Arista Networks Inc (ANET) 2023 Q4 法說會逐字稿

Welcome to the Fourth Quarter 2023 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 at the Arista website following this call. Ms. Liz Stine, Arista's Director of Investor Relations, you may begin.

歡迎參加 Arista Networks 2023 年第四季財務業績電話會議。 （操作員說明）謹此提醒，本次會議正在錄製中，並可在本次電話會議後在 Arista 網站的投資者關係部分重播。 Arista 投資者關係總監 Liz Stine 女士，您可以開始了。

Thank you, operator. 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; Ita Brennan, Arista's outgoing Chief Financial Officer; and Chantelle Breithaupt, Arista's incoming Chief Financial Officer.

謝謝你，接線生。大家下午好，感謝您加入我們。與我一起參加今天電話會議的有 Arista Networks 董事長兼執行長 Jayshree Ullal； Ita Brennan，Arista 即將離任的財務長；以及 Arista 即將上任的財務長 Chantelle Breithaupt。

This afternoon, Arista Networks issued a press release announcing the results for its fiscal fourth quarter ending December 31, 2023. If you would like a copy of this release, you can access it online from our website.

今天下午，Arista Networks 發布了一份新聞稿，宣布截至 2023 年 12 月 31 日的第四財季業績。如果您想要此新聞稿的副本，可以從我們的網站在線訪問。

During the course of this conference call, Arista Networks management will make forward-looking statements, including those relating to our financial outlook for the first quarter of the 2024 fiscal year, longer-term financial outlook for 2024 and beyond, our total addressable market and strategy for addressing these market opportunities, including AI, customer demand trends, 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 管理層將做出前瞻性聲明，包括與我們 2024 財年第一季的財務前景、2024 年及以後的長期財務前景、我們的總目標市場和應對這些市場機遇的策略，包括人工智慧、客戶需求趨勢、供應鏈限制、零件成本、製造產量、庫存管理和我們業務的通膨壓力、交貨時間、產品創新、營運資本優化和收購的好處，這些是取決於我們在向SEC 提交的文件中詳細討論的風險和不確定性，特別是在我們最新的10-Q 表格和10-K 表格中，這可能導致實際結果與這些聲明中預期的結果有重大差異。

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. Also, please note that certain financial measures we use on this call are expressed on a non-GAAP basis and have been adjusted to exclude certain charges. We have provided reconciliations of these non-GAAP financial measures to GAAP financial measures in our earnings press release. With that, I will turn the call over to Jayshree.

這些前瞻性陳述從今天起適用，您不應依賴它們來代表我們未來的觀點。我們不承擔在本次電話會議後更新這些聲明的義務。另請注意，我們在本次電話會議中使用的某些財務指標是在非公認會計原則的基礎上表示的，並已進行調整以排除某些費用。我們在收益新聞稿中提供了這些非公認會計原則財務指標與公認會計原則財務指標的調節表。這樣，我會將電話轉給 Jayshree。

Thank you, Liz. Thank you, everyone, for joining us this afternoon for our fourth quarter 2023 earnings call. 2023 has been another memorable year for Arista. We gave initial guidance of 25% year-over-year revenue growth and instead achieved well beyond that at 33.8%, driving revenue to $5.86 billion, coupled with a record non-GAAP earnings per share for the year of $6.94, up in excess of 50% annually.

謝謝你，莉茲。感謝大家今天下午參加我們 2023 年第四季財報的電話會議。 2023 年對 Arista 來說又是值得紀念的一年。我們最初設定的營收年增 25% 的目標是，最終實現了 33.8% 的成長，將營收推至 58.6 億美元，加上當年非 GAAP 每股盈餘創紀錄的 6.94 美元，成長超過每年 50%。

Back to some Q4 specifics, we delivered revenues of $1.54 billion for the quarter with a non-GAAP record earnings per share of $2.08 due to a one-time favorable tax rate. Services and software support renewals contributed approximately 17% of revenue. Our non-GAAP gross margins of 65.4% was influenced by improving supply chain and greater enterprise mix. International contributions for the quarter registered at 22.3%, with the Americas at 77.7%. This was one of our strongest-performing international quarters in recent history.

回到第四季的具體情況，由於一度優惠的稅率，我們本季實現了 15.4 億美元的收入，非 GAAP 每股收益創歷史新高，達到 2.08 美元。服務和軟體支援續約貢獻了約 17% 的收入。我們的非 GAAP 毛利率為 65.4%，這得益於供應鏈的改善和企業組合的擴大。本季國際貢獻率為 22.3%，其中美洲為 77.7%。這是我們近代史上表現最強勁的國際季度之一。

Shifting to annual sector revenue for 2023, Cloud Titans contributed significantly at approximately 43%. Enterprises, including financials, was strong at approximately 36% while the providers were at 21%. Both Meta and Microsoft are greater than 10% customer concentration at 21% and 18%, respectively. Despite multiple CapEx reductions last year and the normal volatility of Cloud Titan and AI pivots, we cherish our privileged status with both [M&M.]

轉向 2023 年的年度產業收入，Cloud Titans 貢獻顯著，約 43%。包括金融在內的企業表現強勁，約 36%，而提供者則為 21%。 Meta 和 Microsoft 的客戶集中度均超過 10%，分別為 21% 和 18%。儘管去年多次削減資本支出，並且 Cloud Titan 和 AI 支點存在正常波動，但我們仍然珍惜我們在 [M&M] 方面的特權地位。

Speaking of AI, in fall of 2023, Andy and I attended the 50th golden anniversary of Ethernet at the Computer History Museum. It truly is a reminder of how civilian widely deployed Ethernet is with the speed increasing by orders of magnitude from a shared collision 2.95 megabits for file/print/share to a terabit Ethernet switching in the AI and ML era. AI workloads are placing greater demands on Ethernet as they have both data and compute-intensive across thousands of processes today.

說到人工智慧，2023 年秋天，我和 Andy 參加了在電腦歷史博物館舉行的乙太網路 50 週年紀念活動。它確實提醒人們，民用廣泛部署的乙太網路的速度從用於檔案/列印/共享的共享碰撞 2.95 兆位元到 AI 和 ML 時代的太比特乙太網路交換速度提高了幾個數量級。人工智慧工作負載對乙太網路提出了更高的要求，因為它們如今在數千個進程中都具有資料和運算密集。

Basically, AI at scale needs Ethernet at scale. AI workloads cannot tolerate the delays in the network because the job can only be completed after all flows are successfully delivered to the GPU clusters. All it takes is one culprit or worst-case link to throttle an entire AI workload.

基本上，大規模的人工智慧需要大規模的乙太網路。 AI工作負載無法容忍網路延遲，因為只有在所有串流成功交付到GPU叢集後才能完成作業。只需要一個罪魁禍首或最壞情況的連結就能限制整個人工智慧工作負載。

Three improvements are being pioneered by Arista and the founding members of the Ultra Ethernet Consortium to improve job completion time. Number one, packet spring. AI network topology meets packet spring to allow every flow to simultaneously access all parts of the destination. Arista is developing multiple forms of load balancing dynamically with our customers.

Arista 和超以太網聯盟的創始成員率先提出了三項改進，以縮短作業完成時間。第一，包彈簧。 AI網路拓撲與資料包彈簧相結合，讓每個串流同時存取目的地的所有部分。 Arista 正在與我們的客戶一起動態開發多種形式的負載平衡。

Two is flexible ordering. Key to an AI job completion is the rapid and reliable bulk transfer with flexible ordering using Ethernet links to optimally balance AI-intensive operations, unlike the rigid ordering of InfiniBand. Arista is working closely with its leading vendors to achieve this.

二是靈活訂購。完成 AI 作業的關鍵是快速可靠的批量傳輸以及使用乙太網路鏈路的靈活排序，以最佳地平衡 AI 密集型操作，這與 InfiniBand 的嚴格排序不同。 Arista 正在與其領先供應商密切合作以實現這一目標。

Finally, network congestion. In AI networks, there's a common in-cost congestion problem whereby multiple uncoordinated senders can send traffic to the receiver simultaneously. Arista's platforms are purpose-built and designed to avoid these kinds of hotspots, evenly spreading the load across multi-packs across a virtual output queuing VoQ losses fabric.

