Arista Networks Inc (ANET) 2023 Q3 法說會逐字稿

Welcome to the Third Quarter 2023 Arista Networks Financial Results Earnings Conference Call.

歡迎參加 2023 年第三季 Arista Networks 財務業績收益電話會議。

(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.

（操作員指示）提醒一下，本次會議正在錄音，電話會議結束後可從 Arista 網站的投資者關係部分重播。

Ms. Liz Stine, Arista's Director of Investor Relations, you may begin.

您可以請 Arista 投資者關係總監 Liz Stine 女士開始發言了。

Thank you, operator. Good afternoon, everyone, and thank you for joining us. With me on today's call are Jayshree Ullal, Arista Networks President and Chief Executive Officer; and Ita Brennan, Arista's Chief Financial Officer. This afternoon, Arista Networks issued a press release announcing the results for its fiscal third quarter ending September 30, 2023. If you would like a copy of the release, you can access it online at our website.

謝謝您，接線生。大家下午好，感謝大家的參與。今天與我一起參加電話會議的還有 Arista Networks 總裁兼執行長 Jayshree Ullal；以及 Arista 財務長 Ita Brennan。今天下午，Arista Networks 發布了一份新聞稿，宣布了截至 2023 年 9 月 30 日的第三財季業績。

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 2023 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 acquisition, 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. 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.

在本次電話會議期間，Arista Networks 管理層將做出前瞻性陳述，包括與 2023 財年第四季度的財務展望、2024 年及以後的長期財務展望、我們的總體目標市場以及應對這些市場機會的戰略有關的陳述，包括人工智能、客戶需求趨勢、供應鏈約束、組件成本、製造生產、庫存管理和通脹壓力對我們庫存管理和通脹的好處詳細討論的風險和不確定性的影響，特別是在我們最近的 10-Q 表和 10-K 表中，並且可能導致實際結果與這些陳述預期的結果有重大差異。這些前瞻性陳述自今天起適用，您不應依賴它們來代表我們未來的觀點。我們不承擔本次電話會議後更新這些聲明的義務。

Also, please note that certain financial measures we use on the 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 earnings press release.

另請注意，我們在電話會議中使用的某些財務指標是以非 GAAP 為基礎表示的，並且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 GAAP 財務指標與 GAAP 財務指標的對帳表。

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

說完這些，我會把電話轉給 Jayshree。

Thank you, Liz, and happy Halloween, everyone. We delivered revenues of $1.51 billion for the quarter, with a non-GAAP earnings per share of $1.83. Services and software support renewals contributed approximately 16.8% of revenue. Our non-GAAP gross margins of 63.1% was influenced by improving supply chain overheads and higher enterprise contribution. As we have said before, gross margins have consistently improved every quarter this year and will stabilize next year in 2024.

謝謝你，Liz，祝大家萬聖節快樂。本季度，我們的營收為 15.1 億美元，非 GAAP 每股收益為 1.83 美元。服務和軟體支援續約貢獻了約16.8%的收入。我們的非公認會計準則毛利率為 63.1%，受到供應鏈管理費用改善和企業貢獻度增加的影響。正如我們之前所說，今年毛利率每季都在持續提高，並將在明年（2024年）趨於穩定。

International contribution registered at 21.5% with Americas at 78.5%. As predicted, Arista's supply chain and lead times are improving steadily in 2023, and we expect it to normalize in 2024. We are now projecting 33% annual growth versus our prior Analyst Day forecast of 25% growth for the 2023 calendar year.

國際貢獻率為 21.5%，其中美洲貢獻率為 78.5%。正如預測的那樣，Arista 的供應鏈和交貨時間在 2023 年將穩步改善，我們預計它將在 2024 年恢復正常。

During the past year, our Cloud Titan customers have been planning a different mix of AI networking and classic cloud networking for their compute and storage clusters. Our historic classification of our Cloud Titan customers has been based on industry definition of customers with or likely to attain greater than 1 million installed compute service.

在過去的一年裡，我們的 Cloud Titan 客戶一直在為他們的運算和儲存叢集規劃不同的 AI 網路和經典雲端網路組合。我們對 Cloud Titan 客戶的歷史分類是基於行業定義，即已安裝或可能安裝超過 100 萬個運算服務的客戶。

Looking ahead, we will combine cloud and AI customer spend into one category called Cloud and AI Titan sector. And as a result of this combination, Oracle OCI becomes a new member of the sector, while Apple shift to cloud specialty providers. This new Cloud and AI Titan sector is projected to represent greater than 40% of our total revenue mix due to the favorable AI investments expected in the future.

展望未來，我們將把雲端和人工智慧客戶支出合併為一個類別，稱為「雲端和人工智慧泰坦」部門。而此次合併的結果是，Oracle OCI成為該領域的新成員，而蘋果則轉向雲端專業供應商。由於預計未來人工智慧投資將有利，這個新的雲端和人工智慧巨頭部門預計將占我們總收入的 40% 以上。

In terms of enterprise momentum, Arista continues to focus on multi-domain, modern software with architectural superiority based on our single EOS, extensible operating system and CloudVision stack. This is truly a unique foundation and differentiator. We have demonstrated our strong execution and uncompromised quality with predictable release cadence that our customers have come to enjoy and appreciate. The power of our one consistent software stack across a breadth of use cases, be the WAN routing, campus, branch or data center infrastructure is truly unmatched by our industry peers.

在企業發展動力方面，Arista 繼續專注於基於單一 EOS、可擴展作業系統和 CloudVision 堆疊的具有架構優勢的多領域現代軟體。這確實是一個獨特的基礎和差異化因素。我們以可預測的發布節奏展示了我們強大的執行力和不妥協的質量，我們的客戶已經享受並欣賞。我們的一個一致的軟體堆疊在廣泛的使用案例（無論是廣域網路路由、校園、分支機構還是資料中心基礎設施）中的強大功能是業內同行無法比擬的。

Let me illustrate with a few customer wins. Our first customer win is an international one, where the customer is providing services for their interconnect of high-performance compute, HPC clusters, which are often its foundation for GPU as a service offering. Arista's Ethernet modular switch coupled with EOS, created a perfect combination of a switching platform with real-time telemetry, leveraging our EOS state-driven published subscribed model.

讓我透過幾個客戶的成功例子來說明。我們贏得的第一個客戶是國際客戶，為其高效能運算、HPC 叢集的互連提供服務，這通常是其提供 GPU 即服務的基礎。 Arista 的乙太網路模組化交換器與 EOS 結合，創造了交換平台與即時遙測的完美結合，利用了我們 EOS 狀態驅動的發布訂閱模式。

Our next [one] showcases our expansion of Arista in the public sector with their AI initiative. This grant-funded project utilizes Arista simplified operational models with CloudVision. New AI workloads require high scale, high ratings, high bandwidth and low latency as well as a need for granular visibility. This build out of a single EVPN-VXLAN based 400-gig fabric is based on deep buffers fines and underscores the importance of a lossless architecture for AI networking.

我們的下一個 [一] 展示了 Arista 透過其 AI 計劃在公共部門的擴張。此資助計畫採用了 Arista 和 CloudVision 簡化的營運模式。新的人工智慧工作負載需要高規模、高評級、高頻寬和低延遲，以及細粒度的可視性。這種基於單一 EVPN-VXLAN 的 400Gig 結構的建構是基於深度緩衝區精細，並強調了無損架構對於 AI 網路的重要性。

This last but not least, customer is an example of a campus spend. A couple of years ago, the customer was looking to do a complete refresh of their agent campus network, which comprises of 4 major headquarter campuses and several remote sites. The customer was able to leverage the Arista validated design models, AVD, all the way from data center into the campus network. The customer chose Arista because we were able to offer best-of-breed operational excellence as well as security with our 0 trust AVA sensors for threat mitigation across the entire campus of wired switches.

最後但並非最不重要的一點是，顧客是校園消費的一個例子。幾年前，客戶希望徹底更新其代理園區網絡，該網絡由 4 個主要總部園區和幾個遠端站點組成。客戶能夠利用 Arista 驗證的設計模型 AVD，從資料中心一直到校園網路。客戶選擇 Arista 是因為我們能夠透過 0 信任 AVA 感測器提供最佳的卓越營運和安全性，以緩解整個有線交換器園區內的威脅。

Arista's innovative macro segmentation, MSS, combined with leaf access and core spine delivered a compelling 2-tier Cognitive Campus solution. These 3 customers illustrate our power of the platform and software innovation for a modern network model with a low total cost of operations. We are pleased with our trajectory setting the gold standard in our industry with the lowest CVEs and vulnerabilities and the highest Net Promoter Score for cloud networking.

