in 2026 will continue to remain with a handful of customers who know how to -- who have the operational skill to deal with it. So think of that as largely our specialty cloud providers or tighten. It's not going to be mainstream. So single-digit customers probably maybe 10, maybe 20, but it's not going to be hundreds, number one, because we really have to have the operational excellence to take our net DI and our hardware and build upon it and they're open loss or whatever, right?

我認為到 2026 年，藍盒服務趨勢仍將只保留給少數懂得如何操作的客戶——也就是那些具備處理藍盒服務所需操作技能的客戶。所以，你可以把這主要理解為我們的專業雲端服務提供者或加強監管。它不會成為主流。所以個位數的客戶可能只有 10 個，也可能只有 20 個，但肯定不會有幾百個，這是第一點，因為我們真的需要卓越的運營能力，才能利用我們的網絡數據基礎設施和硬件，並在此基礎上發展，而他們卻面臨虧損或其他問題，對吧？

However, in that scenario, you are right to point out that because we may not have the EUS layer, we will take a lower margin on that. And that's factored into our 2026 guide and mix and we think the combination of the Blue Box and the US branded box, if I can call it that, will continue to help us thrive with a profitable and high-growing business.

但是，在這種情況下，您指出我們可能沒有 EUS 層，因此我們將採取較低的利潤率，您說得對。這一點已納入我們的 2026 年指導方針和組合中，我們認為，藍盒和美國品牌包裝盒（如果我可以這樣稱呼的話）的組合將繼續幫助我們蓬勃發展，實現盈利和高速增長的業務。

James Fish, Piper Sandler.

詹姆斯·菲什，派珀·桑德勒。

Just on that topic of Blue Box tracking way maybe not just 2026, but what do you see in terms of the mix regarding the adoption curve as to what percentage of the business could actually represent over not just next year but three, five years from now? And you guys mentioned the convergence of end and back end. Does that take away from kind of your advantage of where you sit today, though, if that mine starts to blur a little more and allows competition to enter?

關於 Blue Box 的追蹤方式，也許不僅僅是 2026 年，而是從採用曲線的角度來看，您認為在未來三到五年內，Blue Box 實際上能佔到企業總業務量的百分比會是多少？你們也提到了前端和後端融合的問題。但是，如果我的優勢開始變得模糊，讓競爭對手有機會進入，這是否會削弱我目前所擁有的優勢？

Yes, please go ahead.

好的，請繼續。

In terms of the front end and back-end converging, this is purely advantageous to us because the front end requires a massive number of features. It's incredibly mission-critical and supports a whole variety of applications. not just the straightforward of demanding communication patterns at the AI back end.

就前端和後端融合而言，這對我們來說完全有利，因為前端需要大量的功能。它至關重要，支援各種各樣的應用，而不僅僅是人工智慧後端對通訊模式的簡單要求。

So we see that our ability to tackle both of them effectively is a significant source strength and a real differentiator and not something that's not easy for competitors to replicate. If you look at NVIDIA, for example, the sales volume is small in the front end, and Cisco was small in the back end. And so I think we'll see that kind of convergence being beneficial to us.

因此，我們能夠有效應對這兩個問題，這是我們的一項重大優勢和真正的差異化優勢，競爭對手很難複製。例如，如果你看看英偉達，你會發現它前端的銷售額很小；而思科的後端銷售額也很小。所以我認為這種融合對我們來說是有利的。

Yes. Thank you, Ken. On the Blue Box, I'm not sure we modeled three to five years. But if I had to venture what I think the evolution of the Blue Box will be. I think it will be more significant in the scale up use cases where there's a higher dependency on the strength of our hardware and our net DI capability and a lower requirement for software. So don't know yet what that will be. I think it will be high end units, low in dollars kind of thing. So the mix may still be small, but it will actually be incremental since that's not a use case we do today.

是的。謝謝你，肯。關於藍盒模型，我不確定我們是否模擬了三到五年的情況。但如果讓我預測藍盒子未來的發展方向，我會怎麼說呢？我認為在規模化應用場景中，這將更加重要，因為這些場景更依賴我們硬體的強大效能和淨數據整合能力，而對軟體的要求則較低。所以現在還不知道會是什麼結果。我認為會是那種高端產品，但價格很低。因此，混合比例可能仍然很小，但實際上是逐步增加的，因為這不是我們目前使用的用例。

Antoine Chkaiban, New Street Research.

Antoine Chkaiban，新街研究公司。

I'd like to ask about the UHC. So can you maybe tell us about the progress that the consortium is making, whether the different voices are aligned and what milestones investors should be looking out for going forward?