最後是網路壅塞。在人工智慧網路中，存在一個常見的成本擁塞問題，即多個不協調的發送方可以同時向接收方發送流量。 Arista 的平台是專門建造和設計的，旨在避免此類熱點，將負載均勻分佈在虛擬輸出排隊 VoQ 損失結構中的多套件上。

In terms of annual 2023 product lines, our core, which consists of cloud, AI and data center products are built upon our highly differentiated Arista extensible operating software system stack. It is successfully deployed across 10, 25, 100, 200 and 400 gig speeds. Our cloud networking products deliver power-efficient, high-availability zones without doubling the cost of redundancy as data centers demand insatiable bandwidth capacity and network speeds for both the front-end and back-end storage and compute clusters. The core drove approximately 65% of our revenue. We continue to gain share in our highest performance switching of 100, 200 and 400 gig ports to attain the #1 position at approximately 40-plus percent according to industry analysts.

就2023年產品線而言，我們的核心產品由雲端、人工智慧和資料中心產品組成，建立在我們高度差異化的Arista可擴展作業系統堆疊之上。它已成功部署在 10、25、100、200 和 400 gig 速度。我們的雲端網路產品可提供高能源效率、高可用性的區域，且不會使冗餘成本增加一倍，因為資料中心對前端和後端儲存和運算叢集的頻寬容量和網路速度的需求無法滿足。該核心業務約占我們收入的 65%。根據行業分析師的說法，我們繼續在最高性能的 100、200 和 400 gig 端口交換領域獲得份額，以大約 40% 以上的份額位居第一。

We have increased our 400-gig customer base of 600 customers in 2022 to approximately 800 customers in 2023. We expect both 400 and 800-gigabit Ethernet will emerge as important pilots for AI back-end GPU clusters. We are cautiously optimistic about achieving our AI revenue goal of at least $750 million in AI networking in 2025.

我們已將 2022 年 600 家客戶的 400 吉比特客戶群增加到 2023 年約 800 家客戶。我們預計 400 吉比特和 800 吉比特乙太網路都將成為 AI 後端 GPU 叢集的重要試點。我們對 2025 年實現人工智慧網路至少 7.5 億美元的人工智慧收入目標持謹慎樂觀態度。

Our second market is network adjacencies comprised of routing, replacing routers and our Cognitive Campus workspaces. We continue to make progress in campus, aiming for the $750 million revenue by 2025 that we have shared at many Analyst Days. Our investments in cognitive wired and wireless, zero-touch provisioning and the introduction of AGNI, Arista Guardian for Network Identities as well as AVA sensors of threat mitigation is resonating well with our campus customers.

我們的第二個市場是由路由、替換路由器和我們的認知校園工作空間組成的網路鄰接。我們在校園中繼續取得進展，目標是到 2025 年實現 7.5 億美元的收入，我們在許多分析師日分享了這一點。我們在認知有線和無線、零接觸配置以及 AGNI、Arista Guardian for Network Identities 以及 AVA 威脅緩解感測器方面的投資與我們的園區客戶產生了良好的共鳴。

The post-pandemic campus is seeking network-as-a-service overlays and Zero Trust network embedded in with high availability, observability and consistency across our OS and management domains. We are also successfully deployed in many routing edge and peering use cases. Just in 2023 alone, we introduced 6 EOS software releases across 600 new features and 50 platforms. In fall of 2023, we introduced WAN Routing System with a focus on scale, encryption and WAN transit routing capabilities. It has positioned us well, giving our customers a seamless enterprise LAN and WAN portfolio.

大流行後的園區正在尋求網絡即服務覆蓋和嵌入的零信任網絡，在我們的作業系統和管理域中具有高可用性、可觀察性和一致性。我們也成功部署在許多路由邊緣和對等互連用例中。光是 2023 年，我們就推出了 6 個 EOS 軟體版本，涵蓋 600 個新功能和 50 個平台。 2023 年秋季，我們推出了 WAN 路由系統，專注於規模、加密和 WAN 傳輸路由功能。它為我們提供了良好的定位，為我們的客戶提供了無縫的企業 LAN 和 WAN 產品組合。

The campus and routing adjacencies together contribute approximately 19% of revenue. Our third category is network software and services based on subscription models such as Arista A-Care, CloudVision, DANZ Monitoring Fabric or DMF observability and advanced threat sensors for network detection and response. Arista's subscription-based network services and software contributed approximately 16% of the total revenue. We surpassed 2,400 cumulative customers with CloudVision, pivotal to building a modern operating model for the enterprise. Please note that perpetual software licenses are not included here and are counted inside the core or adjacent markets.

園區和路由鄰近地區合計貢獻了約 19% 的收入。我們的第三類是基於訂閱模型的網路軟體和服務，例如 Arista A-Care、CloudVision、DANZ Monitoring Fabric 或 DMF 可觀測性以及用於網路偵測和回應的高階威脅感測器。 Arista 基於訂閱的網路服務和軟體貢獻了總收入的約 16%。我們透過 CloudVision 累積客戶數量超過 2,400 個，這對於建立企業現代營運模式至關重要。請注意，此處不包括永久軟體許可證，而是計入核心或相鄰市場。

While 2023's headline has been mostly about AI, we are pleased with the momentum of enterprise and provider customers as well. Arista continues to diversify its business globally with multiple use cases and verticals. We have more than doubled our enterprise revenue in the last 3 years and we are becoming the gold standard for client-to-cloud-to-AI networking with 1 EOS and 1 CloudVision Foundation. Our $1 million customer logos increased steadily in 2023 at approximately 35% as a direct result of our campus and enterprise momentum.

雖然 2023 年的頭條新聞主要是關於人工智慧，但我們也對企業和提供者客戶的發展勢頭感到滿意。 Arista 繼續透過多個用例和垂直領域實現全球業務多元化。在過去 3 年裡，我們的企業收入增加了一倍多，我們正在透過 1 個 EOS 和 1 個 CloudVision Foundation 成為客戶端到雲端到 AI 網路的黃金標準。我們價值 100 萬美元的客戶商標在 2023 年穩步增長約 35%，這是我們園區和企業發展勢頭的直接結果。

Three principles continue to differentiate us as we are poised to be a market share gainer in the enterprise: one, best-in-class, highly available proactive products with resilience and hitless upgrades at multiple levels; two, zero-touch automation for predictive client-to-cloud one-click operations that relies less on human staff or manual operations and is instead software-driven; and finally, prescriptive insights based on AI/ML Autonomous Virtual Assist, AVA algorithms for increased security, observability and root cause analysis.

當我們準備成為企業市場份額的獲得者時，三個原則繼續使我們脫穎而出：一是一流的、高度可用的主動產品，具有彈性和多層次的無中斷升級；第二，零接觸自動化，用於預測性客戶端到雲端的一鍵操作，減少對人員或手動操作的依賴，而是由軟體驅動；最後，基於 AI/ML 自主虛擬輔助、AVA 演算法的規範性見解，以提高安全性、可觀察性和根本原因分析。

Our foundational network data lake architecture and the ability to gather, store and process multiple modalities of network data is the only way to reconcile all the incongruent silos for network operators. While legacy vendors that are 30 to 40 years old are aiming for consolidation, Arista remains the only pure-play networking innovator, earning top spots in Forrester Wave's programmable switching and customer validation in Gartner's Voice of Customer for campus in 2023.

我們的基礎網路資料湖架構以及收集、儲存和處理多種網路資料模式的能力是協調網路營運商所有不一致的孤島的唯一方法。儘管 30 至 40 年的傳統供應商正致力於整合，但 Arista 仍然是唯一的純粹網路創新者，在 Gartner 2023 年校園客戶之聲中，在 Forrester Wave 的可編程交換和客戶驗證中名列前茅。

In December 2023, we conducted one of our largest customer events called Innovate in Vegas. While not my most favorite location, our customers and prospects found it very exciting and compelling for their network transformation initiatives. They resonate deeply with our Arista 2.0 vision, building best-of-breed data-driven networking platforms.

2023 年 12 月，我們在維加斯舉辦了最大的客戶活動之一，名為「創新」。雖然不是我最喜歡的地點，但我們的客戶和潛在客戶發現它對於他們的網路轉型計劃來說非常令人興奮和引人注目。它們與我們的 Arista 2.0 願景產生了深刻的共鳴，建構了同類最佳的數據驅動網路平台。

In summary, as we wrap up another fantastic year in 2023, I am so proud of the team's execution across multiple dimensions. They have all worked tirelessly to improve our operational metrics such as lead times, gross margin and on-time shipments. Simply put, we outpaced the industry in quality, support and innovation.