Arista 的創新宏觀細分 MSS 與葉子訪問和核心主幹相結合，提供了引人注目的 2 層認知園區解決方案。這 3 位客戶證明了我們的平台和軟體創新對於實現低總營運成本的現代網路模式的強大作用。我們對自己的發展軌跡感到滿意，我們以最低的 CVE 和漏洞以及最高的雲端網路淨推薦值設定了業界的黃金標準。

And with that, I'd like to hand to Ita, our CFO, for financial specifics.

接下來，我想將財務細節交給我們的財務長 Ita。

Thanks, Jayshree, and good afternoon. This analysis of our Q3 results and our guidance for Q4 '23 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。我們對第三季業績的分析和 23 年第四季的指引是基於非公認會計準則，不包括所有非現金股票薪酬的影響、某些收購相關費用和其他非經常性項目。我們在收益報告中提供了所選的 GAAP 結果與非 GAAP 結果的完整對帳表。

Total revenues in Q3 were $1.51 billion, up 28.3% year-over-year and well above the upper end of our guidance of $1.45 billion to $1.5 billion. Services and subscription software contributed approximately 16.8% of revenues in the third quarter, up from 15.2% in Q2. International revenues for the quarter came in at $324.7 million or 21.5% of total revenue, up from 20.9% last quarter. This quarter-over-quarter increase largely reflected a healthy contribution from our enterprise customers in EMEA and APAC and some reduction in domestic shipments to our Cloud Titan customers.

第三季總營收為 15.1 億美元，年增 28.3%，遠高於我們預期的 14.5 億美元至 15 億美元的上限。服務和訂閱軟體在第三季貢獻了約 16.8% 的收入，高於第二季的 15.2%。本季國際營收為 3.247 億美元，佔總營收的 21.5%，高於上一季的 20.9%。這一季度的成長主要反映了我們在歐洲、中東和非洲地區以及亞太地區的企業客戶的良好貢獻，以及我們對 Cloud Titan 客戶的國內出貨量有所減少。

Overall gross margin in Q3 was 63.1%, well above guidance of approximately 62% and up from 61.3% last quarter. 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 forecasted product mix.

第三季整體毛利率為 63.1%，遠高於預期的 62% 和上一季的 61.3%。隨著企業出貨量的增加和供應鏈成本的改善，我們繼續看到毛利率逐季逐步改善，但由於客戶改善其預測產品組合，需要增加庫存儲備，這在一定程度上抵消了這一改善。

Operating expenses for the quarter were $255.6 million or 16.9% of revenue, down from last quarter at $287.3 million. R&D spending came in at $164.4 million or 10.9% of revenue, down from $188.5 million last quarter. This primarily reflected increased headcount more than offset by lower new product introduction costs in the period.

本季營業費用為 2.556 億美元，佔營收的 16.9%，低於上一季的 2.873 億美元。研發支出為 1.644 億美元，佔營收的 10.9%，低於上一季的 1.885 億美元。這主要反映了期內員工人數的增加被新產品推出成本的降低所抵消。

Sales and marketing expense was $79 million or 5.2% of revenue, consistent with last quarter, with increased headcount and some reduction in product demo costs. Our G&A costs came in at $12.1 million or 0.8% of revenue, down from last quarter and reflecting the recovery of some bad debt amounts recorded in prior periods.

銷售和行銷費用為 7,900 萬美元，佔收入的 5.2%，與上一季持平，員工人數增加，產品展示成本減少。我們的一般及行政費用為 1,210 萬美元，佔收入的 0.8%，低於上一季度，反映了前期記錄的一些壞帳金額的收回。

Our operating income for the quarter was $696.2 million or 46.1% of revenue. Other income and expense for the quarter was a favorable $42.3 million, and our effective tax rate is 21.3%. This resulted in net income for the quarter of $581.4 million or 38.5% of revenue. Our diluted share number of -- was 317.6 million shares resulting in a diluted earnings per share number for the quarter of $1.83, up 46.4% from the prior year.

本季我們的營業收入為 6.962 億美元，佔營收的 46.1%。本季的其他收入和支出為4,230萬美元，我們的有效稅率為21.3％。這使得本季淨收入達到 5.814 億美元，佔營收的 38.5%。我們的稀釋股數為 3.176 億股，導致本季每股稀釋收益為 1.83 美元，比上年增長 46.4%。

Now turning to the balance sheet. Cash, cash equivalents and investments into the quarter at approximately $4.5 billion. We did not repurchase shares of our common stock in the quarter. To recap our repurchase program to date, we have reported $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。本季的現金、現金等價物和投資約為 45 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，我們報告稱，根據我們目前 10 億美元的董事會授權，回購金額為 8.555 億美元，即 800 萬股，平均價格為每股 107 美元。這樣剩下 1.445 億美元可供未來幾季回購。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 third quarter. We generated approximately $699 million of cash from operations in the period, reflecting strong earnings performance, combined with some increase in deferred revenue and taxes payable. DSOs come in at 51 days, up from 49 days in Q2, reflecting the strong collections quarter and a good linearity of billings.

現在來談談第三季的經營現金表現。我們在此期間從經營活動中產生了約 6.99 億美元的現金，反映了強勁的盈利表現，同時遞延收入和應付稅款有所增加。 DSO 為 51 天，高於第二季的 49 天，反映出本季收款強勁且帳單線性良好。

Inventory turns were 1.1x, down from 1.2% last quarter. Inventory remained flat to last quarter at $1.9 billion, reflecting the ongoing receipt and consumption component from our purchase commitments and an increase in switch-related finished goods. Our purchase commitments at the end of the quarter were $2 billion, down from $2.2 billion at the end of Q2.

庫存週轉率為 1.1 倍，低於上一季的 1.2%。庫存與上一季持平，為 19 億美元，反映了我們採購承諾的持續收貨和消費部分以及轉換相關成品的增加。我們在本季末的採購承諾為 20 億美元，低於第二季末的 22 億美元。

We expect the overall purchase commitment number to continue to decline as we further optimize our supply position. However, we will maintain a healthy position related to key components, especially as we focus on new products. Our total deferred revenue balance was $1.195 billion, up from $1.085 billion in Q2. The majority of the deferred revenue balances are services related and directly link to the timing and term of service tax, which can vary on a quarter-by-quarter basis. Our product deferred revenue increased by $47 million from last quarter. Account payable days were 44 days, down from 57 days in Q2, reflecting the timing of inventory received payments. Capital expenditures for the quarter were $11.2 million.

隨著我們進一步優化供應狀況，我們預計整體採購承諾數量將持續下降。然而，我們將保持與關鍵部件相關的健康地位，特別是在我們專注於新產品時。我們的遞延收入總餘額為 11.95 億美元，高於第二季的 10.85 億美元。大部分遞延收入餘額與服務有關，並與服務稅的時間和期限直接相關，而服務稅可能會按季度變化。我們的產品遞延收入較上一季增加了 4,700 萬美元。應付帳款天數為 44 天，低於第二季的 57 天，反映了庫存收到付款的時間。本季的資本支出為 1,120 萬美元。

Now turning to our outlook for the fourth quarter. Customer planning horizons for new deployments have shortened in concert with steadily improving lead time. On the supply side, we expect to continue to ship against previously committed deployment plans for some time targeting supply improvements were most needed but also careful not to create redundant customer inventory.

現在來談談我們對第四季的展望。隨著交付週期的穩定縮短，客戶新部署的規劃期限也不斷縮短。在供應方面，我們預計將在一段時間內繼續按照先前承諾的部署計劃發貨，目標是最需要改善供應，但同時也注意不要造成多餘的客戶庫存。

As outlined in our guidance, we expect to make incremental improvements to our 2023 outlook, which now calls for year-over-year revenue growth of approximately 33%. On the gross margin front, we expect gross margin of approximately 63% in the fourth quarter, reflecting ongoing supply chain and manufacturing benefits while maintaining a reasonably healthy cloud contribution.

正如我們的指引中所述，我們預計將對 2023 年的前景進行逐步改善，目前預計收入將年增約 33%。在毛利率方面，我們預計第四季度的毛利率約為 63%，這反映了持續的供應鏈和製造效益，同時保持了相當健康的雲端運算貢獻。

Turning to spending and investments. We expect to monitor the overall macro environment carefully, while engaging in targeted hiring in R&D and go-to-market as the team sees the opportunity to acquire talent. On the cash front, while increases in working capital has begun to moderate in recent quarters, our year-to-date 2023 tax payments have been deferred to October. This will represent a significant incremental use of cash in the fourth quarter at approximately $352 million.

