Credo Technology Group Holding Ltd (CRDO) 2025 Q3 法說會逐字稿

Ladies and gentlemen, thank you for standing by. (Operator Instructions)

女士們、先生們，感謝你們的支持。（操作員指示）

And I would now like to turn the conference over to Dan O'Neil. Please go ahead, sir.

現在我想將會議交給丹·奧尼爾。先生，請繼續。

Good afternoon. Thank you for joining our earnings call for the third quarter of fiscal 2025. Today, I am joined by Bill Brennan, Credo's Chief Executive Officer; and Dan Fleming, our Chief Financial Officer.

午安.感謝您參加我們 2025 財年第三季的財報電話會議。今天，與我一起參加會議的還有 Credo 執行長 Bill Brennan 和財務長 Dan Fleming。

During this call, we will make certain forward-looking statements. These forward-looking statements are subject to risks and uncertainties discussed in detail in our documents filed with the SEC, which can be found in the Investor Relations section of the company's website.

在本次電話會議中，我們將做出一些前瞻性的陳述。這些前瞻性陳述受到我們向美國證券交易委員會提交的文件中詳細討論的風險和不確定性的影響，這些文件可以在公司網站的投資者關係部分找到。

It is not possible for the company's management to predict all risks, nor can the company assess the impact of all factors on its business or the extent to which any factor or combination of factors may cause actual results to differ materially from those contained in any forward-looking statement. Given these risks, uncertainties and assumptions, the forward-looking events discussed during this call may not occur, and actual results could differ materially and adversely from those anticipated or implied.

公司管理階層不可能預測所有風險，也無法評估所有因素對其業務的影響，或任何因素或因素組合導致實際結果與任何前瞻性聲明中的結果有重大差異的程度。鑑於這些風險、不確定性和假設，本次電話會議中討論的前瞻性事件可能不會發生，實際結果可能與預期或暗示的結果有重大不利差異。

The company undertakes no obligation to publicly update forward-looking statements for any reason after the date of this call to conform these statements to actual results or to changes in the company's expectations, except as required by law.

除法律要求外，本公司不承擔在本次電話會議召開之日後以任何理由公開更新前瞻性陳述的義務，以使這些陳述符合實際結果或公司預期的變化。

Also, during this call, we will refer to certain non-GAAP financial measures, which we consider to be important measures of the company's performance. These non-GAAP financial measures are provided in addition to, and not as a substitute for or superior to, financial performance prepared in accordance with US GAAP.

此外，在本次電話會議中，我們將參考某些非公認會計準則財務指標，我們認為這些指標是衡量公司績效的重要指標。這些非公認會計準則財務指標是對依照美國公認會計準則編製的財務表現的補充，而不是替代或優於該財務表現。

A discussion of why we use non-GAAP financial measures and reconciliations between our GAAP and non-GAAP financial measures is available in the earnings release we issued today, which can be accessed using the Investor Relations portion of our website.

我們今天發布的收益報告中討論了我們為何使用非 GAAP 財務指標以及 GAAP 和非 GAAP 財務指標之間的對賬，您可以透過我們網站的投資者關係部分存取該報告。

With that, I will now turn the call over to our CEO. Bill?

說完這些，我現在將電話轉給我們的執行長。帳單？

Thanks, Dan, and thank you for joining our earnings for the third quarter of fiscal '25. I'll start with an overview of our third-quarter results and then discuss our outlook. Following my comments, our CFO, Dan Flemming, will provide Q3 financial details and our guidance for the fiscal fourth quarter.

謝謝，丹，感謝您加入我們 25 財年第三季的收益報告。我將首先概述我們的第三季業績，然後討論我們的展望。在我發表評論之後，我們的財務長 Dan Flemming 將提供第三季的財務細節以及我們對第四季財務的指導。

For the third quarter, Credo reported revenue of $135 million, up 87% sequentially and up 154% year over year. Credo's non-GAAP gross margin was 63.8%. Credo achieved record revenue in Q3 as we saw the expected inflection point in our business. This ramp was led by our largest hyperscale customer as they scale production of AI platforms.

Credo 報告第三季營收為 1.35 億美元，季增 87%，年增 154%。Credo 的非 GAAP 毛利率為 63.8%。Credo 在第三季實現了創紀錄的收入，因為我們看到了業務的預期轉折點。這一成長是由我們最大的超大規模客戶推動的，他們正在擴大 AI 平台的生產規模。

Additionally, we received solidified forecasts and saw increased design activity for our products with additional hyperscalers and other customers, as the performance, reliability, and power benefits of our connectivity solutions have become increasingly clear throughout the industry.

此外，隨著我們的連接解決方案的性能、可靠性和功率優勢在整個行業中變得越來越明顯，我們收到了鞏固的預測，並看到我們的產品與更多的超大規模廠商和其他客戶的設計活動增加。

In a world where data drives everything, the demand for faster, more reliable, and energy-efficient connectivity continues to expand rapidly. Key to Credo's competitive advantage is our multi-tiered innovation, which enables us to deliver a broad set of optimized solutions that are tuned at every level.

在數據驅動一切的世界裡，對更快、更可靠、更節能的連線的需求持續快速成長。Credo 競爭優勢的關鍵在於我們的多層次創新，這使我們能夠提供一系列在各個層面都經過優化的解決方案。

The first tier of innovation is in our SerDes technology. Our SerDes technology is purpose-built to tackle the toughest bandwidth challenges, balancing speeds up to 200-gig per lane with exceptional performance and power efficiency. By leveraging advanced signal processing and a programmable design, we've created a flexible architecture that adapts to the unique needs of AI workloads, whether it's long reach data center links or ultrashort connections in dense compute environments.