我想諮詢一下全民健康保險（UHC）的情況。那麼，您能否談談該聯盟目前的進展，各方意見是否一致，以及投資者未來應該關注哪些里程碑？

And can you repeat your question? You weren't coming into clear?

您能再說一次您的問題嗎？你剛才沒說清楚嗎？

I'm asking about the Ultra Ethernet consortium. Maybe can you tell us about the progress that the consortium has?

我問的是超乙太網路聯盟。您能否向我們介紹一下該聯盟的進展？

Yes, yes. Yes, Antoine, yes. So after two years of lots of hard work led by Hugh Holbrook and now Ton Emans, UEC did publish their first specification, I believe it was 1.0 in June of 2025. Arista's ethylene portfolio is entirely capable compatible, and we will continue to add more and more compliance, packet trimming, packet spring, dynamic load balancing. These are all important features that our switches support.

是的，是的。是的，安托萬，是的。經過 Hugh Holbrook 和 Ton Emans 領導的兩年多的辛勤努力，UEC 終於發布了他們的第一個規範，我相信是 2025 年 6 月發布的 1.0 版本。Arista 的乙烯產品組合完全相容，我們將繼續增加合規性、資料包修剪、資料包彈簧、動態負載平衡等功能。這些都是我們交換器支援的重要功能。

And we will augment that with the ESUN specification. As I described, we've been an early pioneer, four vendors started this together, including Broadcom, Arista and a couple of our Titan customers, I'm pretty sure it will be 20, 25, 30 over time. And having a standards-based our OCP ESUN agreement will allow us to expand UEC into the scale-up configuration as well, leveraging UEC and IEEE specs. So this modular framework for Ethernet for scale-up and scale-out is a thing of beauty and Arista is in the middle of it.

我們將根據 ESUN 規範對其進行補充。正如我所描述的，我們是早期開拓者，包括博通、阿里斯塔和我們的幾家泰坦客戶在內的四家供應商共同開始了這項工作，我很確定隨著時間的推移，這個數字會達到 20、25、30。有了基於標準的 OCP ESUN 協議，我們將能夠利用 UEC 和 IEEE 規範，把 UEC 擴展到擴展配置。因此，這種用於乙太網路的模組化框架，無論是向上擴展還是向外擴展，都堪稱完美，而 Arista 正是其中的核心力量。

George Notter, Wolfe Research.

喬治諾特，沃爾夫研究公司。

I think in the monologue, you mentioned Neo Cloud is an area where you're getting more momentum. I think you guys actually said at the Analyst Day as well. I'm just curious like what are you seeing with that customer set? I guess from my perspective, I've historically cut out of that customer as being more focused on the bundle, which isn't necessarily your game, but it sounds like you're maybe talking a bit more positively. I'm just wondering what you're seeing in that space.

我認為在獨白中，你提到 Neo Cloud 是你正在獲得更大發展勢頭的領域。我想你們在分析師日也提到這一點。我只是好奇，你從這組客戶資料看到了什麼？從我的角度來看，我過去一直把這類客戶排除在外，因為我更注重捆綁銷售，這可能不是你的策略，但聽起來你現在的態度可能更積極一些。我只是好奇你在那片空間裡看到了什麼。

Yes. No, George, I think you're right. I think in the beginning, we were looking at them bundling. I can think of two examples where we want to and invite it to the party because you won my GPU, you've got to get the network from me. So we want to -- but there are -- leaving the two aside and even I think those two might be -- might get open-minded over time.

是的。不，喬治，我覺得你是對的。我認為一開始，我們考慮的是將它們綁在一起。我可以想到兩個例子，我們想邀請它參加派對，因為你贏得了我的 GPU，你必須從我這裡獲得網路。所以我們想——但是——先把這兩個人放在一邊，即使我認為那兩個人隨著時間的推移可能會——可能會變得思想開放。

There are many more neoclouds worldwide coming up that are really looking for Arista's help, not only on the product but on the network design on the software capability that they just don't have the staff and expertise to do everything themselves, and they would rather let us satisfy their network needs.

全球還有許多新興的雲端平台正在湧現，它們非常需要 Arista 的幫助，不僅是產品方面，還有網路設計以及軟體功能方面。它們本身缺乏足夠的人員和專業知識來完成所有工作，因此更願意讓我們來滿足它們的網路需求。

So we are taking down many neo cloud and smaller enterprises admittedly smaller numbers of GPU clusters as well. But if they start with 1,000 to a few thousand, then we're hopeful they'll grow because the one advantage they seem to all have is colo space and power, which is, as you know, is a very prestigious asset going forward.

因此，我們正在關閉許多新型雲端平台和規模較小的企業平台，當然，GPU叢集的數量也較少。但如果他們從 1000 到幾千人起步，那麼我們希望他們能夠發展壯大，因為他們似乎都擁有的一個優勢是託管空間和電力，正如你所知，這是未來非常寶貴的資產。

Sebastien Naji, William Blair.