轉向支出和投資。我們希望仔細監控整體宏觀環境，同時在團隊看到獲取人才的機會時，有針對性地招募研發和上市人員。在現金方面，雖然最近幾季營運資本的成長已開始放緩，但我們今年迄今的 2023 年稅款繳納已推遲到 10 月。這意味著第四季的現金使用量將大幅增加，達到約 3.52 億美元。

With all of this as a backdrop, our guidance for the fourth quarter, which is based on non-GAAP results and excludes any noncash stock-based compensation impacts and other nonrecurring items is as follows: Revenues of approximately $1.5 billion to $1.55 billion, gross margin of approximately 63%, operating margin of approximately 42%. Our effective tax rate is expected to be approximately 21.5% with diluted shares of approximately 319 million shares.

綜合上述因素，我們對第四季度的預期（基於非公認會計準則結果且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 億美元至 15.5 億美元，毛利率約為 63%，營業利潤率約為 42%。我們的有效稅率預計約為 21.5%，稀釋股份約為 3.19 億股。

I will now turn the call back to Liz. Liz?

我現在將電話轉回給利茲。莉茲？

Thank you, Ita. We will now move to the Q&A portion of the Arista earnings call.

謝謝你，伊塔。我們現在進入 Arista 收益電話會議的問答部分。

(Operator Instructions) Thank you for your understanding. Operator, take it away.

（操作員指示）感謝您的理解。接線員，把它拿走。

(Operator Instructions) Your first question comes from the line of Samik Chatterjee with JPMorgan.

（操作員指示）您的第一個問題來自摩根大通的 Samik Chatterjee。

Thank you for the question and congrats on the results. I guess just to keep it simple, Jayshree, if you can give us an update on when we think about the last 90 days, how have the 2 sort of verticals, Cloud Titans and enterprise sort of shown up in terms of momentum of orders and demand relative to where your expectations were 90 days ago. I know some of the cloud companies have talked about their CapEx outlook for next year as well. So an update on that would be helpful. And on the last call, you did talk about a target for double-digit growth next year. So how are you thinking on -- in relation to that number still going into the Investor Day.

感謝您的提問，並對所取得的成果表示祝賀。我想只是為了簡單起見，Jayshree，如果你可以給我們提供最新情況，當我們考慮過去 90 天時，相對於你 90 天前的預期，兩種垂直行業，Cloud Titans 和企業在訂單和需求勢頭方面表現如何。我知道一些雲端運算公司也談論了明年的資本支出前景。因此對此進行更新將會很有幫助。在上次電話會議中，您確實談到了明年實現兩位數成長的目標。那麼您對於投資人日當天的這個數字有什麼看法呢？

Okay. Thanks. Samik, first of all, we are looking forward to sharing more detail on Analyst Day. But just to reiterate, our team has always projected at least a double-digit growth for next year and years beyond. So that goal remains unchanged, and we'll share more with you. Coming back to the last 90 days, as you know, as our lead times improve, our visibility declines but we don't see significant change in improvements or declines in the last 90 days. We continue to see good momentum on enterprise, and we continue to see a good, expected push on the combination of both cloud and AI together.

好的。謝謝。 Samik，首先，我們期待在分析師日分享更多細節。但我重申一下，我們的團隊一直預測明年及以後幾年至少會達到兩位數的成長。因此這個目標保持不變，我們將與大家分享更多資訊。回顧過去 90 天，如您所知，隨著交貨時間的縮短，我們的可見度下降，但我們在過去 90 天內沒有看到改善或下降的顯著變化。我們繼續看到企業的良好發展勢頭，並且我們繼續看到雲端運算和人工智慧結合的良好、預期的推動力。

Your next question comes from the line of Antoine Chkaiban with New Street Research.

您的下一個問題來自 New Street Research 的 Antoine Chkaiban。

So accelerated AI cluster deployment is clearly weighting on traditional infrastructure deployment this year. And I'm keen to hear how sustainable you think this is because the vast majority of workloads still run on traditional infrastructure, right? So is it fair to expect a rebound in traditional infrastructure spend next year?

因此，今年加速的AI集群部署顯然對傳統基礎設施的部署產生了影響。我很想聽聽您認為這種做法有多可持續，因為絕大多數工作負載仍然在傳統基礎設施上運行，對嗎？那麼，預計明年傳統基礎建設支出將出現反彈是否合理？

Yes. Thank you, Antoine. I'll share some of my thoughts, and I'd like to hand it over to Anshul for further thoughts. We've always looked at that the cloud network as a front end and the back end. And as we said last year, many of our cloud customers are favoring spending more on the back end with AI, which doesn't mean they stop spending on front end, but they're clearly prioritized and doubled down on AI this year. My guess is as we look at the next few years, they'll continue to double down on AI. But you cannot build an AI bank cluster without thinking of the front end. So we'll see a full cycle here, while today the focus is greatly on AI and the back end of the network. In the future, we expect to see more investments in the front end as well.

是的。謝謝你，安托萬。我將分享一些我的想法，我想將其交給 Anshul 進行進一步的思考。我們一直將雲端網路視為前端和後端。正如我們去年所說的那樣，我們的許多雲端客戶都傾向於在後端人工智慧上投入更多資金，這並不意味著他們停止在前端投入資金，但他們顯然優先考慮人工智慧，並在今年加倍投入。我猜測，展望未來幾年，他們將繼續加倍投入人工智慧。但如果不考慮前端，就無法建構 AI 銀行集群。因此，我們將在這裡看到一個完整的周期，而今天的重點主要放在人工智慧和網路後端。未來，我們預計還會看到更多對前端的投資。

Jayshree, that's right, you said it's spot on, AI is everyone's priority right now, and the rest will get touched at the right time.

Jayshree，沒錯，你說得很對，AI 現在是每個人的首要任務，其餘的事情會在適當的時候得到處理。

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

您的下一個問題來自德意志銀行的 Matt Niknam。

One question, a very simple one on services. Pretty nice improvement, about 13% sequential improvement in the quarter, seasonally, I think we've seen low single digits, mid-single digits. Anything you would call out? And how are we thinking about that for the fourth quarter?

一個問題，一個關於服務的非常簡單的問題。相當不錯的改善，本季環比改善了約 13％，從季節性來看，我認為我們已經看到了低個位數、中等個位數的改善。您想喊出什麼嗎？我們對第四季有什麼看法？

Yes. Look, I think every now and again, you see kind of a pop on the services line. It's usually either somebody has consumed services faster than they intended to or we've been negotiating a contract. And then when we do actually finally signed the renewals contract, there's some flush of prior periods into the quarter. So if you look back historically, you'll see that happens from time to time. I don't think it changes the kind of fundamental growth in services. We've talked about this kind of mid to maybe a little bit higher teens growth on an ongoing basis year-over-year. I don't think it changes that. It's just -- we do have these little spikes from time to time.

是的。看，我認為時不時地，你會看到服務線上出現某種流行現象。通常情況是，有人使用服務的速度比他們預期的要快，或者我們正在協商合約。然後，當我們最終真正簽署續約合約時，本季會出現一些前期波動。所以如果你回顧歷史，你會發現這種情況時有發生。我認為這不會改變服務業的根本成長。我們已經討論過這種每年持續呈現中等到略高一點的青少年成長趨勢。我認為這不會改變這一點。只是——我們確實時不時會出現這些小高峰。

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

您的下一個問題來自法國巴黎銀行的卡爾·阿克曼。

I suppose this is a question for Ita. But is the upside in the quarter and outlook coming from a combination of better bookings and working down some of your prior backlog? Just any thoughts in terms of maybe where your backlog may end up relative to normal levels? Pre-pandemic could be super helpful.

我想這個問題該問 Ita 吧。但本季和前景的上升是否來自於訂單量的增加和部分先前積壓訂單的減少？您能想一想，相對於正常水平，您的積壓訂單最終會達到什麼程度嗎？疫情前的準備可能會非常有幫助。

Yes, Karl, we don't talk about backlog specifically. I think what we have said is as lead times improve, you expect to see some reduction in visibility to (inaudible) customers, the time where they have to place orders changes over time, right? So I think that we are seeing that dynamic. We've talked about that dynamic that we are -- as the lead times get better, we are seeing kind of customer planning horizons are shortening. We will be still deploying -- if you listen to my prepared remarks, I mean, we are still deploying equipment into next year from plans that we made some time ago. And that's just kind of, again, working with customers and laying out their plans. But in terms of giving specific numbers, we haven't done that.