總之，當我們在 2023 年結束又一個美好的一年時，我為團隊在多個方面的執行力感到自豪。他們都孜孜不倦地努力改善我們的營運指標，例如交貨時間、毛利率和準時發貨。簡而言之，我們在品質、支援和創新方面超越了產業。

We set the direction for the future of networking, working intimately with our strategic customers. Despite limited visibility at this time, we reiterate our double-digit growth of 10% to 12% from Analyst Day, aiming for approximately $6.5 billion in 2024. With that, I'd like to turn it one last time to review our financial metrics with Ita Brennan. Ita?

我們與策略客戶密切合作，為網路的未來設定方向。儘管目前能見度有限，但我們重申從分析師日起將實現 10% 至 12% 的兩位數增長，目標是到 2024 年實現約 65 億美元。因此，我想最後一次回顧一下我們的財務指標與伊塔·布倫南。伊塔？

Thanks, Jayshree, and good afternoon. This analysis of our Q4 and full year 2023 results and our guidance for Q1 2024 is based on non-GAAP and excludes all noncash stock-based compensation impacts, certain acquisition-related charges and other nonrecurring items. A full reconciliation of our selected GAAP to non-GAAP results is provided in our earnings release.

謝謝，Jayshree，下午好。我們對 2023 年第四季和全年業績以及 2024 年第一季業績的分析基於非公認會計準則，不包括所有非現金股票薪酬影響、某些收購相關費用和其他非經常性項目。我們的收益報告中提供了我們選擇的 GAAP 與非 GAAP 績效的全面對帳。

Total revenues in Q4 were $1.54 billion, up 20.8% year-over-year and towards the upper end of our guidance of $1.5 billion to $1.55 billion. Services and subscription software contributed approximately 17% of revenue in the fourth quarter, up from 16.8% in Q3. International revenues for the quarter came in at $343.5 million or 22.3% of total revenue, up from 21.5% last quarter. This quarter-over-quarter increase largely reflected a healthy contribution from our in-region EMEA customers. Overall gross margin in Q4 was 65.4%, well above our guidance of approximately 63% and up from 63.1% last quarter. As a recap for the year, we continue to see incremental improvements in gross margin quarter-over-quarter with higher enterprise shipments and better supply chain costs somewhat offset by the need for additional inventory reserves as customers refine their forecast product mix.

第四季總營收為 15.4 億美元，年增 20.8%，接近我們指導的 15 億至 15.5 億美元的上限。服務和訂閱軟體在第四季度貢獻了約 17% 的收入，高於第三季的 16.8%。本季國際營收為 3.435 億美元，佔總營收的 22.3%，高於上季的 21.5%。這一季度環比成長在很大程度上反映了我們歐洲、中東和非洲地區客戶的健康貢獻。第四季的整體毛利率為 65.4%，遠高於我們約 63% 的指導，高於上季的 63.1%。作為今年的回顧，我們繼續看到毛利率環比逐步改善，企業出貨量增加和供應鏈成本改善在一定程度上被客戶完善其預測產品組合所需的額外庫存儲備所抵消。

Operating expenses for the quarter were $262.7 million or 17.1% of revenue, up from last quarter at $255.6 million. R&D spending came in at $165 million or 10.7% of revenue, consistent with last quarter, reflecting lower levels of new product introduction costs versus what we experienced in the first half of 2023 and what we expect for the first half of 2024. This reflects the timing of prototype and other costs associated with the development of next-generation products. Sales and marketing expenses were $83.4 million or 5.4% of revenue, up from $79 million last quarter, with increases -- increased sales compensation and travel costs. Our G&A costs came in at $14.3 million or 0.9% of revenue, up from $12.1 million last quarter, reflecting some seasonal fourth quarter spending.

本季營運費用為 2.627 億美元，佔營收的 17.1%，高於上季的 2.556 億美元。研發支出為 1.65 億美元，佔收入的 10.7%，與上季度一致，反映出新產品推出成本水準低於我們在 2023 年上半年的經驗以及我們對 2024 年上半年的預期。原型的時間表以及與下一代產品開發相關的其他成本。銷售和行銷費用為 8,340 萬美元，佔收入的 5.4%，高於上一季的 7,900 萬美元，其中銷售報酬和差旅費用增加。我們的一般管理費用為 1,430 萬美元，佔收入的 0.9%，高於上季度的 1,210 萬美元，反映了第四季度的一些季節性支出。

Our operating income for the quarter was $744 million or 48.3% of revenue. Other income and expenses for the quarter was a favorable $54.5 million, and our effective tax rate was 16.8%. The lower-than-normal quarterly tax rate reflected the release of tax reserves due to the expiration of the statute limitations and some true-up of jurisdictional earnings mix. This resulted in net income for the quarter of $664.3 million or 43.1% of revenue. Our diluted share number was 318.85 million shares, resulting in a diluted earnings per share number for the quarter of $2.08, up 47.5% from the prior year.

我們本季的營業收入為 7.44 億美元，佔營收的 48.3%。本季的其他收入和支出為 5,450 萬美元，有效稅率為 16.8%。低於正常水平的季度稅率反映了由於法規限製到期而釋放的稅收儲備以及司法管轄區收益組合的一些調整。這使得該季度的淨利潤達到 6.643 億美元，佔營收的 43.1%。我們的稀釋後股票數量為 31885 萬股，導致本季稀釋後每股收益為 2.08 美元，比上年增長 47.5%。

Now turning to the balance sheet. Cash, cash equivalents and investments ended the quarter at approximately $5 billion. We did not repurchase shares of our common stock in the quarter. To recap our repurchase program to date, we have repurchased $855.5 million or 8 million shares at an average price of $107 per share under our current $1 billion Board authorization. This leaves $144.5 million 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.

現在轉向資產負債表。本季末現金、現金等價物和投資約 50 億美元。本季我們沒有回購普通股。回顧我們迄今為止的回購計劃，根據目前 10 億美元的董事會授權，我們以每股 107 美元的平均價格回購了 8.555 億美元或 800 萬股股票。這使得未來幾季可以回購 1.445 億美元。未來回購的實際時間和金額將取決於市場和業務狀況、股價和其他因素。

Now turning to operating cash performance for the fourth quarter. We generated approximately $526.5 million of cash from operations in the period, affecting strong earnings performance, combined with some increase in deferred revenue, offset by reductions in taxes payable. [DSOs] come at 61 days, up from 51 days in Q3, reflecting the timing of shipments and seasonal strength in service renewal billings. Inventory turns were 1.07x, down slightly to 1.1x last quarter. Inventory increased slightly to $1.95 billion, reflecting the ongoing receipt and consumption of components from our purchase commitments and an increase in switch-related finished goods. Our purchase commitments at the end of the quarter were $1.59 billion, down from $2 billion at the end of Q3. We expect to continue to reduce our overall purchase commitment number. However, we will maintain a healthy position related to key components, especially as we focus on new products.

現在轉向第四季的營運現金表現。在此期間，我們從營運中產生了約 5.265 億美元的現金，影響了強勁的盈利表現，同時遞延收入有所增加，但被應付稅款的減少所抵消。 [DSO] 為 61 天，高於第三季的 51 天，反映了出貨時間和服務續訂帳單的季節性強度。庫存週轉率為 1.07 倍，略降至上季的 1.1 倍。庫存小幅增加至 19.5 億美元，反映出我們採購承諾中組件的持續接收和消耗以及與開關相關的成品的增加。本季末我們的採購承諾為 15.9 億美元，低於第三季末的 20 億美元。我們預計將繼續減少整體採購承諾數量。然而，我們將在關鍵零件方面保持健康的地位，特別是當我們專注於新產品時。

Our total deferred revenue balance was $1.51 billion, up from $1.195 billion in Q3. The majority of the deferred revenue balance is services related and directly linked to the timing and term of service contracts, which can vary on a quarter-by-quarter basis. Our product deferred revenue balance increased approximately $153 million over last quarter. This was ahead of our expectations for the quarter and yet again shows that this balance can move significantly on a quarterly basis.

我們的遞延收入餘額總額為 15.1 億美元，高於第三季的 11.95 億美元。大部分遞延收入餘額與服務相關，並與服務合約的時間和期限直接相關，而服務合約的時間和期限可能會因季度而異。我們的產品遞延收入餘額比上季增加了約 1.53 億美元。這超出了我們對本季度的預期，並再次表明這種平衡可以按季度顯著變化。

As of now, we expect the spends to decline somewhat in Q1 '24 and still be up significantly from Q3 '23 levels. Accounts payable days were 72 days, up from 44 days in Q3, reflecting the timing of inventory receipts and payments. Capital expenditures for the quarter were $6 million. I would now like to turn the call back to Jayshree. Jayshree?