塞巴斯蒂安·納吉，威廉·布萊爾。

I'd like to understand a little bit more about the investments you're making in the enterprise go-to-market looks like sales and marketing expenses stepped up in the quarter. Where do you think you make the most progress as we go into 2026? Is it geographic expansion? Is it investing more into the channel? Is it just trying to cross-sell more into the existing enterprise customer base? I'd love to get your thoughts there.

我想更深入了解您在企業市場推廣方面的投資情況，看起來本季的銷售和行銷費用有所增加。您認為在進入 2026 年之際，我們在哪些方面能夠取得最大的進步？這是地域擴張嗎？公司是否加大了對該通路的投入？是不是只是想向現有企業客戶群進行更多交叉銷售？我很想聽聽你的想法。

Yes. You're hitting on a really important spot because we really have two sites to our coin. On one side, the AI and cloud makes us dizzy, but we're just as excited and buzzy about the huge $30 billion TAM, and that's why we're so happy to have Todd here. We were just at an international innovate in London that Ken, Todd and I all got a chance to see and the excitement and enthusiasm for a relevant high-quality network vendor has never been higher. So indeed, we want to invest there. Todd, do you want to say a few words?

是的。你問到了一個非常重要的問題，因為我們的幣其實有兩個網站。一方面，人工智慧和雲端運算讓我們眼花繚亂，但我們對高達 300 億美元的潛在市場規模同樣感到興奮和激動，這就是為什麼我們非常高興 Todd 能來到這裡。我們剛參加了在倫敦舉行的國際創新大會，我和肯、托德都有機會親眼目睹了大會上人們對高品質網路供應商的熱情和興奮程度從未如此之高。所以，我們確實想在那裡投資。托德，你想說幾句話嗎？

Yes. We're excited about the growth here across the board from the enterprise space, but there's three real dimensions we're staying really focused on. One is expansion into the campus. The VeloCloud acquisition completes our portfolio there, getting great traction, pushing extremely hard around the world. It's a ton a ton of white space accounts for us that we haven't gotten.

是的。我們對企業領域的全面成長感到興奮，但我們仍然專注於三個真正的方面。一是向校園內擴張。收購 VeloCloud 完善了我們在該領域的佈局，獲得了巨大的發展勢頭，並在全球範圍內大力推進。對我們來說，還有大量的空白市場帳戶沒有被我們利用。

I think you mentioned geographic expansion. That's great. We saw good numbers in Asia this quarter. We've got a lot of opportunity, I think, to accelerate there. And we like the progress -- but the last is just reaching new logos, and we're investing in our channel to really deliver that and bring us more opportunities more at bats to find folks and introduce them to Arista for the first time.

我想你提到了地域擴張。那太棒了。本季我們在亞洲的業績表現良好。我認為我們有很多機會可以加速發展。我們對取得的進展感到滿意——但最後一步是接觸到新的品牌標識，我們正在投資我們的管道，以真正實現這一目標，並為我們帶來更多機會，讓我們找到更多人，並讓他們第一次了解 Arista。

That was a cricket analogy.

那是一個關於板球的比喻。

Yes, it was quick.

是的，很快。

Yes, there you go. So Sebastien, we're feeling really good, and it clearly is the other half of our numbers.

對，就是這樣。所以塞巴斯蒂安，我們感覺非常好，這顯然是我們數據的另一半。

Ryan Koontz, Needham.

Ryan Koontz，Needham。

This is Jeff Hobson on for Ryan Koontz. We've seen a lot of the deals with the hyperscalers or the AI model companies with new data center buildouts, probably not a level since we've seen with the cloud buildout. So I was just curious, is there a way to think about Arista's opportunity with new network builds versus refreshing or upgrading existing networks?

這裡是傑夫霍布森，替瑞安昆茨為您報道。我們看到許多超大規模資料中心營運商或人工智慧模型公司都在進行新的資料中心建設，其規模可能不如雲端建設。所以我很好奇，對於 Arista 而言，新建網路與更新或升級現有網路相比，有哪些機會？

Yes, that's exactly the way to think about it because in the past with the cloud, we really got to talk about gigawatts and beyond. So much of them are multi-megawatts, so these are newly constructed AI buildouts as opposed to the traditional CPU or storage-driven cloud buildouts. Of course, they will have refresh too. But frankly, they're not getting the attention, all the attention is going to the new buildouts for AI. So that's the right way to look at it.