是的，卡爾，我們不會具體談論積壓問題。我認為我們所說的是，隨著交貨時間的縮短，您預計會看到（聽不清楚）客戶的可見度有所降低，他們下訂單的時間會隨著時間而改變，對嗎？所以我認為我們正在看到這種動態。我們已經討論過這種動態——隨著交貨時間的縮短，我們看到客戶的規劃期限正在縮短。我們仍將繼續部署——如果你聽我準備好的發言，我的意思是，我們仍將按照前段時間制定的計劃，在明年部署設備。這只是與客戶合作並製定他們的計劃。但就給出具體數字而言，我們還沒有這樣做。

Your next question comes from the line of Amit Daryanani with Evercore.

您的下一個問題來自 Evercore 的 Amit Daryanani。

Congrats on a nice set of numbers here. I was hoping you could talk a little bit more on the enterprise side. You're seeing some really good strength over here clearly. But maybe you can talk about, is the strength more coming from campus versus the data side, maybe just qualitatively, where are you seeing better trends? And really, the context of this is, I think a lot of your peers are seeing a very severe drop in the growth rates as their backlogs have gone away. You don't seem to be having that issue. So I'm wondering like what is the offset to that and what's enabling the growth. And to the extent you can talk about campus versus the data center, that would be really helpful.

恭喜您取得了一組不錯的成績。我希望您能就企業方面再多談一點。您可以清楚的看到這裡確實存在著一些非常強大的力量。但是也許您可以談談，優勢是否更多地來自校園而不是數據方面，也許只是從品質上講，您在哪裡看到更好的趨勢？而實際上，我認為，由於積壓訂單的消失，許多同業的成長率都出現了非常嚴重的下降。您似乎沒有遇到這個問題。所以我想知道，這種抵銷作用是什麼，以及什麼推動了成長。如果你能談論校園與資料中心，那將會非常有幫助。

Okay, Amit. Again, I'll share a few words and ask Anshul to step in and say some too. Look, if you look back 3 years ago, we started seriously investing in the enterprise. And back in 2020, we had a small enterprise business that it was largely comprised of, you rightly pointed out, data center and some high-performance compute and low latency HFT, can't never forget our original heritage. But in the last 3 years, we have made an investment and seen a significant uptick in enterprise customers wanting to do business with Arista. Historically, it's been the high-tech enterprise and the financials. And today, we're seeing a much better cross-section of verticals, including health care, education, we expect to see more and more distributed enterprises.

好的，阿米特。再次，我將分享一些話，並請安舒爾介入並說一些話。如果你回顧三年前，我們就開始認真地投資企業。早在 2020 年，我們就有一家小型企業業務，其主要由資料中心和一些高效能運算和低延遲 HFT 組成，您正確指出了這一點，我們永遠不能忘記我們的初衷。但在過去 3 年裡，我們進行了投資，並且看到希望與 Arista 開展業務的企業客戶數量顯著增加。從歷史上看，它是高科技企業和金融企業。今天，我們看到垂直產業的橫斷面積越來越大，包括醫療保健、教育等，我們預期會看到越來越多的分散式企業。

And to your question on data center versus campus, the answer is yes to both. We actually see one uniform architecture where you can have a universal spine that connects into a wired leaf, a wireless leaf, a storage cluster, a compute cluster and a border leaf for routing and WAN transit. It's pretty exciting that Arista surely and remarkably setting the tone for a 2-tier leaf spine architecture across the enterprise and building that modern operating model based on CloudVision.

對於您關於資料中心與校園的問題，答案都是肯定的。我們實際上看到了一種統一的架構，其中您可以擁有一個通用主幹，它可以連接到有線葉、無線葉、儲存叢集、計算叢集和邊界葉，以進行路由和 WAN 傳輸。令人興奮的是，Arista 確實且出色地為整個企業的 2 層葉脊架構定下了基調，並建立了基於 CloudVision 的現代營運模式。

Amit, this is Anshul here. We have a great team being led by Chris Schmidt and Ashwin Kohli in this space. And now we sell in many, many countries around the world. And as Jayshree mentioned, both data center and campus customers are coming to us for the automation for the higher quality, for the visibility but able to bring to them across the board in one architecture, one OS and one CloudVision. That message resonates with every CIO today, they are no longer worried about Arista being this new kid on the block that's a risky move for them. We are, in fact, becoming the de facto and they like it. So this is why the momentum just continues. It's good execution by the team, and getting to more and more customers around the world.

阿米特 (Amit)，我是安舒爾 (Anshul)。我們在這個領域擁有一支由 Chris Schmidt 和 Ashwin Kohli 領導的優秀團隊。現在我們的產品銷往世界許多國家。正如 Jayshree 所提到的，資料中心和校園客戶都來找我們，希望實現更高品質的自動化和視覺性，並且能夠透過一個架構、一個作業系統和一個 CloudVision 為他們提供全面的服務。這個訊息如今引起了每位 CIO 的共鳴，他們不再擔心 Arista 這個新來者會為他們帶來風險。事實上，我們正​​在成為事實上的，他們喜歡這樣。這就是為什麼這種勢頭能夠持續下去。團隊的執行非常出色，並且獲得了世界各地越來越多的客戶。

Your next question comes from the line of Ben Bollin with Cleveland Research.

您的下一個問題來自克利夫蘭研究公司的 Ben Bollin。

Jayshree and Anshul, I was hoping you might be able to comment a little bit about your thoughts as you make progress in the back-end network around GPU cluster opportunity, how you see that developing versus what you've shared with us previously? And any color in particular around both preexisting and the opportunity for net new wins would be helpful.

Jayshree 和 Anshul，我希望你們能夠就你們在 GPU 叢集機會的後端網路方面取得的進展發表一些想法，與你們之前與我們分享的內容相比，你們如何看待這一發展？任何特定的顏色，無論是圍繞現有顏色還是淨新勝利機會的顏色，都會有所幫助。

Sure. Again, this is an area that Anshul lives and breathes more than I do. So I'll give you some executive comments. But Ben as I see it, the back-end network was something we didn't even see a few months or years ago, and was largely dominated by InfiniBand. Today if I look at the 5 major designs for AI networking, one of them is still very InfiniBand dominated, all the others we're looking at is -- are adopting on dual strategy of both Ethernet and InfiniBand.

當然。再說一遍，這是 Anshul 比我更生活、更活躍的領域。因此我將給你一些執行性的評論。但在我看來，後端網路是我們幾個月或幾年前都看不到的東西，並且主要由 InfiniBand 主導。今天，如果我看一下 AI 網路的 5 種主要設計，其中一種仍然以 InfiniBand 為主導，我們正在研究的所有其他設計都採用了乙太網路和 InfiniBand 的雙重策略。

So I think AI networking is going to become more and more favorable to Ethernet, particularly with the Ultra Ethernet Consortium and the work they're doing to define a spec, you're going to see more products based on UEC. You're going to see more of a connection between the back end and the front-end using IP as a singular protocol. And so we're feeling very encouraged that especially in 2025, there will be a lot of production rollout of back end and, of course, front end based on Ethernet.

因此，我認為人工智慧網路將越來越有利於以太網，特別是隨著超級以太網聯盟及其為定義規範所做的工作，您將看到更多基於 UEC 的產品。您將會看到後端和前端之間使用 IP 作為單一協定的更多連線。因此，我們感到非常鼓舞，特別是在 2025 年，將會有大量基於乙太網路的後端和前端的生產推出。

Over to you, Anshul.

交給你了，安舒爾。

Sure. Thanks, Jayshree. Ben our Cloud Titan customers as well as the specialty providers have been great partners of ours. So the level of partnership and co-development that's going on in this space, is high. It's just like in previous cycles, previous products that we've done with them. There's a lot of fine-tuning needed in these back-end networks to get the maximum utilization of DPUs. And as you know, we are good at these engineering projects.

當然。謝謝，Jayshree。 Ben，我們的 Cloud Titan 客戶以及專業供應商一直是我們優秀的合作夥伴。因此，該領域的合作與共同開發水準很高。這就像我們在之前的周期中與他們合作製作的產品一樣。這些後端網路需要進行大量微調才能最大限度地利用 DPU。如你所知，我們擅長這些工程項目。

So the teams are enjoying it, the activity is much, much higher than before. And the goal is to scale these clusters as quickly as possible, so our customers can run their jobs faster. We're feeling good about it. You've heard comments from Jayshree as well in the past, and you have more on the analyst here on this topic, too, but all good on the activity front that we have.

因此團隊很享受它，活動比以前多得多。我們的目標是盡快擴展這些集群，以便我們的客戶可以更快地完成他們的工作。我們對此感覺很好。您過去也聽過 Jayshree 的評論，並且您對這個主題的分析師也有更多了解，但就我們的活動而言，一切都很好。

I think one thing to just add is the entropy and efficiency of these large language models and the job completion time has become so critical that it's not just about packet latency. It's really about end-to-end latency. And this is something our team, especially our engineers know a lot about from the early days. So we're really working this end-to-end.