截至目前，我們預計 24 年第一季的支出將有所下降，但仍較 23 年第三季的水平大幅上升。應付帳款天數為 72 天，高於第三季的 44 天，反映了庫存收付的時間。該季度的資本支出為 600 萬美元。我現在想將電話轉回 Jayshree。傑什裡？

Thank you, Ita, first of all, for an incredible 8.5 years as our Chief Financial Officer, we're going to miss you a lot, and wish you all the best in your next innings. And if you ever miss an earnings call, please come, we'll invite you for one.

首先感謝你，Ita，在擔任我們財務長的 8.5 年裡度過了令人難以置信的時光，我們會非常想念你，並祝你在下一局一切順利。如果您錯過了財報電話會議，請來，我們會邀請您參加。

Now to describe our Q1 2024 guidance, it's my pleasure to introduce our incoming Chief Financial Officer, Chantelle Breithaupt for her very first earnings call at Arista. Welcome, Chantelle.

現在，為了描述我們的 2024 年第一季指引，我很高興介紹我們即將上任的財務長 Chantelle Breithaupt，她將參加 Arista 的第一次財報電話會議。歡迎，尚特爾。

Thank you, Jayshree. Ita, congratulations on all that you achieved during your tenure at Arista. Your partnership during our transition is greatly appreciated.

謝謝你，傑什裡。 Ita，恭喜您在 Arista 任職期間所取得的一切成就。非常感謝您在我們過渡期間的合作。

Since joining Arista, I've been impressed by both the outstanding leadership team and the highly innovative engineering team to both serve a set of marquee customers that are redefining the future of networking.

自從加入 Arista 以來，傑出的領導團隊和高度創新的工程團隊給我留下了深刻的印象，他們為一群正在重新定義網路未來的大客戶提供服務。

Arista began shipping products in 2008. And in 15 years, the annual bandwidth of the data centers has grown 350-fold. In just the past 2 years, the annual bandwidth has doubled with Arista shipping, a cumulative 75 million ports in that time frame. Our acceleration of the data center switching market in recent quarters is evidenced by our market share gains in the 20-plus percent range of both ports and dollars. I am thrilled to be joining Arista at such an exciting time.

Arista於2008年開始出貨。15年來，資料中心的年頻寬成長了350倍。僅在過去 2 年中，Arista 航運的年頻寬就翻了一番，在此期間累計連接埠數量達到 7,500 萬個。近幾季我們資料中心交換市場的加速發展體現在我們的連接埠和美元市佔率均成長了 20% 以上。我很高興能在如此令人興奮的時刻加入 Arista。

Now turning to our outlook for the first quarter of 2024 and the remainder of the fiscal year. We remain confident with our Analyst Day view, which called for fiscal year 2024 revenue growth of 10% to 12%. This reflects our outlook for moderated cloud spending after multiple years of accelerated growth, combined with a continued growth trajectory in the enterprise business.

現在轉向我們對 2024 年第一季和本財年剩餘時間的展望。我們對分析師日的觀點仍充滿信心，該觀點預計 2024 財年營收將成長 10% 至 12%。這反映了我們對經過多年加速成長後雲端支出放緩以及企業業務持續成長軌跡的展望。

For gross margin, we reiterate the range for the fiscal year of 62% to 64% with Q1 '24 expected to be at the lower end due to a heavier cloud mix, including some expected release of deferred revenue. In terms of spending, we expect to invest in gross spending faster than revenue. In line with our Analyst Day view, with an operating margin of approximately 42% in 2024. This incremental investment may include go-to-market resourcing and increased new product introduction costs to support our product roadmap. This latter trend is already evident in Q1 '24 as R&D is expected to rebound from the unusually low levels in the second half of 2023.

對於毛利率，我們重申本財年的毛利率範圍為62% 至64%，預計24 年第一季毛利率將處於較低水平，原因是雲混合情況較重，包括一些預期釋放的遞延收入。在支出方面，我們預期總支出的投資速度將快於收入的投資速度。與我們分析師日的觀點一致，2024 年營業利潤率約為 42%。這種增量投資可能包括上市資源和增加的新產品推出成本，以支持我們的產品路線圖。後一種趨勢在 24 年第一季已經很明顯，因為研發預計將從 2023 年下半年的異常低水準反彈。

On the cash front, we will continue to work to reduce our working capital investments and drive some further reduction in inventory as we move through the year. Our structural tax rate is expected to remain at 21.5%, back to the usual historical rate, up from the unusually low onetime rate of 16.8% experienced last quarter Q4 FY '23. With all of this as a backdrop, our guidance for the first quarter, which is based on our non-GAAP results and excludes any noncash stock-based compensation impacts and other nonrecurring items is as follows: revenues of approximately $1.52 billion to $1.56 billion, gross margin of approximately 62% and operating margin at approximately 42%. Our effective tax rate is expected to be approximately 21.5%, with approximately 319.5 million diluted shares.

在現金方面，我們將繼續努力減少營運資本投資，並在今年內進一步減少庫存。我們的結構性稅率預計將保持在 21.5%，回到通常的歷史稅率，高於 23 財年第四季 16.8% 的異常低的一次性稅率。以此為背景，我們對第一季度的指導基於我們的非 GAAP 業績，不包括任何非現金股票薪酬影響和其他非經常性項目，如下：收入約為 15.2 億至 15.6 億美元，毛利率約為62%，營業利益率約42%。我們的有效稅率預計約為 21.5%，稀釋後股份約為 3.195 億股。

In summary, I am excited to lead Arista 2.0 journey as CFO. We will migrate our best-of-breed products to best-of-breed data-driven platforms, enabling our impressive TAM of $60 billion. With that, I now turn the call back to Liz for Q&A. Liz?

總之，我很高興能夠作為 CFO 領導 Arista 2.0 之旅。我們將把我們最好的產品遷移到最好的數據驅動平台，使我們的 TAM 達到 600 億美元。現在，我將電話轉回給莉茲進行問答。麗茲？

Thank you, Chantelle. We will now move to 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 question. Thank you for your understanding. Operator, take it away.

謝謝你，尚特爾。我們現在將進入 Arista 財報電話會議的問答部分。為了讓更多人參與，我想請大家只回答一個問題。感謝您的體諒。接線員，把它拿走。

(Operator Instructions) Your first question comes from the line of Aaron Rakers from Wells Fargo.

（操作員說明）您的第一個問題來自富國銀行的 Aaron Rakers。

Ita, it's been great working with you. I wish you the best in retirement. I guess my question is, Jayshree, just obviously, the focus on AI and the build-out of back-end networks based on 400, 800 gig Ethernet. I'm just curious like as you -- as we progress through these last 3 months, how has your views evolved? And just reminds us of the cadence of kind of product cycles that really set the table for Arista and this AI opportunity as we move through '24 and particularly into '25.

伊塔，與你合作真是太好了。我祝你退休後一切順利。我想我的問題是，Jayshree，很明顯，重點是人工智慧和基於 400、800 gig 乙太網路的後端網路的建構。我和你一樣好奇 - 隨著我們過去 3 個月的進展，你的觀點發生了怎樣的變化？只是提醒我們，當我們進入 24 年，特別是進入 25 年時，真正為 Arista 和人工智慧機會奠定基礎的產品週期的節奏。

Thank you, Aaron. And yes, we will all miss Ita. So our AI performance continues to track well for the $750 million revenue goal that we set last November at Analyst Day. To give you some color on the last 3 months, I would say difficult to project anything in 3 months. But if I look at the last year, which maybe last 12 months is a better indication, we have participated in a large number of AI bids and when I say large, I should say they are large AI bids, but there are a small number of customers actually to be more clear. And in the last 4 out of 5, AI networking clusters we have participated on Ethernet versus InfiniBand, Arista has won all 4 of them for Ethernet, one of them still stays on InfiniBand. So these are very high-profile customers. We are pleased with this progress. But as I said before, last year was the year of trials. This is the year of pilots and true production truly sets in only in 2025.