第一層創新是我們的 SerDes 技術。我們的 SerDes 技術專為解決最嚴峻的頻寬挑戰而設計，可平衡每通道高達 200 千兆的速度以及卓越的性能和功率效率。透過利用先進的訊號處理和可程式設計，我們創建了一種靈活的架構，可以適應 AI 工作負載的獨特需求，無論是長距離資料中心鏈路還是密集運算環境中的超短連接。

The second tier of innovation is in our integrated circuit design. Our IC designs, including retimers, DSPs and, chiplets are engineered to deliver the best combination of performance, power, and cost for a given application. Our LRO DSP is a great example of innovation on the customers' behalf to deliver a compelling IC solution optimized for power efficiency and optical links.

第二層次的創新是我們的積體電路設計。我們的 IC 設計（包括重定時器、DSP 和小晶片）旨在為特定應用提供性能、功率和成本的最佳組合。我們的 LRO DSP 是代表客戶進行創新的一個很好的例子，它提供了針對電源效率和光學鏈路進行最佳化的引人注目的 IC 解決方案。

The third tier of innovation is our system-level approach. We don't stop at chips. Our best example of system-level innovation is Credo pioneering the active electrical cable market.

第三層次的創新是我們的系統級方法。我們不會止步於晶片。我們系統級創新的最佳例子是 Credo 開拓有源電纜市場。

By taking accountability for the system-level solution, we raised the bar to deliver end-to-end connectivity solutions that go beyond industry standards to deliver unique functionality and best-in-class reliability. We see this system-level approach creating a larger opportunity for Credo with the emergence of AI clusters due to the intense demand for reliability and power efficiency.

透過承擔系統級解決方案的責任，我們提高了提供超越行業標準的端到端連接解決方案的標準，以提供獨特的功能和一流的可靠性。我們看到，隨著人工智慧叢集的出現，由於對可靠性和功率效率的強烈需求，這種系統級方法為 Credo 創造了更大的機會。

Finally, wrapped around each tier of innovation is our development and diagnostics software and firmware platform to deliver predictive signal integrity, link optimization, and tuning. From the SerDes to the system level, this software platform helps our customers navigate system development to achieve best performance, yields, and reliability.

最後，圍繞每一層創新的是我們的開發和診斷軟體和韌體平台，以提供預測訊號完整性、鏈路優化和調整。從 SerDes 到系統級，此軟體平台可協助我們的客戶進行系統開發，以實現最佳效能、產量和可靠性。

As we look forward, we'll expand our solutions to the PCIe protocol with the same ingenuity that creates our differentiated Ethernet solutions. With the introduction of a full suite of PCIe products on the near-term horizon, Credo will soon be addressing a larger connectivity opportunity with AI scale out and scale up networks, substantially expanding our overall TAM.

展望未來，我們將以創造差異化乙太網路解決方案的相同獨創性將我們的解決方案擴展到 PCIe 協定。隨著近期推出全套 PCIe 產品，Credo 很快就會透過 AI 橫向擴展和縱向擴展網路來解決更大的連接機會，從而大幅擴展我們的整體 TAM。

Now I'll discuss our business in more detail. Regarding our AEC product line. As expected, our revenue surged in the third quarter driven by our largest hyperscale customer. Compared to alternatives, the benefits of AECs have become clearer.

現在我將更詳細地討論我們的業務。關於我們的 AEC 產品線。正如預期的那樣，在我們最大的超大規模客戶的推動下，我們的收入在第三季飆升。與其他替代方案相比，AEC 的優勢更加明顯。

More than ever, data centers are highly focused on back-end network reliability. With billions of hours operating in the field, AECs have become the de facto standard for intra-rack connections for NIC to TOR and switch to switch applications.

資料中心比以往任何時候都更加關注後端網路的可靠性。憑藉數十億小時的現場運行，AEC 已成為 NIC 到 TOR 和交換器到交換器應用的機架內連接的事實標準。

However, we are now seeing a new expansion of AEC usage. Our ZeroFlap AECs deliver more than 100 times better reliability than laser-based optical solutions. And as a result, we're seeing AECs replacing optics for rack-to-rack solutions for lengths up to seven meters.

然而，我們現在看到 AEC 用途有了新的擴展。我們的 ZeroFlap AEC 的可靠性比基於雷射的光學解決方案高出 100 倍以上。因此，我們看到 AEC 取代了機架到機架解決方案的光學元件，長度可達七公尺。

We continue to make significant progress with additional hyperscalers for our Ethernet AEC solutions. We've achieved volume production with three hyperscalers, and we're in qualification with two additional hyperscalers expecting production in fiscal '26. With broad traction, we feel confident we'll continue to see increasing diversification of our revenue base across more customers in the coming quarters and years.

我們繼續透過為乙太網路 AEC 解決方案添加額外的超大規模器取得重大進展。我們已經與三家超大規模企業實現了批量生產，我們正在與另外兩家超大規模企業進行資格審查，預計在 26 財年實現量產。憑藉廣泛的吸引力，我們有信心在未來幾季和幾年內，我們將繼續看到更多客戶的收入基礎日益多樣化。

Additionally, Credo continues to make progress with our PCIe AEC solutions. Our Gen6 64-gig PAM-4 AECs will deliver the same compelling benefits for AI scale-up networks as deployments move to rack scale architectures. Credo will demonstrate our PCIe AECs at NVIDIA's GTC show later this month.

此外，Credo 繼續憑藉 PCIe AEC 解決方案取得進展。隨著部署轉向機架規模架構，我們的 Gen6 64-gig PAM-4 AEC 將為 AI 擴展網路帶來同樣引人注目的優勢。Credo 將於本月稍後在 NVIDIA 的 GTC 展會上展示我們的 PCIe AEC。

We expect customer design engagements and qualifications for our PCIe AECs in the upcoming quarters with a significant revenue opportunity in the upcoming years. For our AEC business in total, we expect continued revenue growth based on customer forecasts, new qualifications, new design engagements, and TAM expansion.

我們預計我們的 PCIe AEC 將在未來幾季獲得客戶設計參與和認證，並在未來幾年帶來巨大的收入機會。對於我們的整個 AEC 業務，我們預計基於客戶預測、新資格、新設計合作和 TAM 擴展的收入將持續成長。

Now I'll turn to our optical business. Our optical DSP business is on track to achieve the growth objectives we set out at the beginning of fiscal '25. We have opportunities across the global customer base with revenue currently driven by 50-gig and 100-gig per-lane designs for AOC and transceiver applications at port speeds up to 800 gig.