我認為需要補充的一點是，這些大型語言模型的熵和效率以及作業完成時間變得如此關鍵，以至於它不僅僅與資料包延遲有關。這實際上與端到端延遲有關。我們的團隊，特別是我們的工程師從早期就非常了解這一點。所以我們確實在端到端地進行這項工作。

Your next question comes from the line of Aaron Rakers with Wells Fargo.

您的下一個問題來自富國銀行的 Aaron Rakers。

I just want to kind of dovetail off that last question a little bit. I know, Jayshree, last quarter, I think it was you commented that you'd expect to see pilot deployments for these AI opportunities in '24 and then meaningful volume in 2025. First of all, do you reaffirm that view? Or has that changed at all? And then on that, can you give us some context of how you see network spend intensity for these AI fabrics relative to, I think in the past, it's been kind of high single-digit percent of compute spend on networking in classical cloud infrastructure environments?

我只是想稍微回答一下最後一個問題。我知道，Jayshree，上個季度，我認為您曾評論說，您希望在 2024 年看到這些 AI 機會的試點部署，然後在 2025 年實現有意義的規模部署。或者說情況已經完全改變了？然後，關於這一點，您能否向我們介紹一下您如何看待這些 AI 結構的網路支出強度，相對於過去，在傳統的雲端基礎設施環境中，網路運算支出佔比一直很高，只有個位數？

First of all, Aaron, the first question is easy. I reaffirm that view and more later on November 9 at our Analyst Day. So if I tell you everything now, you may not attend that session. Coming back to this networking spend versus the rest of the GPUs and et cetera, I would say it started to get higher and higher with 100-gig, 400-gig and 800-gig, where the optics and the switches are more than 10%, perhaps even 15% in some cases, 20, a lot of its governed by the cables and optics too. But the percentage hasn't changed a lot in high-speed networking. In other words, it's not too different between 10, 100, 200, 400 and 800. So we -- you'll continue to see that 10% to 15% range.

首先，Aaron，第一個問題很簡單。我將於 11 月 9 日的分析師日上重申這一觀點以及其他更多內容。所以，如果我現在告訴你一切，你可能不會參加那個會議。回到網路支出與其他 GPU 等的比較，我想說，隨著 100G、400G 和 800G 的普及，網路支出開始越來越高，其中光學元件和交換器佔比超過 10%，在某些情況下甚至可能達到 15%，20%，其中許多也由電纜和光學元件控制。但在高速網路中，這一比例並沒有太大變化。換句話說，10、100、200、400 和 800 之間差別不大。 所以我們——您將繼續看到 10% 到 15% 的範圍。

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

您的下一個問題來自美國銀行的塔爾·利阿尼 (Tal Liani)。

Jayshree, your tone is definitely better this quarter than last quarter. And you sound more confident in the numbers. And I want to understand if something changed in the last 3 months that made you more optimistic. I'm looking at the consensus estimates. And it looks like the growth rate has been declining for 4 quarters from like 54% to about 20% next quarter. And any troughs that Q1 stays there and recovers after that. Do you agree that we are nearing kind of the end of the down adjustment to the growth rates and then it's going to stabilize and go up from there? Or how do you look at the risks of that not materializing?

Jayshree，本季你的語氣肯定比上季好多了。而且你聽起來對這些數字更有信心。我想了解過去三個月內是否發生了一些變化讓您變得更加樂觀。我正在查看一致預期。而且看起來成長率已經連續 4 個季度下降，從 54% 下降到下個季度的 20% 左右。並且 Q1 的任何低谷都會停留在那裡，然後恢復。您是否同意我們即將結束成長率的下調階段，然後開始穩定並上升？或者您如何看待這一情況未能實現的風險？

What do you think Ita?

Ita，您覺得怎樣？

So Tal, I think, look, we've been talking about kind of the growth decelerating as we move through the year just because of the comps are so high. I think if you look at the discussion we've had so far about '24 and obviously, the more to come next week, we've talked about double-digit growth. But again, we are expecting that there is some moderation on the cloud side of the business next year.

所以塔爾，我認為，看，我們一直在談論隨著時間推移增長會逐漸減速，只是因為同店銷售額太高了。我想，如果你看看我們迄今為止關於‘24’的討論，顯然，下週我們會談到更多的是兩位數的增長。但我們再次預計，明年雲端業務將會放緩。

So I think within the balance of kind of the plan that we've laid out and the discussions that we've laid out, I think we're executing well, right? We're giving you some upside on the -- in the guide for '23 and by default, to almost kind of some upside of '24, right? So I think we're executing well. But within the balance of what we talked about, we do believe that there's a moderation of price lending as we head into 2024.

因此，我認為，在我們所製定的計劃和所進行的討論的平衡範圍內，我們的執行情況很好，對嗎？我們在 23 年的指南中為您提供了一些好處，預設情況下，幾乎可以說是 24 年的一些好處，對嗎？所以我認為我們的執行得很好。但根據我們所討論的內容，我們確實相信，隨著進入 2024 年，價格貸款將有所緩和。

And Tal, I think I need to focus on my tone and maybe sign a song or something because that I felt really enthusiastic last quarter and this quarter.

塔爾，我想我需要關注我的語氣，也許簽一首歌或做點什麼，因為上個季度和這個季度我感到非常熱情。

But you're pretty happy last quarter. So...

但上個季度你相當開心。所以...

I'm a happy kind of [gal] at the moment.

我現在是一個很快樂的女孩。

We are reading between the lines.

我們正在讀懂字裡行間的東西。

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

您的下一個問題來自威廉布萊爾的塞巴斯蒂安納吉。

I wanted to ask about the change in revenue breakdown and the inclusion of OCI in this new Cloud Titan and AI segment. Was this the result of the material change in Arista's wallet share at Oracle? Or is that business becoming a larger portion of revenue? Anything you can provide there?

我想詢問一下收入組成的變化以及將 OCI 納入這個新的 Cloud Titan 和 AI 部門的情況。這是 Arista 在 Oracle 的份額發生重大變化的結果嗎？或者說這項業務正在成為公司收入的更大一部分嗎？您能提供什麼嗎？

Yes. No, we don't do it based on wallet share of Arista. We do it based on definition. So I think OCI has become a meaningful top-tier cloud customer and they belong in the cloud tightening category and in addition to their AI investments as well. So for reasons of classification and definition, the change is very warranted. And yes, they happened to be a good customer of Arista, that's nice as well.

是的。不，我們並不是根據 Arista 的錢包份額來這麼做的。我們根據定義來做這件事。因此，我認為 OCI 已成為一個有意義的頂級雲端客戶，他們屬於雲端緊縮類別，並且也在 AI 方面進行了投資。因此，出於分類和定義的原因，這種改變是非常合理的。是的，他們恰好是 Arista 的好客戶，這也很好。

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

您的下一個問題來自摩根士丹利的 Meta Marshall。

Jayshree or Anshul, maybe just some commentary on the Tier 2 and specialty providers. And just what you're seeing in terms of other people kind of building out some of these AI clusters, if you classify some of those customers as largely focused on back end today, and those represent opportunities going forward? Or just kind of what the discussion is outside of the Cloud Titans amongst some of these other guys that are building very large networks?

Jayshree 或 Anshul，也許只是對 Tier 2 和專業提供者的一些評論。而您看到其他人正在建立一些 AI 集群，如果您將其中一些客戶歸類為當今主要關注後端，那麼這些是否代表著未來的機會？或者只是在 Cloud Titans 之外與其他一些正在建立非常大的網路的人討論什麼？

Sure. Meta, this is Anshul. The Tier 2 cloud providers are doing exactly what the Tier 1 is doing just at a smaller scale. So the activity is out there. Many companies are trying to build these clusters, maybe not hundreds of thousands GPUs but thousands of GPUs together in their real estate if they can get them. But the designs that we're working on with them, the type of sort of features, fine-tuning is actually very, very similar to the cloud, just at a smaller scale. So we're very happy with that activity and this is across the board. It's very positive to see this in the ecosystem that it's not limited just 4 or 5 customers.

當然。 Meta，我是 Anshul。二級雲端提供者所做的事情與一級雲端提供者完全相同，只是規模較小。活動就在那裡。許多公司都在嘗試建立這些集群，也許不是數十萬個 GPU，而是在他們可以獲得的空間內建立數千個 GPU。但我們與他們合作進行的設計、功能類型和微調實際上與雲端非常非常相似，只是規模較小。所以我們對這項活動感到非常高興，而且這是全面的。在生態系統中看到這一點是非常積極的，它不僅限於 4 或 5 個客戶。

I think they're also waiting for GPUs like everyone else is. So there's that common problem that we're not the only one with lead time issues. But to clarify the comment on scale, Anshul and I are also seeing some very interesting enterprise projects against smaller scale. So a lot of customers are trying AI for small clusters, not too different from what we saw with HPC clusters back in the day.