謝謝你，亞倫。是的，我們都會想念 Ita。因此，我們的 AI 表現繼續良好地實現了我們去年 11 月分析師日設定的 7.5 億美元收入目標。為了讓您了解過去 3 個月的情況，我想說在 3 個月內很難預測任何事情。但如果我看看去年，也許過去 12 個月是一個更好的跡象，我們參與了大量的人工智慧投標，當我說大型時，我應該說它們是大型人工智慧投標，但數量很少客戶實際上更清楚。在過去 5 場人工智慧網路集群中，我們在乙太網路與 InfiniBand 上參加的人工智慧網路集群中，Arista 贏得了乙太網路的所有 4 場比賽，其中一場仍然停留在 InfiniBand 上。所以這些都是非常引人注目的客戶。我們對此進展感到高興。但正如我之前所說，去年是考驗的一年。今年是試點之年，真正的量產要到 2025 年才會開始。

Your next question comes from the line of Tal Liani from Bank of America.

您的下一個問題來自美國銀行的塔爾·利亞尼 (Tal Liani)。

I'm trying to find -- because we don't have the backlog contribution of last year. I'm trying to kind of dissect the numbers and see what's the correlation with core data center business and traditional compute. So if service sales -- the service sales cycle is low, and we see some declines in servers, does it mean that at least in the short run, excluding the macro contribution, there is also a decline in the orders? Just how does it work between server demand and switching demand?

我正在努力尋找——因為我們沒有去年的積壓貢獻。我正在嘗試剖析這些數字，看看與核心資料中心業務和傳統運算之間的相關性。那麼，如果服務銷售——服務銷售週期較低，我們看到伺服器下降，是否意味著至少在短期內，排除宏觀貢獻，訂單也會下降？伺服器需求和交換需求之間到底是如何運作的？

Yes. So Tal, first of all, as you know, Ita and I or Chantelle and I would never -- really comment on bookings, orders. We find these all to be kind of useless metrics because ultimately, what matters is what we ship, which is revenue. But just to sort of answer your question on ratio of CPUs or for that matter, GPUs in the future to the network, typically, we have to have the CPUs or GPUs come in before we can outfit the network. To kind of go hand-in-hand, but as you know, in AI, we've been waiting for the GPUs. And in the last couple of years, they've been waiting for everything with a long lead time. But I would say, generally, in the leaf architecture, they go hand in hand where you could create a rack of 1,000 servers or whether they are CPUs and GPUs. And generally, they look to rack and stack the cable, the CPUs and the network together.

是的。所以，塔爾，首先，如你所知，伊塔和我或尚特爾和我永遠不會真正評論預訂、訂單。我們發現這些都是無用的指標，因為最終重要的是我們交付的內容，也就是收入。但為了回答你關於 CPU 或未來 GPU 與網路的比例的問題，通常，我們必須先安裝 CPU 或 GPU，然後才能裝備網路。兩者是齊頭並進的，但正如你所知，在人工智慧領域，我們一直在等待 GPU。在過去的幾年裡，他們一直在等待一切，等待時間很長。但我想說的是，一般來說，在葉架構中，它們是齊頭並進的，您可以創建一個包含 1,000 台伺服器的機架，或者無論它們是 CPU 還是 GPU。一般來說，他們希望將電纜、CPU 和網路架起來並堆疊在一起。

On the spine, which is our -- which connects all of our leaves, that decision can be made independently even if the processors are not available. So on the leaf, it's more correlated, on the spine, it's not.

在連接我們所有葉子的脊椎上，即使處理器不可用，也可以獨立做出該決定。所以在葉子上，它的相關性更強，而在脊柱上，則不是。

Your next question comes from the line of Sebastien Naji from William Blair.

你的下一個問題來自威廉·布萊爾的塞巴斯蒂安·納吉。

I just want to start an echo of everyone's commentary and wish you the best, Ita. It's been a pleasure. My question has to do with white box. People have been talking about the threat for white box since Arista has been around, and it hasn't really impacted Arista's ability to grow. Can you maybe articulate why you believe in the world of AI networks, more of the market would not move to white box or vice versa, maybe why more of the market would move away from white box?

我只是想回應大家的評論並祝你一切順利，Ita。這是我的榮幸。我的問題與白盒子有關。自從 Arista 出現以來，人們一直在談論白盒子的威脅，但這並沒有真正影響 Arista 的發展能力。您能否闡明為什麼您相信人工智慧網路的世界，更多的市場不會轉向白盒，反之亦然，也許為什麼更多的市場會遠離白盒？

It's a good question, Sebastien. Look, I think white box is here to stay for a very long time if somebody just wants a throwaway commodity product, but how many people want throwaway commodity in the data center? They're still mission-critical, and they're even more mission-critical for AI. If I'm going to spend multimillion dollars on a GPU cluster, and then the last thing I'm going to do is put a toy network in, right? So to put this sort of in perspective, that we will continue to coexist with a white box. There will be use cases where Arista's blue box or a stand-alone white box can run either SONiC or FBOSS but many times, the EOS software stack is really, really something they depend on for availability, analytics, automation, and there's -- you can get your network for 0 cost, but the cost of downtime is millions and millions of dollars.

這是一個很好的問題，塞巴斯蒂安。聽著，我認為如果有人只想要一次性商品，那麼白盒會在這裡停留很長一段時間，但是有多少人想要資料中心的一次性商品呢？它們仍然是任務關鍵型，對人工智慧來說更是關鍵任務。如果我要在 GPU 叢集上花費數百萬美元，那麼我要做的最後一件事就是放入玩具網絡，對吧？因此，從這個角度來看，我們將繼續與白盒共存。在某些用例中，Arista 的藍色盒子或獨立白盒可以運行 SONiC 或 FBOSS，但很多時候，EOS 軟體堆疊確實是他們在可用性、分析、自動化方面所依賴的東西，而且 - 您可以以0 成本獲得您的網絡，但停機成本卻高達數百萬美元。

So we have always embraced white box, we coexist with it, but it continues to be a relatively small use case in the larger mission-critical data centers for enterprise and cloud companies.

因此，我們一直擁抱白盒，我們與之共存，但在企業和雲端公司的大型關鍵任務資料中心中，它仍然是一個相對較小的用例。

Your next question comes from the line of Matt Niknam from Deutsche Bank.

您的下一個問題來自德意志銀行的 Matt Niknam。

Maybe a higher-level strategy question. We've seen 2 of your key networking peers scale up through sizable M&A over the last several months. So can you talk a little bit about how you view the value of such scale in order to maybe better serve and target both the cloud and AI -- Titan as well as enterprise verticals?

也許是更高層次的策略問題。我們看到你們的兩個主要網路同行在過去幾個月中透過大規模併購擴大了規模。那麼，您能否談談您如何看待這種規模的價值，以便更好地服務和瞄準雲端和人工智慧——泰坦以及企業垂直領域？

Yes. No, Matt, that's a good question. I think on the cloud and AI, we feel pretty bulked up to deal with those customers because they don't look for size and bulk. They look for, as you know, networking innovation capabilities, and this has been Arista's heritage for 10 years and will continue to be with the AI cycle for the next foreseeable 10 years.

是的。不，馬特，這是個好問題。我認為在雲端和人工智慧方面，我們感覺在與這些客戶打交道時感覺相當笨重，因為他們不追求規模和體積。如你所知，他們尋求網路創新能力，這是 Arista 10 年來的傳統，並將在未來可預見的 10 年中繼續伴隨人工智慧週期。

On the enterprise, there are multiple markets and size health. I think if you're targeting the early adopters, service has traditionally done very, very well there. And the last 3 years is a good example of how well we've done that both in the data center and in the campus. If you look at the next category of sort of the -- not necessarily the screening early adopters, but maybe the fast follower, I think Arista will continue to do well there in the large enterprise. We are so underserved and underpenetrated in both the Fortune 1000 and the Global 2000. We've got a long, long ways to go. We probably have 20% of those customers. We've got 80% of them left to go. And that I'm not even talking about the mid-market and the SMB, which is a whole lot of markets that we are underserved in.

對企業而言，存在著多個市場和規模健康。我認為，如果你的目標是早期採用者，那麼傳統上服務就做得非常非常好。過去三年是我們在資料中心和園區中做得很好的一個很好的例子。如果你看看下一個類別——不一定是篩選早期採用者，但可能是快速追隨者，我認為 Arista 將繼續在大型企業中表現出色。我們在《財星 1000 強》和《全球 2000 強》中的服務和滲透率都非常低。我們還有很長很長的路要走。我們可能擁有其中 20% 的客戶。我們還有 80% 的人要離開。我甚至沒有談論中端市場和中小企業，這是我們服務不足的一大堆市場。

So absolutely, we need to make more investments in enterprise there. When I look at what Anshul, (inaudible), Ashwin are doing, this is exactly where we're doubling down. This is exactly where we doubled down in the last 3 years, post-pandemic, and we have more than doubled our revenue and increased our logo presence because of this investment in the enterprise. I can't comment about consolidation of vendors, but vendors don't grow, 5 plus 5, sometimes is 10. But to be careful on integration, 5 plus 5 can sometimes be 7 too. So that's somebody else's responsibility, not mine. I think we can get a lot of organic growth.