現在我將談談我們的光學業務。我們的光學 DSP 業務預計將實現我們在 25 財年年初制定的成長目標。我們在全球客戶群中都擁有機會，目前的收入主要來自 AOC 和收發器應用的每通道 50 千兆和 100 千兆的設計，連接埠速度高達 800 千兆。

Credo is actively engaged in opportunities with more than 10 transceiver vendors for multiple hyperscale end users. We work with our optical transceiver partners to provide full DSP and LRO options to meet a wide range of networking architectures.

Credo 正積極與 10 多家收發器供應商合作，為多個超大規模終端用戶提供合作機會。我們與光收發器合作夥伴合作，提供完整的 DSP 和 LRO 選項，以滿足各種網路架構。

We see a large and growing market for these offerings as well as for 1.6T port deployments in the future. With our recent 3-nanometer tape-out, Credo is well positioned for these leading-edge opportunities with our 200-gig per-lane DSPs, where we expect to again have a compelling combination of performance, power, and features.

我們看到這些產品以及未來 1.6T 連接埠部署的市場規模龐大且不斷成長。憑藉我們最近的 3 奈米流片，Credo 憑藉每通道 200 千兆的 DSP 在這些前沿機會中佔據了有利地位，我們期望再次擁有性能、功率和功能的引人注目的組合。

We see the market opportunity for optical connectivity continuing to be very dynamic as reliability and energy efficiency become more important. As a result, we've seen an increasing opportunity for Credo to deliver system-level advantages to our partners, activating Credo's third tier of innovation I outlined earlier.

我們看到，隨著可靠性和能源效率變得越來越重要，光纖連接的市場機會將繼續保持活躍。因此，我們看到 Credo 有越來越多的機會為我們的合作夥伴提供系統級優勢，從而激活我之前概述的 Credo 第三層創新。

Next month, at the OFC conference in San Francisco, we'll demonstrate a full suite of optical solutions, including 200-gig per lane in conjunction with our optical module partners. Based on all of our progress, the breadth of customer engagements, and the expanding market opportunity, we remain excited about the increasing revenue prospects given our role as an innovator in the optical connectivity market.

下個月，在舊金山舉行的 OFC 會議上，我們將與我們的光學模組合作夥伴一起展示全套光學解決方案，包括每通道 200 千兆的解決方案。基於我們取得的所有進展、廣泛的客戶參與度以及不斷擴大的市場機會，鑑於我們作為光纖連接市場創新者的角色，我們對不斷增長的收入前景感到興奮。

Now regarding our retimer business. Credo continued to gain momentum with our retimer business in the third quarter. Over the past several years, Credo has established leadership in the Ethernet retailer market, delivering advanced capabilities such as MAX second encryption, gearboxing, and other software-enabled functionality.

現在談談我們的重定時器業務。Credo 的定時器業務在第三季度繼續保持強勁成長勢頭。在過去的幾年中，Credo 在乙太網路零售商市場確立了領導地位，提供了 MAX 秒加密、變速箱和其他軟體支援功能等先進功能。

Existing customer wins and future opportunities here include 100-gig and 200-gig per-lane applications for both traditional switching and increasingly for AI servers requiring retimers for scale-out networks. This year, Credo has entered the market for PCIe retimers used in scale up networks.

現有客戶的成功和未來的機會包括每通道 100 千兆和 200 千兆的應用，既可用於傳統交換，也可用於日益增長的需要重定時器進行橫向擴展網絡的 AI 伺服器。今年，Credo 進入了用於擴展網路的 PCIe 重定時器市場。

Our strategy is in alignment with our three-tier innovation approach. We believe credos PCIe SerDes IP will establish new benchmarks for the combination of latency, reach performance, and power; and that our implementation of the two-CAN PCIe retimer will deliver compelling advantages to our customers.

我們的策略與我們的三層創新方法一致。我們相信 credos PCIe SerDes IP 將為延遲、性能和功率的組合建立新的基準；並且我們對雙 CAN PCIe 重定時器的實現將為我們的客戶帶來引人注目的優勢。

In February, Credo participated in the PCI SIG compliance workshop in Taipei, and we are pleased that our two-CAN retimer achieved full PCIe compliance. It is notable that Credo is only the second vendor to achieve this level of compliance certification for PCIe Gen5.

今年 2 月，Credo 參加了在台北舉行的 PCI SIG 合規性研討會，我們很高興我們的雙 CAN 重定時器完全符合 PCIe 標準。值得注意的是，Credo 是第二家獲得 PCIe Gen5 此等級合規認證的供應商。

This significant milestone demonstrates our capability to bring best-in-class PCIe products to market. Credo will be added to the PCI SIG integrators list in the coming weeks.

這一重要里程碑證明了我們有能力將一流的 PCIe 產品推向市場。Credo 將在未來幾週內被添加到 PCI SIG 整合商清單中。

During Q3, we engaged with key customers who evaluated our PCIe silicon. I'm pleased to say that the feedback was very encouraging, and we received our first platform commitment from a large AI server ODM. We are on track for production revenue in calendar year 2026.

在第三季度，我們與評估我們的 PCIe 矽片的關鍵客戶進行了接觸。我很高興地說，反饋非常令人鼓舞，我們從一家大型 AI 伺服器 ODM 獲得了第一個平台承諾。我們預計在 2026 日曆年實現生產收入。

Market forecasters believe the TAM for PCIe retimers will exceed $1 billion by 2027, and Credo is very well positioned to compete for material market share.

市場預測人士認為，到 2027 年，PCIe 重定時器的 TAM 將超過 10 億美元，而 Credo 在爭奪材料市場份額方面處於非常有利的地位。

In summary, I'd like to first comment about our team's incredible execution over the past quarter. To successfully navigate a ramp of this magnitude requires extremely tight operational control, supply chain coordination, and customer communication.