我認為他們也和其他人一樣在等待 GPU。因此，存在一個常見的問題，即我們不是唯一面臨交貨時間問題的公司。但為了澄清關於規模的評論，Anshul 和我也看到了一些規模較小的非常有趣的企業專案。因此，許多客戶正在嘗試在小型叢集中使用 AI，這與我們過去在 HPC 叢集中看到的情況並沒有太大區別。

Yes.

是的。

Your next question comes from the line of Michael Ng with Goldman Sachs.

您的下一個問題來自高盛的 Michael Ng。

I just had one on the OpEx outperformance in the quarter. We saw an unseasonal decline quarter-on-quarter. And I think you mentioned lower product introduction costs that may have helped R&D. I was just wondering if you could talk a little bit more about that aspect of it. Any way we should think about product introductions going forward to help us understand the trajectory of OpEx?

我剛剛發布了本季營運支出超額表現的報告。我們看到了季度間不尋常的下滑。我認為您提到了較低的產品引進成本可能有助於研發。我只是想知道您是否可以進一步談談這方面的情況。我們應該以何種方式考慮未來的產品推出，以幫助我們了解營運支出的軌跡？

Yes. I mean a lot of it is timing, right? We've got a lot of different projects, a lot of different products kind of sold through the R&D labs right now. So there is going to be some kind of volatility in terms of when the spend shows up, when the project spend happens, et cetera. So I think we were lower this quarter in Q3 than maybe we even anticipated coming into the quarter. I expect that to come back in kind of the guide for Q4. And again, there may be some volatility in that even going forward, just because it's all about timing, nothing unusual in that. There's just a lot of products kind of going through the R&D labs.

是的。我的意思是很多都是時間問題，對吧？我們目前有很多不同的項目，許多不同的產品透過研發實驗室銷售。因此，在支出出現的時間、專案支出發生的時間等方面都會出現某種波動。因此，我認為本季第三季的業績甚至可能低於我們預期。我預計這將以第四季度指南的形式回歸。而且，未來可能還會出現一些波動，這只是因為時機問題，沒有什麼不尋常的。許多產品正在經過研發實驗室的研發。

So Michael, when the chips are down, our spending is down. But when the chips come on hard, our spending is hard too. So -- and expect our prototype to have some high variability, and we've got a lot, a lot of new products in the pipeline that Andy, Anshul, Ken, you are all working on. So we expect that number to go up over the next 4 quarters.

所以邁克爾，當情況惡化時，我們的支出就會減少。但當籌碼變得困難時，我們的支出也會變得困難。所以 — — 預計我們的原型會具有很高的可變性，而且我們已經有很多新產品正在籌備中，Andy、Anshul、Ken 你們都在致力於這些產品。因此我們預計未來 4 個季度該數字將會上升。

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

您的下一個問題來自花旗的阿蒂夫馬利克 (Atif Malik)。

Jayshree, at the recent Open Compute Project Conference, Marvell and Broadcom leading Ethernet switch merchant chip providers sounded very confident in terms of Ethernet adoption at hyperscalers like Meta and Oracle as well. And then one of your peers has talked about $500 million in AI orders whether it's custom chip. So I was curious about your thoughts on the dynamics between custom ship and merchant switch chip providers? And how does that help Arista?

Jayshree，在最近的開放運算專案會議上，領先的乙太網路交換器商用晶片供應商 Marvell 和 Broadcom 對 Meta 和 Oracle 等超大規模企業採用乙太網路也非常有信心。然後您的一位同行談到了 5 億美元的人工智慧訂單，無論是定制晶片。所以我很好奇您對定制船舶和商家交換晶片提供者之間的動態的看法？這對 Arista 有何幫助？

Yes. Atif, we have been strong proponents in our last 15, 17 years of Arista Korea on merchant silicon. We look for the best-of-breed chips. It's something my team -- engineering team has built a lot of chips in their past before, but we decided to work with the best-of-breed companies, Broadcom being one of our favorite and major suppliers. Of course, in the past, we worked with Intel, Cavium, and we don't rule out other suppliers as well. But this is clearly an area where you can't just build one chip, you have to build a portfolio of silicon and what Broadcom has done in building that portfolio not only for cloud networking but for campus and AI is impressive. And you have to not just look at performance, you have to look at price density, power, these are all very important metrics as we look ahead.

是的。 Atif，在過去的 15 到 17 年裡，我們一直是 Arista Korea 在商用矽片領域的堅定支持者。我們尋找同類最佳的晶片。我的團隊——工程團隊以前曾經製造過很多晶片，但我們決定與最優秀的公司合作，而博通是我們最喜歡和最主要的供應商之一。當然，過去我們曾與英特爾、Cavium合作過，我們也不排除其他供應商。但顯然，在這個領域你不能只製造一種晶片，你必須建立一個矽片產品組合，而博通不僅在雲網絡，而且在校園和人工智慧領域構建這個產品組合的努力令人印象深刻。而且你不僅要看性能，還要看價格密度、功率，這些都是我們展望未來非常重要的指標。

The root issue here, and you -- we'll share this more with you going forward as well. It's not just the merchant silicon but how you can enable the merchant silicon with the right software and drivers, and this is an area that really Arista excels, and if you just have chips, you can't build the system. But our system-wide features, whether it's a genetic load balancing, or latency analyzer to really improve the job completion time and deal with that frequent communication and generative AI is also fundamentally important. You're going to hear a lot more about this next week. So stay tuned.

這裡的根本問題是──我們今後也會與你們分享更多這個問題。這不僅僅是商用矽片，還包括如何使用正確的軟體和驅動程式來啟用商用矽片，這是 Arista 真正擅長的領域，如果你只有晶片，你就無法建立系統。但是我們的系統範圍的功能，無論是遺傳負載平衡還是延遲分析器，對於真正改善作業完成時間並處理頻繁的通訊和生成人工智慧也至關重要。下週您將會聽到更多有關此事的消息。敬請關注。

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

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

Jayshree and Ita, can you discuss a little more your gross margin commentary that it should moderate next year from the 63% levels in the back half. I mean, are we talking about it going to the first half '23 kind of levels or just a little bit of a degradation next year? And what would be the reason behind it? Other than lead times, is there any other mix or other issues that would cause it to go down?

Jayshree 和 Ita，你們能否進一步討論一下你們的毛利率評論，即明年毛利率應該會從下半年的 63% 水平有所緩和。我的意思是，我們談論的是它明年會恢復到 23 年上半年的水平還是會略有下降？這背後的原因又是什麼呢？除了交貨時間之外，還有其他因素或其他問題會導致其價格下降嗎？

So Ben, I think what Jayshree's commentary and my commentary is, we have been seeing it incrementally improved as we've gone through the year. We expect it to stabilize. So not that we expect it to go down next year but more that it will stabilize and then it will become more dependent on customer mix another thing and again, similar to where we've had before. We obviously provide some more outlook on discussion on this next week, too. But -- so the intention was not to say that we think it starts to decline. Again, it was more than we think it will stabilize after a period where we've been seeing these incremental improvements.

所以本，我認為，正如傑什里和我的評論所言，隨著時間的推移，我們已經看到情況逐漸改善。我們預計它會穩定下來。因此，我們不認為明年它會下降，而是預計它會穩定下來，然後變得更加依賴客戶組合，這與我們之前的情況類似。顯然，我們也對下​​週的討論提供了更多的展望。但是——我們的意圖並不是說我們認為它開始下降。再者，在我們看到這些漸進式改進的一段時間之後，它比我們想像的還要穩定。

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

您的下一個問題來自巴克萊銀行的蒂姆·朗。

I just wanted to hit on the Cloud Titan vertical or Cloud Titan AI vertical now, if I could. I think, Ita one of your comments was down a little or something in the quarter. Could you just -- 2 parts here, talk a little bit about that command. Is this just timing? Or are there some different market share dynamics there? And then secondly, if you could talk a little bit about opportunities at other hyperscalers. I know that's something where there's been trial activity and potential and it sounds like it might take a little while but any updates on other cloud titans that could become larger customers.