因此，我們絕對需要對那裡的企業進行更多投資。當我看到 Anshul（聽不清楚）和 Ashwin 正在做的事情時，我發現這正是我們加倍努力的地方。這正是我們在大流行後的過去三年加倍努力的地方，由於對企業的這項投資，我們的收入增加了一倍多，並增加了我們的徽標存在。我無法評論供應商的整合，但供應商不會成長，5加5，有時是10。但要小心整合，5加5有時也可以是7。所以這是別人的責任，不是我的。我認為我們可以獲得很多有機成長。

Your next question comes from the line of Meta Marshall from Morgan Stanley.

您的下一個問題來自摩根士丹利的 Meta Marshall。

Jayshree, maybe just a question. You noted limited visibility and understand that this early -- during the year. But would you say that it's timing of when some of these back-end pilots scale into production, is it kind of level of front-end spending? Is it enterprise projects? Just like where you're finding just more commentary on the visibility comment? And then second question, you guys noted on the gross margin, but it's a portion of mix and kind of supply chain costs coming down. But just if there's any one bias towards what led to the gross margin upside in the quarter.

Jayshree，也許只是一個問題。您注意到可見度有限，並了解這一點——今年早些時候。但您是否會說，現在是一些後端試點規模投入生產的時機，這是否是前端支出的水平？是企業專案嗎？就像您在哪裡找到有關可見性評論的更多評論一樣？第二個問題，你們注意到了毛利率，但這是混合和供應鏈成本下降的一部分。但如果對導致本季毛利率上升的因素存在任何偏見的話。

Yes, maybe I'll take that last one first. I mean a lot of the upside in the fourth quarter was really just customer mix, right? I mean we awaited heavily towards enterprise in Q4, not for any particular reason. It just happened to be that way and that kind of dropped the margins higher.

是的，也許我會先拿最後一個。我的意思是，第四季度的許多好處實際上只是客戶組合，對吧？我的意思是，我們在第四季對企業版充滿期待，並不是出於任何特殊原因。碰巧就是這樣，這導致了利潤率的下降。

And Meta, to answer your question on enterprise and AI activity, I think Arista continues to drive the concept of EOS, multi-domain routing, campus, high availability, mission-critical enterprises for multiple verticals. We're making good progress there, this is going to be the part of our mainstream innovation and go-to-market.

Meta，為了回答您關於企業和人工智慧活動的問題，我認為 Arista 繼續推動 EOS、多域路由、園區、高可用性、多個垂直領域的關鍵任務企業的概念。我們在那裡取得了良好的進展，這將成為我們主流創新和進入市場的一部分。

On the AI side, we continue to track well. I think we're moving from what I call trials, which is connecting hundreds of GPUs to pilots, which is connecting thousands of GPUs this year, and then we expect larger production clusters.

在人工智慧方面，我們繼續保持良好的追蹤。我認為我們正在從我所說的試驗（連接數百個 GPU）轉向今年的試點（連接數千個 GPU），然後我們期望更大的生產叢集。

I think one of the questions that we will be asking ourselves and our customers is how these production clusters evolve. Is it going to be 400, 800 or a combination thereof? The role of Ultra Ethernet Consortium and standards and the ecosystem all coming together, very similar to how we had these discussions in 400 gig will also play a large part. But we're feeling pretty good about the activity. And I think moving from trials to pilots this year will give us considerable confidence on next year's number.

我認為我們要問自己和客戶的問題之一是這些生產集群如何發展。是 400、800 還是兩者的組合？超乙太網路聯盟、標準和生態系統的作用都將發揮重要作用，這與我們在 400 gig 中進行的討論非常相似。但我們對這項活動感覺非常好。我認為今年從試驗轉向試點將使我們對明年的數字充滿信心。

Your next question comes from the line of James Fish from Piper Sandler.

你的下一個問題來自 Piper Sandler 的 James Fish。

Maybe Ita for you, and I'll miss having you on here, by the way, congrats on retirement. But what's causing the delay being able to ship that we saw that product deferred revenue jump as much as we did or should we think about this as normal to see this level of jump in Q4 is based on what you've disclosed in the past. It doesn't seem like this is a normal jump. I guess what's the hang-up and with supply chain starting to go the other way, it's supplied more readily available, could we actually see the price increases you guys have enacted in the past now have to be given back at some point in '24 or '25?

也許 Ita 適合你，我會想念你在這裡的，順便說一句，恭喜你退休。但是，導致發貨延遲的原因是我們看到該產品遞延收入的增長與我們一樣多，或者我們是否應該將其視為正常現象，以看到第四季度的這種增長水平，這是基於您過去揭露的內容。這看起來不像是正常的跳躍。我想問題是什麼，隨著供應鏈開始走向相反的方向，它的供應更加容易獲得，我們是否真的可以看到你們過去製定的價格上漲現在必須在 24 年的某個時候歸還或'25？

Yes, Jim, I think the deferred, if you think back to how this works, I mean obviously, it's been shipped for it to actually be in deferred, right? So I think that's -- it's just timing. And we've talked about this last over the past -- I'm sure Chantelle is going to talk about it, again, in the future, right, is that, it really is just purely timing of shipments and where we have some new type projects, new capabilities that we're trialing with the customer, that's causing it to get caught in the deferred. But it's not a fundamental underlying driver of the business.

是的，吉姆，我認為延期，如果你回想一下這是如何運作的，我的意思是顯然，它已經被運送，實際上是延期的，對吧？所以我認為這只是時機。我們過去已經討論過這個問題 - 我相信 Chantelle 將來會再次討論這個問題，對吧，這真的只是純粹的發貨時間以及我們有一些新的東西類型項目，我們正在與客戶一起試驗的新功能，這導致它陷入延期。但這並不是業務的根本驅動力。

I think on pricing and the very little that's happening in terms of pricing adjustments, that's kind of out of the order, just normal pricing environment where we continue to compete for business. I don't think there's anything particularly different there that we've seen.

我認為在定價方面以及在定價調整方面發生的很少的事情，這有點不合常理，只是我們繼續競爭業務的正常定價環境。我認為我們沒有看到什麼特別不同的地方。

Your next question comes from the line of Ittai Kidron from Oppenheimer.

您的下一個問題來自奧本海默的 Ittai Kidron。

Ita, I'll miss you and Chantelle, good luck, of course, to you in your new role. I have a couple of ones for me. First of all, on the cloud mix, it kind of declined a little bit on the year. Maybe you can tell us what are your underlying working assumptions for '24? And then more broadly on the '24 guide, Chantelle, it feels like you're talking about $600 million increase year-over-year in revenue. It feels like half of it can already come from the AI networking, given your '25 targets, and you seem very comfortable about your '25 targets, I would think your '24 should be comfortable as well. So why -- if I assume that $200 million, $300 million come from AI networking this year, why should the rest of the business generate only $300 million to get to your annual targets? Why such an aggressive conservatism here on the guidance?

Ita，我會想念你和 Chantelle，當然，祝你在新職位上好運。我有幾個給我的。首先，在雲組合方面，今年略有下降。也許您可以告訴我們您對「24」的基本工作假設是什麼？然後，從更廣泛的角度來看，Chantelle，在 24 小時指南中，感覺您正在談論收入同比增長 6 億美元。感覺一半已經可以來自人工智慧網絡，考慮到你的 '25 目標，而且你似乎對你的 '25 目標很滿意，我認為你的 '24 也應該很舒服。那麼為什麼——如果我假設今年 2 億美元、3 億美元來自人工智慧網絡，為什麼其他業務只產生 3 億美元來實現你的年度目標？為什麼指導意見如此激進保守？

Okay, Ittai, let me take the first question, and then I'll pass it over to Ita and Chantelle for what you call conservatism. So first of all, our cloud mix is very strong, very good. But I think what you should take away from this is not that our cloud mix came down, but our enterprise did really, really well. And since 100% is the total time when something does really well, then the others look less so. So we're doing well on all 3 sectors and we're very proud of the enterprise momentum. AI is going to come. It is yet to come -- certainly in 2023, as I've said to you many, many times, it was a very small part of our number, but it will gradually increase. Okay, which are one of my fantastic CFOs wants to take the conservatism question?