總而言之，我想先評論一下我們團隊在過去一個季度的出色表現。要成功通過如此大規模的坡道，需要極其嚴格的營運控制、供應鏈協調和客戶溝通。

Just as Credo works tirelessly with customers to innovate on solving their pressing connectivity needs, the Credo team is clearly rising to the occasion to deliver on the significant demand ramp we're experiencing. As we more broadly ramp customers across our products, we will continue to closely manage our execution.

正如 Credo 孜孜不倦地與客戶合作，透過創新解決他們迫切的連結需求一樣，Credo 團隊顯然正在努力滿足我們所經歷的巨大的市場需求。隨著我們更廣泛地擴大客戶對我們產品的支持，我們將繼續密切管理我們的執行情況。

I remain enthusiastic about the expanding market opportunity for high-speed connectivity, driven by the promise of AI and the investment it's spurring. Credo's tiered approach to innovation has and will continue to be an advantage as we serve our customers. Based on our progress with customers and the increasing demand for leading-edge connectivity solutions, Credo remains on track for continued scaling of revenue and profit.

我仍然對高速連接的不斷擴大的市場機會充滿熱情，這得益於人工智慧的前景及其所刺激的投資。Credo 的分層創新方法已經並將繼續成為我們服務客戶的優勢。基於我們與客戶共同取得的進展以及對尖端連接解決方案日益增長的需求，Credo 的收入和利潤仍有望持續增長。

I'll now turn the call over to our CFO, Dan Fleming, who will provide more detail.

現在我將把電話轉給我們的財務長丹‧弗萊明 (Dan Fleming)，他將提供更多細節。

Thank you, Bill, and good afternoon. I'll first review our Q3 results and then discuss our outlook for Q4 of fiscal year '25.

謝謝你，比爾，下午好。我將首先回顧我們的第三季業績，然後討論我們對 25 財年第四季的展望。

In Q3, we reported revenue of $135 million, up 87% sequentially and up 154% year over year and well above the high end of our guidance range. Our product business generated $132 million of revenue in Q3. up 91% sequentially and up 155% year over year.

第三季度，我們的營收為 1.35 億美元，季增 87%，年增 154%，遠高於我們預期範圍的上限。我們的產品業務在第三季創造了 1.32 億美元的收入，季增 91%，年增 155%。

Notably, our AEC product line grew strong triple digits sequentially to achieve new record revenue levels. Our product business, excluding product engineering services, generated another record at $129.4 million of revenue in Q3, 101% higher than our previous product record in the prior quarter.

值得注意的是，我們的 AEC 產品線連續三位數強勁成長，實現了新的創紀錄收入水準。我們的產品業務（不包括產品工程服務）在第三季創造了新的紀錄，收入達到 1.294 億美元，比上一季的產品紀錄高出 101%。

Our IP business generated $3 million of revenue in Q3. As demonstrated by our product revenue ramp, we are seeing substantial opportunities with customer programs on the product side which we are prioritizing. This prioritization does not impact our long-term model for company-wide non-GAAP gross margin of 63% to 65%.

我們的 IP 業務在第三季創造了 300 萬美元的收入。正如我們的產品收入成長所表明的那樣，我們在產品方面看到了客戶計劃中的巨大機遇，我們對此予以優先考慮。這種優先排序不會影響我們全公司非 GAAP 毛利率 63% 至 65% 的長期模型。

Our largest end customer was 86% of revenue in Q3. As a reminder, customer mix will vary from quarter to quarter, and we continue to make progress in diversifying our customer base. As we shared last quarter, we had seven customers that contributed more than 5% of revenue. And going forward, we expect that three to four customers will be greater than 10% of revenue in the coming quarters and fiscal year as additional hyperscalers ramp to more significant volumes, as Bill described.

我們最大的終端客戶佔第三季營收的 86%。提醒一下，客戶結構每季都會有所不同，我們將繼續在客戶群多元化方面取得進展。正如我們上個季度所分享的，我們有七位客戶貢獻了超過 5% 的收入。展望未來，我們預計，隨著更多超大規模企業的業務量不斷增加，三到四家客戶將在未來幾季和財年貢獻超過 10% 的收入，正如比爾所描述的那樣。

Our team delivered Q3 non-GAAP gross margin of 63.8%, above the high end of our guidance range and up 17 basis points sequentially. Our product non-GAAP gross margin was 63% in the quarter, up 85 basis points sequentially and up 152 basis points year over year.

我們的團隊實現了第三季非公認會計準則毛利率 63.8%，高於我們的預期範圍高端，季增 17 個基點。本季度，我們產品的非 GAAP 毛利率為 63%，較上一季成長 85 個基點，比去年同期成長 152 個基點。

Our products non-GAAP gross margin, excluding product engineering services, was 62.4% in the quarter, up 229 basis points sequentially and up 934 basis points year over year, primarily due to increasing scale. Total non-GAAP operating expenses in the third quarter were $43.8 million, within our guidance range and up 16% sequentially due primarily to higher headcounts.

本季度，我們產品的非 GAAP 毛利率（不含產品工程服務）為 62.4%，比上一季成長 229 個基點，比去年同期成長 934 個基點，這主要歸功於規模的擴大。第三季非公認會計準則總營運費用為 4,380 萬美元，在我們的指導範圍內，季增 16%，主要原因是員工人數增加。

Our non-GAAP operating income was $42.4 million in Q3, compared to non-GAAP operating income of $8.3 million in Q2, up demonstrably due to the leverage attained by achieving 87% sequential top-line growth. Our non-GAAP operating margin was 31.4% in the quarter compared to a non-GAAP operating margin of 11.5% in the prior quarter, a sequential increase of nearly 20% points.