如果可以的話，我現在只想涉足 Cloud Titan 垂直領域或 Cloud Titan AI 垂直領域。我認為，您的某條評論在本季有所下降。您能否 -- 這裡有兩個部分，稍微討論一下該指令。這只是時機嗎？或者那裡存在一些不同的市場份額動態？其次，您能否談談其他超大規模企業的機會？我知道這是一個有試驗活動和潛力的事情，聽起來可能需要一點時間，但有沒有關於其他可能成為更大客戶的雲端巨頭的最新消息。

Just in terms of cloud, I mean, it's going to be a good cloud year again in 2023 for us. I think but we did come into the year saying we wanted, if we could, to balanced supply a little bit towards enterprise, and we have been doing that because there's been some -- you'll see it's not a huge mix shift but there has been some mix shift towards enterprise when we can, and we're pleased that we've been able to do that.

就雲而言，我的意思是，2023 年對我們來說又將是一個雲端繁榮的一年。我認為，但我們確實在今年說過，如果可以的話，我們希望稍微平衡一下對企業的供應，我們一直在這樣做，因為你會發現這不是一個巨大的混合轉變，但當我們可能的時候，確實有一些混合轉變向企業轉變，我們很高興能夠做到這一點。

Anshul, I don't know if you want to take that other cloud.

安舒爾，我不知道你是否想拿走那另一片雲。

Sure. Tim, the engagement with other cloud titans for our customers but small customers are still very positive. The good customers are, many of you know, in routed levels, backbone, WAN use cases as well. Then next week, we'll touch a little bit more on the whole build versus my topic.

當然。提姆，我們與其他雲端巨頭的合作對我們的客戶和小客戶來說仍然非常積極。你們許多人都知道，優質客戶也存在於路由層級、主幹網路、廣域網路用例中。然後下週，我們將更多地討論整個構建與我的主題。

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

您的下一個問題來自派珀·桑德勒 (Piper Sandler) 的詹姆斯·菲什 (James Fish)。

So great quarter. Just on the product side, you guys released a new 25-gig offering recently. I guess what's been the early feedback? What kind of differentiates down there? And Jayshree, just to clarify here, when you talk about that double-digit growth rate for next year and years beyond, are you talking about a multiyear CAGR or for '24 specifically and then for '25 and '26 and beyond? Just trying to clarify here.

真是一個美好的季度。僅在產品方面，你們最近發布了一款新的 25G 產品。我猜測早期的回饋是什麼樣的？那裡有什麼區別？ Jayshree，我在這裡澄清一下，當您談到明年及以後幾年的兩位數增長率時，您指的是多年的複合年增長率，還是專門指2024年的複合年增長率，然後是2025年、2026年及以後的複合年增長率？只是想在此澄清一下。

Okay. Well, Anshul, do you want to answer the product question first?

好的。好吧，Anshul，你想先回答產品問題嗎？

Sure. Jim, the recent announcement was the launch of our 25-gig ultralow-latency switches. These are the 7130 Series. And now the whole world can upgrade the high-frequency trading infrastructure going from 10 to 25, at very, very low latency. You're talking about with cross-point technology, you're talking about 7 nanoseconds, but we also now introduced layer 2, layer 3 features at about 100 to 130 nanoseconds in these...

當然。吉姆，我們最近宣布推出 25G 超低延遲交換器。這些是 7130 系列。現在，全世界可以將高頻交易基礎設施從 10 個升級到 25 個，並且延遲非常低。您談到了交叉點技術，談到了 7 奈秒，但我們現在還引入了第 2 層、第 3 層功能，時間約為 100 到 130 奈秒...

And Anshul, just to put this in perspective, back in the day, it used to be 500 nanoseconds.

安舒爾，為了方便理解，我們來回顧一下，在過去，這個時間是 500 奈秒。

That's right, when it keeps going down.

沒錯，當它繼續下降時。

Yes, yes, faster than the speed of light. And James, just to give you a clarification. I was saying, as a company, either myself or Anshul, we're aiming for at least double digits in '24 and years beyond, but I wasn't making any forecast for exact numbers.

是的，是的，比光速還快。詹姆斯，我只是想向你澄清一下。我說過，作為一家公司，無論是我自己還是安舒爾，我們都在努力實現 24 年及以後至少兩位數的成長，但我沒有對具體數字做出任何預測。

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

您的下一個問題來自奧本海默的伊泰·基德倫 (Ittai Kidron)。

A quick question on gross margin. Nice improvements there. Ita, maybe you can go into the details of how much room is there more to go? And I'm just kind of wondering with your customers now looking at your excellent financials and in recovering gross margins. What are the odds that pricing pressures start coming back, something you probably have not seen much in the last couple of years since COVID, now that margins are normalizing, what -- could prices come down potentially perhaps even for the -- more specifically to the larger customers of yours?

關於毛利率的一個快速問題。那裡有很好的改進。 Ita，也許你可以詳細說一下還有多少空間？我只是有點好奇，您的客戶現在是否看到了您出色的財務狀況和正在恢復的毛利率。價格壓力再次回升的可能性有多大？ 自從新冠疫情以來，過去幾年您可能沒有看到太多這樣的情況，現在利潤率正在正常化，價格有可能下降嗎？ 甚至對於您的大客戶而言，價格有可能下降嗎？

Ittai, I'll just start by saying prices are always coming down as we go from one speed factor to another between the SerDes technology and the density, the dollar per gigabit is always coming down. So where pricing pressure doesn't change independent of our gross margin, we're always in competitive deals. Where the value really comes in, and again, as I alluded to this, is CapEx versus OpEx. So we expect pricing to be reasonably stable but we expect the operational cost to be significantly advantageous with the Arista technology. The total -- the TCO because of singular cloud vision, because of our software-driven approach, because of the fact that we have single-digit vulnerabilities while our industry peers have 100 to 500 of them in a given 5-year factor. These are all now paying customers enterprises, especially a are very fatigued with the poor quality of our competitors and obtain a lot of attention to that and willing to pay for that quality.

Ittai，我首先要說的是，隨著我們從一個速度因素轉向另一個速度因素，在 SerDes 技術和密度之間，每千兆位元的美元價格總是在下降。因此，當定價壓力不會獨立於我們的毛利率而變化時，我們始終保持有競爭力的交易。價值的真正體現在於資本支出 (CapEx) 與營運支出 (OpEx)，正如我再次提到的。因此，我們預計價格將相當穩定，但我們預計 Arista 技術的營運成本將具有明顯優勢。整體擁有成本 (TCO) 是由於我們擁有單一的雲端願景，由於我們採用軟體驅動的方法，由於我們只有個位數的漏洞，而我們的行業同行在給定的 5 年內有 100 到 500 個漏洞。這些都是現在付費客戶的企業，尤其是對競爭對手的低品質感到非常厭倦，對此給予了高度關注並願意為這種品質付費。

Ittai, as Jayshree mentioned, I want to emphasize this, the market is very competitive, and it has been ever since we started. The gross margin that we report is not the reason why customers try to negotiate pricing. The gross margin is simply a result of what we've been executing on. I think we'll continue doing so.

Ittai，正如 Jayshree 所提到的，我想強調這一點，市場競爭非常激烈，自從我們開始以來就一直如此。我們報告的毛利率並不是客戶試圖協商定價的原因。毛利率只是我們一直執行的結果。我想我們會繼續這樣做。

Your next question comes from the line of Simon Leopold with Raymond James.

您的下一個問題來自雷蒙德詹姆斯 (Raymond James) 的西蒙利奧波德 (Simon Leopold)。

I wanted to see if you would be able willing to comment on your customer concentration year-to-date. I appreciate it can be lumpy quarter-to-quarter. But given sort of where you were in 2022, I'd just like to get a better understanding what essentially the progress has been in 2023. And in that context, how big is enterprise as a percent of revenue this year-to-date versus where it was last year?

我想看看您是否願意對今年迄今為止的客戶集中度發表評論。我知道每個季度的情況可能會有所不同。但考慮到 2022 年的情況，我只想更了解 2023 年的進展。

Simon, I -- we're very proud of our customers, even if they're concentrated, we love it. And as you know, the last year, we had some outsized concentration, if I recall the numbers, Meta was at 26%, and what was Microsoft, Ita?

西蒙，我——我們為我們的客戶感到非常自豪，即使他們很專注，我們也喜歡他們。如你所知，去年我們的集中度非常高，如果我沒記錯的話，Meta 佔了 26%，而微軟、Ita 佔了多​​少？

Microsoft was 16%.

微軟佔16%。

16% to 17%. While we expect due to many of the CapEx news you've seen and shift in AI spending, that it's possible they come down but there's still going to be very strong north of 10% contributors to our 2023 results. And we continue to -- even as the denominator may get larger in the forthcoming years, we continue to look at them as 2 very important and strategic customers for us.

16% 至 17%。雖然我們預計，由於您所看到的許多資本支出新聞和人工智慧支出的轉變，這些支出可能會下降，但對我們的 2023 年業績來說，仍有 10% 以上的強勁貢獻者。而且我們會繼續下去——即使未來幾年分母可能會變得更大，我們仍會繼續將他們視為兩個非常重要且具有戰略意義的客戶。

Your next question comes from the line of David Vogt with UBS.