好的，Ittai，讓我回答第一個問題，然後我將其轉交給 Ita 和 Chantelle，因為你們所謂的保守主義。首先，我們的雲端組合非常強大、非常好。但我認為你應該從中得到的並不是我們的雲端組合下降了，而是我們的企業做得非常非常好。由於 100% 是某件事做得非常好的總時間，所以其他事情看起來就沒那麼好。因此，我們在所有三個領域都做得很好，我們對企業的發展勢頭感到非常自豪。人工智慧即將到來。它尚未到來——當然在 2023 年，正如我多次對你們說過的那樣，它只占我們人數的很小一部分，但它會逐漸增加。好的，我的哪位出色的財務長想回答保守主義問題？

Thank you for the well wishes. I think coming into 2024, it's a balanced view in the sense that we want to have multiple options to get to our year and so we'll work through what those mixes are and how to get to that performance that we've laid out for our guidance. I think that Jayshree very eloquently put in the sense of '23, '24, '25 on what we expect from AI going from trials to pilots to production. And so we'll work through what that means in 2024. But I think to change anything in Q1 at this time, we're just going to go a quarter at a time, especially with me coming in and we'll see how the year progresses.

謝謝您的良好祝愿。我認為進入 2024 年，這是一個平衡的觀點，因為我們希望有多種選擇來實現這一年，因此我們將研究這些組合是什麼以及如何實現我們所規劃的性能我們的指導。我認為 Jayshree 非常雄辯地表達了「23」、「24」、「25」的含義，表達了我們對人工智慧從試驗到試點再到生產的期望。因此，我們將在 2024 年研究這意味著什麼。但我認為，此時要改變第一季的任何內容，我們只會一次進行一個季度，特別是在我加入的情況下，我們將看到如何年進步。

Your next question comes from the line of Alex Henderson from Needham & Company.

您的下一個問題來自 Needham & Company 的 Alex Henderson。

Ita, I can't believe you're leaving us. I'm going to miss you.

伊塔，我不敢相信你要離開我們。我會想念你的。

So the question I have really is what are you hearing from the field, particularly in the enterprise segment? There's been a lot of noise about indigestion of large amounts of volume that have been shipped to various companies. And clearly, there's some concern that there's some oversupply over the last year into the enterprise market. And I think you've talked to a lot of CEOs. What are they telling you in terms of where their IT spending intentions are for '24? Where are they saying the spending is going relative to the networking gear versus alternative spending priorities?

所以我真正的問題是，您從該領域，特別是在企業領域聽到了什麼？關於運往各公司的大量貨物消化不良的聲音很多。顯然，人們擔心去年企業市場出現了供應過剩的情況。我認為您已經與許多執行長交談過。他們告訴您什麼是 24 世紀的 IT 支出意圖？他們說相對於網路設備和其他支出優先事項的支出流向在哪裡？

That's a good question, Alex. I certainly talked to a lot of CIOs and CEOs. And if I rewind the clock to January last year, I think price was a lot (inaudible), we were going through this whole financial crisis, Silicon Valley Bank, this, that, the other. And if I now fast forward to a year later, our momentum in the enterprise is actually stronger now than it was a year ago. So all this distinct customers are looking for that innovation, modern network model, CI/CD principles, bringing DevOps, NetOps, SecOps, all of this together. And so Arista continues, in my view, with the large TAM we have in the enterprise of at least $30 billion out of that $60 million to have the opportunity to really deliver that vision of client to cloud, break down the operational silos. And I would say today, the CIOs recognize us as the pure-play innovator more than any other company.

這是個好問題，亞歷克斯。我當然與許多首席資訊長和執行長進行過交談。如果我把時鐘倒回去年一月，我認為價格很高（聽不清楚），我們正在經歷整個金融危機，矽谷銀行，這個，那個，另一個。如果我現在快進到一年後，我們在企業中的勢頭實際上比一年前更強。因此，所有這些不同的客戶都在尋求創新、現代網路模型、CI/CD 原則，將 DevOps、NetOps、SecOps，所有這些結合在一起。因此，在我看來，Arista 繼續利用我們在 6,000 萬美元中至少有 300 億美元的企業中擁有的大型 TAM，有機會真正將客戶願景交付到雲端，打破營運孤島。今天我想說的是，資訊長們比其他任何公司都更認可我們作為純粹的創新者。

Your next question comes from the line of Atif Malik from Citi.

您的下一個問題來自花旗銀行的 Atif Malik。

Thanks for providing that comments on 4 wins against InfiniBand. Now your networking competitor announced a collaboration with NVIDIA on Ethernet AI enterprise solutions last week. Can you talk about what this means for your Ethernet back-end business, if anything?

感謝您提供有關 4 場戰勝 InfiniBand 的評論。現在，您的網路競爭對手上週宣布與 NVIDIA 就乙太網路 AI 企業解決方案進行合作。您能談談這對您的乙太網路後端業務意味著什麼嗎？

Yes. I don't understand the announcement as well as probably my competitor does. I think it has more to do with UCS and Cisco-validated designs. Specific to our partnership, you can be assured that we'll be working with the leading GPU vendors. And as you know, NVIDIA has 90% or 95% of the market. So Jensen and I are going to partner closely. It is vital to get a complete AI network design going. We will also be working with our partners in AMD and Intel so we will be the Switzerland of XPUs, whatever the GPU might be, and we look to supply the best network ever.

是的。我對這個公告的理解可能不如我的競爭對手。我認為這與 UCS 和思科驗證的設計有更多關係。具體到我們的合作關係，您可以放心，我們將與領先的 GPU 供應商合作。如您所知，NVIDIA 擁有 90% 或 95% 的市佔率。因此，詹森和我將密切合作。讓完整的人工智慧網路設計順利進行至關重要。我們也將與 AMD 和 Intel 的合作夥伴合作，因此無論 GPU 是什麼，我們都將成為 XPU 領域的瑞士，並且我們希望提供有史以來最好的網路。

Your next question comes from the line of Tim Long from Barclays.

你的下一個問題來自巴克萊銀行的提姆朗。

Yes, Ita, going to miss you as well, good luck. So I wanted to follow up a little bit more on that AI, Jayshree. You talked about those wins. Could you just talk a little bit about -- a little bit more color there? Do you think these deployments are going to be more sole-sourced or will there be multiple vendors? Did you face kind of a different competitive landscape than normal in these? And what are you thinking about breadth of this business? I'm sure it's a lot of the really large customers as you said right now. But can you talk a little bit about how you see this moving into whether it's other service providers or the enterprise vertical?

是的，伊塔，也會想念你的，祝你好運。所以我想對 AI Jayshree 進行更多的跟進。你談到了那些勝利。可以談談——多一點顏色嗎？您認為這些部署會更單一來源還是會有多個供應商？您在這些方面是否面臨與平常不同的競爭格局？您對這項業務的廣度有何看法？我確信正如您現在所說，有很多真正的大客戶。但您能否談談您如何看待這種趨勢，無論是其他服務提供者還是企業垂直領域？

Yes. Thanks, Tim. Okay. So let me just step back and say the first real consultative approach from Arista is to provide our expertise on how to build a robust back-end AI network. And so the whole discussion of Ethernet become -- versus InfiniBand becomes really important because as you may recall, a year ago, I told you we were outside looking in, everybody had an Ethernet -- everybody had an InfiniBand HPC cluster that was kind of getting bundled into AI. But a lot has changed in a year. And the popular product we are seeing right now and the back-end cluster for our AI is the Arista 7800 AI spine, which in a single chassis with north of 500 terabit of capacity can give you a substantial number of ports, 400 or 800. So you can connect up to 1,000 GPUs just doing that. And that kind of data parallel scale-out can improve the training time dimensions, large LLMs, massive integration of training data. And of course, as we shared with you at the Analyst Day, we can expand that to a 2-tier AI leaf and spine with a 16-way CMP to support close to 10,000 GPUs nonblocking.