我們第三季的非公認會計準則營業收入為 4,240 萬美元，而第二季的非公認會計準則營業收入為 830 萬美元，這一增長明顯是由於實現了 87% 的連續營業收入增長而產生的槓桿作用。本季我們的非公認會計準則營業利潤率為 31.4%，而上一季的非公認會計準則營業利潤率為 11.5%，環比增長近 20 個百分點。

Our non-GAAP net income was $45.4 million in Q3 compared to non-GAAP net income of $12.3 million in Q2. And our non-GAAP net margin was 33.6% in the quarter, above the high end of our long-term net margin model of 28% to 33%.

我們第三季的非 GAAP 淨收入為 4,540 萬美元，而第二季的非 GAAP 淨收入為 1,230 萬美元。本季我們的非公認會計準則淨利率為 33.6%，高於我們長期淨利潤​​率模型的高端 28% 至 33%。

Cash flow from operations in the third quarter was $4.2 million, down sequentially due to working capital increases driven by the significant sequential product ramp. CapEx was $4.6 million in the quarter, driven largely by purchases of production equipment. And free cash flow was negative $0.4 million, an improvement of $11.3 million from the second quarter.

第三季的營運現金流為 420 萬美元，季減，原因是產品產量持續大幅增加導致營運資本增加。本季資本支出為 460 萬美元，主要源自於生產設備的購買。自由現金流為負 40 萬美元，較第二季增加 1,130 萬美元。

We ended the quarter with cash and equivalents of $379.2 million, a decrease of $3.7 million from the second quarter. We remain well capitalized to continue investing in our growth opportunities while maintaining a substantial cash buffer. Our Q3 ending inventory was $53.2 million, up $16.9 million sequentially.

本季末，我們的現金和等價物為 3.792 億美元，比第二季減少 370 萬美元。我們仍然擁有充足的資本來繼續投資於我們的成長機會，同時保持大量的現金緩衝。我們第三季的期末庫存為 5,320 萬美元，比上一季增加 1,690 萬美元。

Now turning to our guidance, we currently expect revenue in Q4 of fiscal '25 to be between $155 million and $165 million, up 19% sequentially at the midpoint. We expect Q4 non-GAAP gross margin to be within a range of 63% to 65%.

現在談到我們的指導，我們目前預計 25 財年第四季的營收將在 1.55 億美元至 1.65 億美元之間，中位數環比成長 19%。我們預計第四季非公認會計準則毛利率將在 63% 至 65% 之間。

We expect Q4 non-GAAP operating expenses to be between $50 million and $52 million. We expect Q4 diluted weighted average share count to be approximately 188 million shares.

我們預計第四季非公認會計準則營運費用將在 5,000 萬美元至 5,200 萬美元之間。我們預計第四季稀釋加權平均股數約為 1.88 億股。

As we approach the start of fiscal year 206, we expect revenue growth from fiscal year '25 to fiscal year '26 to be greater than 50%, and we expect non-GAAP operating expenses to grow at half the rate of revenue from fiscal year '25 to fiscal year '26. As a result, we look forward to continue driving operating leverage while expanding our net margin throughout the year.

隨著 206 財年即將開始，我們預計 2025 財年到 2026 財年的營收成長率將超過 50%，而我們預計非 GAAP 營運費用成長率將是 2025 財年到 2026 財年收入成長率的一半。因此，我們期待全年繼續提高經營槓桿，同時擴大淨利率。

And with that, I will open it up for questions.

現在，我將開始回答大家的提問。

(Operator Instructions) Vivek Arya, Bank of America Securities.

（操作員指示）美國銀行證券的 Vivek Arya。

Just for the first one, if you could give us a sense of how large was the largest customer. I think Dan mentioned the number, but I didn't catch it. But I guess I have two parts to the question.

就第一個問題而言，您能否告訴我們最大的客戶有多大？我認為丹提到了這個號碼，但我沒聽清楚。但我想我的問題有兩個部分。

One is, where are you in the adoption of AEC at that customer? And if you exclude that customer, how are you looking at the growth of your business among other customers? Because depending on that customer concentration, rightly, we get a slightly different trend outside of that large customer.

一是，您在該客戶中採用 AEC 的情況如何？如果您排除該客戶，您如何看待其他客戶中您的業務成長？因為根據客戶集中度，我們會發現大客戶以外的趨勢略有不同。

So Vivek -- so what I mentioned in the prepared remarks is our largest customer was 86% of revenue. So let me just give you some historical context there.

所以 Vivek——我在準備好的發言中提到，我們最大的客戶佔了我們收入的 86%。那麼讓我來介紹一些歷史背景吧。

As we entered our fiscal '25, we described a second half inflection point really to be driven by our largest hyperscaler, which is exactly what we saw play out with that 87% sequential growth into Q3. So as far as customer concentration goes, Q3 was a bit of an outlier for us, especially considering the customer diversity that we had in Q2 and what we expect in the coming quarters.

當我們進入 25 財年時，我們描述了下半年的拐點，這實際上是由我們最大的超大規模企業推動的，這正是我們在第三季度看到的 87% 的連續增長。因此，就客戶集中度而言，第三季對我們來說有點異常，特別是考慮到我們在第二季擁有的客戶多樣性以及我們對未來幾季的預期。

And again, as I mentioned in the prepared remarks, we expect 3% to 4% or 10%-plus end customers in the coming quarters and fiscal year, which is simply based on the forecast that we received from our customers. So just broadly speaking, there's clearly a broad-based need for our innovative solutions as lane speeds are increasing at these hyperscale deployments.

正如我在準備好的演講中提到的那樣，我們預計未來幾個季度和財政年度的最終客戶數量將達到 3% 到 4% 或 10% 以上，這只是基於我們從客戶收到的預測。因此，從廣義上講，隨著這些超大規模部署中車道速度的提高，顯然對我們的創新解決方案有廣泛的需求。

And our AEC opportunities, in particular, are expanding in the back-end network from originally the scale-out network to now the scale-up network as well, which represents a significant expansion of our TAM. So I'll let Bill add some additional color to those comments on customer diversification.