是的，正是這樣想的，因為過去使用雲端運算時，我們確實討論過千兆瓦甚至更大的容量。其中很多都是兆瓦級的，所以這些都是新建的 AI 構建，而不是傳統的 CPU 或儲存驅動的雲端構建。當然，他們也會刷新頁面。但坦白說，它們並沒有得到關注，所有的注意力都集中在人工智慧的新建設上。所以，這才是正確的看待這個問題的方式。

So we have time for one last question.

所以我們還有時間問最後一個問題。

Ben Bollin, Cleveland Research.

本‧博林，克利夫蘭研究公司。

Jayshree, you talked a little bit about some of the tightening lead time conditions out there. Curious what you're seeing from these cloud customers around engineering and delivery lead times, how that has evolved? And changes you're seeing on your confidence in delivering their needs whatever, in the next 12, 18 months. That's it.

Jayshree，你剛才稍微談到了目前交貨週期日益縮短的情況。很好奇您從這些雲端客戶那裡了解到的工程和交付週期方面的情況，以及這些情況是如何演變的？以及在接下來的 12 到 18 個月內，您在滿足客戶需求方面的信心所發生的變化。就是這樣。

Ben, as you know, forecast is a very delicate science. I hardly get it right. So I do rely as does Tyson and Ken and the whole team on early preview and early forecasts from our large customers, without which we couldn't do proper planning even before the put in their purchase orders, we've got to have a good idea of what they want. And you're seeing that reflected in Chantelle's purchase commitment. So when it comes to our large and intimate customer engagement, they understand, and they've gotten burned with the 2022 supply crisis and are absolutely planning with us.

本，你也知道，預測是一門非常精細的科學。我幾乎總是做得不好。因此，我和泰森、肯以及整個團隊一樣，都依賴大客戶的早期預覽和早期預測，如果沒有這些，我們甚至在他們下達採購訂單之前都無法進行適當的規劃，我們必須清楚地了解他們想要什麼。這點從 Chantelle 的購買行為就能看出來。所以，當我們與客戶進行大規模、密切的互動時，他們能夠理解，並且他們已經從 2022 年的供應鏈危機中吸取了教訓，他們絕對會與我們一起制定計劃。

Some of that is true in Todd's areas, too, with the large enterprises because in a large data center, you have to plan ahead. It's not like they miraculously show up. They need power, they need space, and those are one- or two-year lead times. Where we have to be more vigilant and this is something Todd and my campus and the entire manufacturing team is working on is realized as a campus business, we had one of our best quarters. Congratulations, Todd, Kumar, this quarter on the campus.

在 Todd 負責的領域，大型企業也存在一些類似的情況，因為在大型資料中心，你必須事先做好規劃。他們不會憑空出現。他們需要電力，他們需要空間，而這些都需要一到兩年的準備時間。我們必須更加警惕，而托德、我的校園以及整個製造團隊正在努力實現這一點，作為一項校園業務，我們取得了最好的季度業績之一。恭喜 Todd 和 Kumar，本季在校園表現出色。

That planning cycle is a lot shorter. That tends to be diving weeks, not months or half a year or longer, right? So we're working again on this dichotomy in our business and planning as much as we can for the AI but also planning ahead as much as we can for the enterprise and campus. Would you like to add something, Todd?

規劃週期短得多。那通常是指潛水幾週，而不是幾個月、半年甚至更長時間，對吧？因此，我們正在再次努力解決業務中的這種二元性問題，盡可能為人工智慧進行規劃，同時也盡可能為企業和園區進行規劃。托德，你還有什麼要補充的嗎？

Yes. I'll just add, we are getting aggressive as Jayshree said on to improve our CapEx lead times and really accelerate that business and help drive the enterprise growth that we feel pretty passionately about. And the only other thing is that the investment here and the amount of dollars being put into purchasing taking the purchase commitment is key as we look for improvement in that.

是的。我還要補充一點，正如 Jayshree 所說，我們正在積極採取措施，縮短資本支出週期，真正加快業務發展，並推動我們非常熱衷的企業成長。還有一點要注意的是，這裡的投資以及投入採購的資金數量，以及採購承諾，對於我們尋求改進至關重要。

Thanks, Jayshree and Todd. That concludes Arista Networks' third quarter 2025 earnings call. We have posted a presentation that provides additional information on our results, which you can access on the Investors section of our website. Thank you for joining us today and for your interest in Arista.

謝謝Jayshree和Todd。Arista Networks 2025 年第三季財報電話會議到此結束。我們已發布一份演示文稿，提供有關我們業績的更多信息，您可以在我們網站的投資者關係部分訪問該演示文稿。感謝您今天蒞臨本站，也感謝您對 Arista 的關注。

Thank you for joining. Ladies and gentlemen, this concludes today's call. You may now disconnect.

感謝您的參與。女士們、先生們，今天的電話會議到此結束。您現在可以斷開連線了。