是的。謝謝，蒂姆。好的。因此，我退一步說，Arista 的第一個真正的諮詢方法是提供我們關於如何建立強大的後端 AI 網路的專業知識。因此，關於以太網與InfiniBand 的整個討論變得非常重要，因為正如您可能還記得，一年前，我告訴過您，我們從外部看，每個人都有一個以太網，每個人都有一個InfiniBand HPC 集群，這有點像與人工智慧捆綁在一起。但一年內發生了很多變化。我們現在看到的流行產品和 AI 後端叢集是 Arista 7800 AI 主幹，它在容量超過 500 太比特的單一機箱中可以為您提供大量連接埠（400 或 800 個）。因此，您最多可以連接1,000 個GPU。而這種資料並行橫向擴展可以提高訓練時間維度、大型LLM、訓練資料的海量整合。當然，正如我們在分析師日與您分享的那樣，我們可以將其擴展到具有 16 路 CMP 的 2 層 AI 葉子和主幹，以支援近 10,000 個 GPU 無阻塞。

This lossless architecture for Ethernet. And then the overlay we will have on that with the Ultra Ethernet Consortium in terms of congestion controls, packet spring and working with a suite of [UC] mix is what I think will make Ethernet the default standard for AI networking going forward. Now will it be sole sourced, I -- I would be remiss if I didn't tell you that our cloud networking isn't sole sourced. So probably our AI won't be (inaudible). But today's models are moving very rapidly, relying on a high bandwidth, predictable latency, the focus on application performance requires you to be sole sourced initially. And over time, I'm sure it'll move to multiple sources, but I think Arista is very well positioned for the first innings of AI networking, just like we were for the cloud networking decade.

這種無損乙太網路架構。然後，我們將在擁塞控制、數據包彈簧和使用一套[UC]混合方面與超以太網聯盟進行疊加，我認為這將使以太網成為未來人工智慧網路的預設標準。現在它會是唯一來源的嗎？如果我沒有告訴你我們的雲端網路不是唯一來源的，那我就是失職了。所以我們的人工智慧可能不會（聽不清楚）。但當今的模型發展非常迅速，依賴高頻寬、可預測的延遲，對應用程式效能的關注要求您最初是唯一的來源。隨著時間的推移，我確信它會轉移到多個來源，但我認為 Arista 在人工智慧網路的第一局中處於非常有利的位置，就像我們在雲端網路十年中一樣。

And one other thing I want to say is, although a lot of these customers are doing AI pivots, these AI pivots will result in revisiting the front-end cloud network, too. So this AI anatomy is being really well understood. And if you take a deep look at the centerpiece of it, which is all the GPUs, they have to connect to something very reliable and this is really where we come in. And so this -- being actively involved has -- is going to pay a lot of dividends, but we're still very much in our first innings of AI.

我想說的另一件事是，雖然很多客戶都在做人工智慧樞紐，但這些人工智慧樞紐也將導致重新審視前端雲端網路。所以這個人工智慧的解剖結構已經被很好地理解了。如果你深入研究它的核心，即所有 GPU，它們必須連接到非常可靠的東西，這正是我們的切入點。因此，積極參與這一點將會支付了很多紅利，但我們仍處於人工智慧的第一局。

Your next question comes from the line of Ben Reitzes from Melius Research.

您的下一個問題來自 Melius Research 的 Ben Reitzes。

And obviously, Ita, it's been great working with you. Thanks for all you've done for us. I wanted to ask about your guidance and the conservatism from another lens here. With regard to 2024, since November 9 Analyst Day, some things have changed. Microsoft, Meta and Google have all raised their CapEx forecast for 2024. Obviously, your guidance for 2024 stays the same, and I know you're usually conservative. And then for 2025, AMD upped their TAM very significantly for AI and -- by a multiple. And I guess they're seeing something that many of us are seeing with regard to the future demand. And you've kept your guidance at $750 million. I just -- with that backdrop and the changes since November 9 and you guys keeping your guidance, and I understand you're conservative, do you mind addressing your conservativism or your guidance from those lenses, both with regard to '24 and '25, Jayshree?

顯然，Ita，與你合作非常愉快。感謝您為我們所做的一切。我想從另一個角度詢問您的指導和保守主義。關於2024年，自11月9日分析師日以來，有些事情發生了變化。微軟、Meta 和谷歌都上調了 2024 年的資本支出預測。顯然，你們對 2024 年的指導保持不變，我知道你們通常很保守。然後到 2025 年，AMD 大幅提高了 AI 的 TAM，並且提高了數倍。我想他們看到的東西和我們許多人看到的有關未來需求的東西一樣。您將指導價維持在 7.5 億美元。我只是──在這樣的背景下，自11 月9 日以來發生的變化，你們保持你們的指導，我知道你們是保守派，你們介意從這些角度來闡述你們的保守主義或你們的指導嗎，無論是關於“24”還是“25” ，傑什裡？

So Ben, I'm going to let my 2 CFOs speak to the conservatism, and then I'll add more color, how about that? Who wants to go first...

Ben，我將讓我的兩位財務長談論保守主義，然後我將添加更多色彩，怎麼樣？誰想先走...

I think that a change from November to January, February time frame, I don't think it would change our guidance on the year. Kind of similar to the question before. I think that our guide right now resembles where we think we're at in the sense of what will materialize in '24, we'll take it 1 quarter at a time. The reflections of the changes you're mentioning, the timing of that, we have to wait and see. There's no guarantee that's within our 12-month guidance time frame, and we'll watch and wait and see, but Jayshree?

我認為從 11 月到 1 月、2 月的時間框架的變化，我認為這不會改變我們對今年的指導。和之前的問題有點類似。我認為我們現在的指南類似於我們認為在 24 年將實現的情況，我們將一次一個季度。你提到的變化的反映，以及發生的時間，我們必須拭目以待。無法保證這會在我們 12 個月的指導時間範圍內，我們將拭​​目以待，但 Jayshree 呢？

Yes. I think that says it all. I mean, all the drivers that you mentioned are great drivers, the timing of everything that's always complex, right? So we'll take it a quarter at time and see how things play out.

是的。我認為這一切都說。我的意思是，你提到的所有車手都是偉大的車手，所有事情的時機總是很複雜，對吧？所以我們會花一個季度的時間看看事情會如何發展。

And look, if our conservatism changes to more optimism in the second half or more likely in 2025, we'll keep you posted.

看，如果我們的保守主義在下半年或更可能在 2025 年變得更加樂觀，我們會及時通知您。

Operator, we have time for one last question.

接線員，我們還有時間回答最後一個問題。

Your final question comes from the line of Karl Ackerman from BNP Paribas.

你的最後一個問題來自法國巴黎銀行的卡爾·阿克曼。

Good evening from Paris. So there have been several companies with me, optimal supply chain that indicate the market for 800 gig and early deployments of [1.60] ports will begin to inflect later this year for actually front-end networks. And so I guess why would I be wrong to conclude that your hardware sales would be a leading indicator of that? And I guess, as a result, shouldn't cloud tighten revenue grow at least in line with your outlook for 2024 of double-digit growth?

來自巴黎的晚上好。因此，有幾家公司與我合作，最佳供應鏈表明 800 gig 市場和 [1.60] 連接埠的早期部署將在今年稍後開始影響實際的前端網路。所以我想為什麼我得出的結論是你的硬體銷售將是一個領先指標是錯誤的？我想，因此，雲端運算收入成長不應該至少與您對 2024 年兩位數成長的展望一致嗎？

Yes. Thank you, Karl. Again, I'll step -- history is a good indicator of future. And if you look at our 400-gig, everybody asked me the same question, how come 400 gig, is it taking off in 2019 or '20? And it turned out, it took our ecosystem several years and of course, the pandemic didn't help whether it was optics or NICs or the whole entire thing to come together. And I don't doubt we will have trials for 800 gig this year, but I think real production, 800 gig will happen in 2025. I'd like to be proven wrong and maybe it'll come in sooner in which case, like I said, we'll let you know. But at the moment, this is our best case prediction.

是的。謝謝你，卡爾。我再說一次——歷史是未來的良好指標。如果你看看我們的 400 場演出，每個人都問我同樣的問題，為什麼 400 場演出，它會在 2019 年還是 20 年起飛？事實證明，我們的生態系統花了幾年的時間，當然，無論是光學器件、網卡還是整個系統，這場流行病都沒有幫助我們整合起來。我毫不懷疑今年我們將進行800 場演出的試驗，但我認為真正的生產，800 場演出將在2025 年進行。我希望被證明是錯誤的，也許在這種情況下它會更快推出，比如我說，我們會讓你知道的。但目前，這是我們最好的情況預測。

This concludes the Arista Networks Fourth Quarter 2023 Earnings Call. We have posted a presentation, which provides additional information on our results, which you can access on the Investors section of our website. Thank you for joining us today, and thank you for your interest in Arista.

Arista Networks 2023 年第四季財報電話會議到此結束。我們發布了一份演示文稿，其中提供了有關我們業績的更多信息，您可以在我們網站的投資者部分訪問這些信息。感謝您今天加入我們，並感謝您對 Arista 的興趣。