尤其是我們的 AEC 機會，正在後端網絡中從最初的橫向擴展網絡擴展到現在的縱向擴展網絡，這代表著我們 TAM 的顯著擴展。因此，我會讓比爾對客戶多樣化的評論添加一些額外的內容。

First thing I would say is that my takeaways are that this is really a great confirmation of the AEC product category end market and the overall TAM. I think that it's also a great confirmation of Credo's leadership in the space.

首先我想說的是，我的結論是，這確實對 AEC 產品類別終端市場和整體 TAM 進行了很好的確認。我認為這也是對 Credo 在該領域領導地位的極大肯定。

And so as we've talked about on past calls, we see each one of hyperscalers capable of driving a very sizable TAM as we take our engagements from first deployment to a more populated accretive AMC deployment longer term.

正如我們在過去的電話會議中所討論的那樣，隨著我們的合作從首次部署轉向更長期、更密集的增值 AMC 部署，我們看到每一個超大規模企業都能夠推動非常可觀的 TAM。

So I think if we look at our customer base, it's broadening. I mentioned that we've taken now three customers, three hyperscalers, to volume production. And we've got two additional hyperscalers in qualification and expecting ramp in our fiscal '26.

因此我認為，如果我們看一下我們的客戶群，它正在擴大。我提到過，我們現在已經帶領三家客戶、三家超大規模企業實現了大量生產。我們還有另外兩家超大規模企業正在獲得資格認證，預計在 26 財年將實現成長。

So we expect to see really solid customer diversity long term within our AEC business. But also, I think that that I'll add to that that the expanding opportunity that we see for optical as well as PCIe drive really further diversity across products and end customers as we look towards the future.

因此，我們期望在 AEC 業務中長期看到真正穩固的客戶多樣性。但同時，我認為我還要補充一點，展望未來，我們看到光學和 PCIe 不斷擴大的機會確實會進一步促進產品和最終客戶的多樣化。

Got it. So my follow-up, Bill -- if let's say, more of the market turns towards inference rather than training, right, whether more (technical difficulty) of the compute build-out, what does that do to these AI clusters? And what does that do to connectivity requirements and the AEC potential? Is that good, bad, neutral for AEC?

知道了。所以我的後續問題是，比爾——如果說更多的市場轉向推理而不是訓練，那麼計算構建的更多（技術難度）會對這些人工智慧集群產生什麼影響？這對連線需求和 AEC 潛力有何影響？這對 AEC 來說是好事、壞事還是中性事？

I think as the inference market really takes off, I think, if anything, we would see a larger opportunity from an AEC standpoint. And that's just based on sheer number of deployments related to inference.

我認為，隨著推理市場真正起飛，從 AEC 的角度來看，我們將看到更大的機會。而這僅僅是基於與推理相關的部署數量。

So I think that as we look at the customer activity going forward, there's a lot more focus now on inference, and we're seeing really an uptick in the amount of activity.

因此，我認為，當我們展望未來的客戶活動時，現在會更加關注推理，而且我們確實看到活動數量增加。

Tom O'Malley, Barclays.

巴克萊銀行的湯姆·奧馬利。

So I think a lot of the tone of the call here has been about like a broadening of the portfolio. You're talking about PCIe retimers. You're talking about PCIe cables.

所以我認為這次電話會議的基調很大程度上與擴大投資組合有關。您正在談論 PCIe 重定時器。您說的是 PCIe 電纜。

To push you maybe one step further, if you're looking in the PCIe realm, an area of a lot of value is in the switching ecosystem as well. Do you guys have plans to move into the switching ecosystem?

再進一步說，如果您關注 PCIe 領域，那麼交換生態系統中也有一個非常有價值的領域。你們有進入轉換生態系的計畫嗎？

Is that like a natural progression from the road you're traveling down already? And maybe talk about the challenges of doing retimers versus moving to switching and the time line that it would take to transition from product in the market today to maybe some new products on the switching side.

這是否是您所走過的路的自然發展？也許可以談談使用重定時器與轉向開關相比所面臨的挑戰，以及從當前市場上的產品過渡到開關方面的一些新產品所需的時間表。

Sure. I appreciate the question. So we've made a lot of progress with PCIe over the past quarters. I think the comment that I made earlier was regarding Credo passing PCI Special Interest Group, PCI-SIG, most recently in February. That's a great confirmation of the technology, great confirmation of interoperability for Gen5.

當然。我很感謝你提出這個問題。過去幾季我們在 PCIe 方面取得了很大進展。我想我之前發表的評論是關於 Credo 透過 PCI 特別興趣小組 (PCI-SIG) 的，最近一次是在二月。這是對該技術的極大肯定，也是對 Gen5 互通性的極大肯定。

So as we move towards Gen6, I think we're increasingly bullish on our ability to compete in the market. We see that for us, first step is really going after the retimer market as well as the PCIe AEC market as scale-up architectures move to rack-scale.

因此，隨著我們邁向 Gen6，我認為我們對我們在市場上的競爭力越來越有信心。我們看到，對我們來說，隨著擴展架構轉向機架規模，第一步實際上是進軍重定時器市場以及 PCIe AEC 市場。

I think it's a really natural progression to think about moving to PCIe switching. There's different challenges associated with that, but it is really not such a huge step to transition from retimers to the type of switches that are being built and talked about being deployed in the future.

我認為考慮轉向 PCIe 交換是一個非常自然的過程。這其中存在著不同的挑戰，但從重定時器過渡到正在建置並討論在未來部署的交換器類型，其實並不是一個巨大的進步。

So I look at that as a possible spot for us to grow into over time. Right now, we're very much focused on getting things right related to the retimer as well as the AECs that we're building.

因此，我認為隨著時間的推移，這將是我們成長的一個可能點。現在，我們非常注重正確處理與重定時器以及我們正在建造的 AEC 相關的事情。

I appreciate you answering despite you just announcing the prior product before, so I appreciate it. But just in terms of the market outlook -- so you're talking about the scale-up network as -- and also the scale-out network, you have PCIe cables potentially more aligned with the scale-up network.