您的下一個問題來自瑞銀的 David Vogt。

I just want to follow up on Simon's question, maybe put it a little bit differently. So if I think about your market sector trend update, how much of the shift to that 40% to 45% cloud and AI Titans reflects the inclusion of new AI use cases going forward and the shift of Oracle combined with maybe some normalization at Meta and Microsoft. Can you kind of help us think through the dynamics there? And if it looks like share is going to be unchanged with enterprise and financials, does that suggest to you that those markets are going to grow comparably over the long term across the cycle? Is that the way to look at it?

我只是想跟進西蒙的問題，或許表達得有點不同。因此，如果我考慮您的市場部門趨勢更新，那麼向 40% 到 45% 的雲端和 AI 巨頭的轉變在多大程度上反映了未來新 AI 用例的納入以及 Oracle 的轉變以及 Meta 和 Microsoft 的一些正常化。您能幫助我們思考那裡的動態嗎？如果看起來企業和金融的份額將保持不變，這是否意味著這些市場將在整個週期內長期保持同等增長？是這樣看待問題的嗎？

Yes. David, your analysis is really deep on this one. Let me just say how innocently we reported this, which is, Oracle is a greater than $1 million installed server company right now. And their -- both their cloud spend as OCI and AI is significant, both as a company and for Arista. But we're not making any assumptions and that will vary every year, of course, on the mix of Meta, Microsoft or Oracle or any other for that matter. We're simply seeing AI is going to become such an important component of all our cloud titans that is now a combined vertical. Don't read too much more into that.

是的。大衛，你對這個問題的分析非常深入。我只想說，我們報告這件事是多麼的天真，Oracle 現在是一家安裝伺服器價值超過 100 萬美元的公司。無論對於公司還是對於 Arista 來說，他們在 OCI 和 AI 方面的雲端運算支出都非常重要。但我們不做任何假設，當然，每年都會有所不同，取決於 Meta、Microsoft、Oracle 或任何其他公司的組合。我們只是看到人工智慧將成為我們所有雲端運算巨頭的重要組成部分，現在是一個綜合的垂直領域。不要對此進行過多的解讀。

Yes. It's more of a forward-looking impact, to be honest David. Historically, this doesn't really change the trends that we've been talking about previously. It's really more about the future and how do you think AI will impact these numbers going forward.

是的。說實話，大衛，這更是一種前瞻性的影響。從歷史上看，這並沒有真正改變我們之前談論的趨勢。這實際上更多的是關於未來，以及您認為人工智慧將如何影響這些數字。

Yes, I AI is just (inaudible) that much important right now, it's right. You know that. But as it starts to become important, then this combined will go north of the 39% we have normally forecast.

是的，我認為 AI 現在（聽不清楚）非常重要，沒錯。你知道的。但隨著它開始變得重要，這個數字將超過我們通常預測的 39%。

Your next question comes from the line of Erik Suppiger with JMP Securities.

您的下一個問題來自 JMP Securities 的 Erik Suppiger。

Thanks for taking my question and congrats. I know you don't want to talk about backlog but can you give us a sense at what point or at what time you think your book-to-bill will return back to 1 or greater than 1? Or when will your lead times reach a normalized level? And then I have a quick follow-up after that.

感謝您回答我的問題，恭喜您。我知道您不想談論積壓訂單，但您能否告訴我們，您認為您的訂單出貨比將在什麼時候或什麼時間回到 1 或大於 1？或者你的交貨時間什麼時候才能達到正常水平？然後我會快速跟進。

Jayshree is not going to talk about book-to-bill, if you don't talk about backlog, Erik. I think, honestly, we're making improvements. And Jayshree talked about how lead times are much improved, right? So we'll continue to do that. That's a positive thing for the customers and for the business. And we're still not back to kind of the turns business that we had some time ago. We're making progress, but we're still not there.

艾瑞克，如果你不談論積壓訂單，傑什裡就不會談論訂單出貨比。我認為，說實話，我們正在取得進展。 Jayshree 說過交貨時間是如何大幅改善的，對嗎？因此我們會繼續這樣做。這對客戶和企業來說都是一件好事。我們還沒有恢復到前段時間那種轉變狀態。我們正在取得進展，但尚未到達目標。

May be I just want to add, and I'm quite proud of the progress the team has made. When you look back 2 years ago, we were short of components. We were making multiyear purchases. There was a risk of a very large exposure because you can't get all these forecasts right. And then obviously, the mix changes from time to time, especially with the cloud and AI. So it's very hard to measure our business on book-to-bill and backlog at a given time. But if you have to look at it as an overall multiyear trend.

也許我只是想補充一下，我對團隊的進步感到非常自豪。回顧兩年前，我們缺少零件。我們正在進行多年期的採購。由於你無法正確做出所有預測，因此有非常大的風險。顯然，這種組合會不時發生變化，尤其是雲端運算和人工智慧。因此，很難在特定時間內根據訂單出貨比和積壓訂單來衡量我們的業務。但如果你把它看作是一個多年的總體趨勢。

Can you comment then on just when will the lead times be at a normalized level?

那麼您能否評論一下交貨時間何時才能達到正常水平？

Yes. And I think we said this. It's been improving consistently and we expect it to be normalized just like our gross margins in 2024.

是的。我想我們說過這件事。它一直在持續改善，我們預計它將像我們的 2024 年毛利率一樣正常化。

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 last question today comes from the line of Woo Jin Ho with Bloomberg.

您今天的最後一個問題來自彭博社的 Woo Jin Ho。

Great. [I made a cut]. Happy Halloween folks. So I think there was a mention on merchant silicon earlier in the Q&A. And one of your merchant silicon partners has actually moved up the stack towards the service provider routing. I'm just curious if there's any intention on going after that piece if that chip is made available to you?

偉大的。 [我剪掉了]。祝大家萬聖節快樂。所以我認為早些時候在問答環節中提到了商用矽片。您的一位商業矽片合作夥伴實際上已經將堆疊向上移動到服務提供者路由。我只是好奇，如果該晶片可用，您是否有意追求該晶片？

Sure. Woo Jin, I believe you are referring to the latest announcement Broadcom on their 25.60 Jericho chip that was announced recently.

當然。 Woo Jin，我相信您指的是 Broadcom 最近發布的最新 25.60 Jericho 晶片公告。

Yes, the Qumran3D.

是的，Qumran3D。

Qumran3D, exactly. So it's the same family, same features. And as you know, we've been a great partner of Broadcom for a long time, and we will continue to build new products. This is not a new entry, so to speak. We've been building these products that can be used on switches our orders for a while, and that bandwidth just doubled going to now 25.6. So you can expect some products from us in the future with those variants as well. But really -- nothing really changed. Just innovation continues and merchant silicon continues to succeed.

Qumran3D，確實如此。因此，它們是同一個家族，具有相同的特徵。如您所知，我們長期以來一直是博通的優秀合作夥伴，我們將繼續打造新產品。可以這麼說，這並不是一個新的條目。我們一直在建立這些可以在交換機上使用的產品，我們的訂單已經有一段時間了，頻寬剛剛翻了一番，達到 25.6。因此，您也可以期待我們將來推出具有這些變體的一些產品。但事實上──什麼都沒有改變。只是創新仍在繼續，商用矽片將繼續取得成功。

And the investment, Woo Jin, we have made in our routing stack over the last 10 years, I want to say, has just gotten better and stronger. Power in the Internet, power in the cloud, power in the AI, these are hard problems. And they require thousands of engineers of investment to build the right VXLAN, BGP routing, EVPN, et cetera. So it's not just a chip. It's how we name the chip to do these complicated routing algorithms.

而且，我想說的是，我們過去 10 年在路由堆疊上所做的投資已經變得更好、更強大了。網路中的力量、雲端中的力量、人工智慧中的力量，這些都是難題。他們需要數千名工程師的投入來建置正確的 VXLAN、BGP 路由、EVPN 等。所以它不僅僅是一個晶片。這就是我們如何命名晶片來執行這些複雜的路由演算法。

This concludes the Arista Networks Third 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. As a reminder, Arista will be posting our 2023 Cloud and AI Innovators Analyst Day on Thursday, November 9. If you are interested in attending virtually, you may register from the Investors section of our website.

Arista Networks 2023 年第三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。提醒一下，Arista 將於 11 月 9 日星期四發布我們的 2023 年雲端和 AI 創新者分析師日。

Thank you for joining us today, and thank you for your interest in Arista.

感謝您今天加入我們，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.

感謝您的加入。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。