儘管您剛剛發布了之前的產品，但我仍然感謝您的回答，所以我很感激。但就市場前景而言 - 所以您正在談論擴展網絡 - 以及擴展網絡，您的 PCIe 電纜可能與擴展網絡更加一致。

You've got some of the AEC product, particularly looking more at the scale-out network. Can you talk about the opportunity size? Because I think we're going to see in a couple of weeks here like maybe a redefinition of what scale up and scale out is.

您已經了解一些 AEC 產品，特別是關於橫向擴展網路的產品。能談談機會大小嗎？因為我認為我們將在幾週內看到擴大規模和擴展程度的重新定義。

Is it one rack? Is it multiple racks? Maybe your thoughts on scale up versus scale out, where does the opportunity set lie for you? Are you indifferent? Just want to get your temperature on that transition.

是一個架子嗎？是多個機架嗎？也許您對擴大規模與縮小規模有什麼看法，對您來說機會在哪裡？你漠不關心嗎？只是想了解你在這個轉變過程中的溫度。

Sure. So we've talked about the opportunity in the past. So scale-out, we really see it being an Ethernet protocol. And so we see that continuing on the path that we're -- that we've been on as the market moves to 100-gig per lane and as the market will subsequently move to 200-gig per lane solutions.

當然。我們過去曾討論過這個機會。因此，對於橫向擴展，我們確實將其視為一種乙太網路協定。因此，我們看到，隨著市場轉向每通道 100 千兆，以及市場隨後將轉向每通道 200 千兆的解決方案，我們將繼續沿著我們一直以來的道路前進。

So scale up is really probably more interesting in the sense of talking about where we are today and where we're going. And we do see by the innovations previously shown by NVIDIA and what we expect coming up that you can see that there's a lot of activity. It's really a dynamic space from their perspective.

因此，從談論我們目前所處的位置以及我們要去哪裡的角度來看，擴大規模實際上可能更有趣。我們確實從 NVIDIA 之前展示的創新以及我們預期即將發生的事情中看到，有很多活動。從他們的角度來看，這確實是一個充滿活力的空間。

Now they're a little bit different from the market broadly. But what we see within the customer base that we're talking to is that this scale-up network that has really been a network that exists within an AI appliance really going rack-scale. And then of course, there's an opportunity for it to go row-scale long term.

現在它們與市場總體上略有不同。但我們在與之交談的客戶群中看到，這種擴展網絡實際上是存在於 AI 設備中的網絡，並且正在真正實現機架規模。當然，它有機會長期實現行規模化。

We've talked about the volumes being larger than the scale-out network opportunities. So we really see this as a big new TAM. As the market moves from Gen5 to Gen6, we're talking about moving from 32-gig NRZ, which is really very old technology -- and it is really not competitive if you compare it to the market leader from a bandwidth standpoint per lane.

我們已經討論過容量大於橫向擴展網路機會的問題。因此我們確實將其視為一個巨大的新 TAM。隨著市場從 Gen5 轉向 Gen6，我們正在談論從 32 千兆 NRZ 轉向，這確實是一項非常古老的技術 - 如果從每個通道的頻寬角度將其與市場領導者進行比較，它確實沒有競爭力。

Gen6 represents the move to 64-gig PAM-4 modulation. So that's a step-up in complexity and difficulty from the existing suppliers in the market moving towards that direction for Credo. We've been there for many years, so we feel like we're going to bring the same compelling advantages that we've brought to Ethernet.

Gen6 代表向 64 千兆 PAM-4 調製的轉變。因此，對於 Credo 來說，市場上現有供應商朝這個方向發展會使其複雜性和難度提升。我們已經在那裡很多年了，所以我們覺得我們將帶來與乙太網路相同的引人注目的優勢。

The interesting part of the future is that when you look at that scale of network and the opportunity for improving performance with an increase in bandwidth -- we've talked about the Ultra Accelerator Link, the UAL conversation in the market, taking the scale-up network from 64 gig all the way to 224 gig.

未來有趣的部分是，當你看到網路規模和透過增加頻寬來提高效能的機會時——我們已經討論了超級加速器鏈路、市場上的 UAL 對話，將擴展網路從 64 千兆一直擴展到 224 千兆。

And so we're going to be in a unique position to take advantage of that transition when that happens with the other market leaders that are outside of the NVIDIA ecosystem. So we're pretty excited, absolutely, about both opportunities, but I would say the scale-up opportunity is probably larger if we look out the two- to three-year time frame.

因此，當 NVIDIA 生態系統之外的其他市場領導者也參與其中時，我們將處於獨特的地位來利用這種轉變。因此，我們對這兩個機會都非常興奮，但我想說，如果我們著眼於兩到三年的時間框架，擴大規模的機會可能會更大。

Karl Ackerman, BNP Paribas.

法國巴黎銀行的卡爾‧阿克曼。

Two questions, if I may. The first one, I guess, what's driving the uptick in gross margins in the April quarter? Is it IP licensing revenue-related? Or is there some other thing that we should be thinking about?

請問我有兩個問題。第一個問題，我想，是什麼推動了四月季度毛利率的上升？這與 IP 授權收入有關嗎？或者我們還應該考慮其他一些事情？

So to sum it up, we're really seeing a huge benefit from scale, which we've talked about -- or we've foreseen it for quite some time. And as you know, our overall gross margin in Q3 was almost 64%, up a little bit from last quarter.

總而言之，我們確實看到了規模帶來的巨大好處，我們已經討論過這一點——或者說我們已經預見到它很長一段時間了。如您所知，我們第三季的整體毛利率接近 64%，比上一季略有上升。

But if you dig into the numbers, the thing that's more indicative of what's happening underneath is if you look at our product gross margin, excluding product engineering services, our gross margin was 62.4%. So that was up over 200 basis points sequentially, up over 900 basis points year over year, principally driven by scale. It's really as simple as that.

但如果你深入研究這些數字，更能說明根本情況的是，如果你看看我們的產品毛利率，不包括產品工程服務，我們的毛利率為 62.4%。因此，該數字比上一季上漲了 200 多個基點，比去年同期上漲了 900 多個基點，主要得益於規模的成長。確實就是這麼簡單。

The other -- or the lesser factor that I'll mention, we have a warrant with Amazon. The contra revenue associated with that rolled off during the quarter as we attain that or as we fulfill the $200 million of total gross revenue shipments to them in the quarter. So that was also accretive to margin, which will continue to be accretive next quarter as well.

另一個——或者說我要提到的次要因素是，我們與亞馬遜有合作關係。當我們實現這一目標或在本季向他們完成 2 億美元的總收入出貨量時，與此相關的抵銷收入將在本季度滾動。因此這也增加了利潤率，下個季度也將繼續增加。

Got it. Just a follow-up, on your prepared comments, you spoke about how you may have three or four customers that would be 10% plus in the coming quarters or in calendar '26. Obviously, the first three today are AEC-related. Is the fourth one also AEC-related or would that suggest perhaps a broadening into other aspects and product portfolios of your business?

知道了。再問一下，根據您準備好的評論，您談到了在未來幾個季度或 26 年，您可能會有三到四個客戶，其成長率將超過 10%。顯然，今天的前三個都與 AEC 有關。第四個也與 AEC 有關嗎？或者這是否意味著可能擴展到您業務的其他方面和產品組合？

Of the ones we've referred to in those comments, they are AEC-related. I mean that by far is the largest driver in absolute dollars of our revenue growth.

我們在這些評論中提到的那些都與 AEC 有關。我的意思是，到目前為止，這是我們收入成長的最大驅動力。

Now having said that, if I just look simply year over year from fiscal '24 to '25, for instance, we've had our main three product categories between AEC -- our retimer business, our Ethernet retimer business, and optical, they're all experiencing sequential growth, but AEC just has -- it's growing from a bigger base into an even bigger base as we go into the future.

話雖如此，如果我只簡單回顧一下 24 財年到 25 財年的同比情況，我們的主要產品類別包括 AEC（重定時器業務、以太網重定時器業務和光學產品），它們都在經歷連續增長，但 AEC 已經——隨著我們邁向未來，它將從更大的基數增長到更大的基數。

Quinn Bolton, Needham & Company.

奎因·博爾頓，Needham & Company。

If I've got my numbers right, revenue outside of your largest customer went from about $48 million in October to about $19 million in January. You talked about revenue diversifying again over the next few quarters and into fiscal '26 with three to four customers that could be over 10%.

如果我的數字正確的話，那麼除最大客戶之外的收入從 10 月的約 4800 萬美元下降到了 1 月的約 1900 萬美元。您談到了未來幾個季度以及 26 財年的收入將再次多樣化，其中三到四個客戶的收入可能會超過 10%。

Can you just give us some sense, what do you expect your largest customer to do? Does it stay 80%-plus of revenue? Are you anticipating that that pulls back and that you see ramps at some of the other customers in the April quarter? Just any sense of revenue rediversifying would be helpful.

您能否告訴我們，您希望您的最大客戶做什麼？它是否能維持80%以上的收入？您是否預期四月份的銷售額會回落，而其他一些客戶的銷售額會上升？任何有關收入再多樣化的意識都會有所幫助。

We've talked in the past about -- actually, Amazon is a great example. So our largest hyperscaler, if you look at their Q1 revenue, was $30 million, then it went down a bit in Q2. Now it obviously surged in our Q3. Our internal expectation is they'll probably be in the same ZIP code as to where they were in absolute dollar terms this -- in Q3 or where they were in Q3.

我們過去曾討論過——實際上，亞馬遜就是一個很好的例子。因此，如果您看一下我們最大的超大規模提供商的第一季收入，其收入為 3000 萬美元，而第二季度則略有下降。現在，它在我們的第三季度明顯飆升。我們的內部預期是，他們可能與今年第三季或第三季的絕對美元金額處於同一郵遞區號區域。

So if you look at that, being what it is and knowing that we guided 19% sequentially up quarter over quarter into Q4 at the midpoint, that would imply that maybe they're two-thirds our revenue in Q4, would be what that math would imply.

因此，如果你看一下這個情況，就會知道我們在第四季度中期預計將環比增長 19%，這意味著他們可能占我們第四季度收入的三分之二，這就是這個數學計算所暗示的。

Got it. And then I guess just looking at that customer, obviously, it's a significant percentage of revenue in the January and April quarters. I think in the past, you've talked about having 12-month rolling forecast from some of your larger hyperscale customers.

知道了。然後我想，只要看一下那個客戶，顯然，它佔了一月和四月季度收入的很大一部分。我認為過去您曾談到一些大型超大規模客戶的 12 個月滾動預測。

Just generally speaking, can you give us any sense -- what gives you the confidence that the surge you've seen here in January and April may not just be somewhat related to an inventory build ahead of server deployments that, at some point in fiscal '26, you go into kind of a depletion mode or just a deployment mode where you could see, I guess, for a lack of a better term, an air pocket at that customer at some point next year? Just any help to hold hands would be appreciated.

總的來說，您能否給我們一些啟發——是什麼讓您有信心，您在 1 月和 4 月看到的激增可能不僅僅與服務器部署之前的庫存建立有關，在 26 財年的某個時候，您會進入一種耗盡模式或只是一種部署模式，我想，在明年的某個時候，您可能會看到該客戶出現氣穴？只要能提供任何幫助，我們都會感激不盡。

So I guess when we talk about visibility, I think it's very normal at this point to get 12-month forecasts from customers. I think this is one of the positives that came out of COVID, is that really, a top priority for our customers is to secure their supply across all materials, including the things that Credo supplies.

因此，我想當我們談論可見性時，我認為從客戶那裡獲得 12 個月的預測是非常正常的。我認為這是 COVID 帶來的正面影響之一，即我們的客戶的首要任務是確保所有材料的供應，包括 Credo 供應的材料。

And so when there is a sizable ramp, it's really imperative to work together with our customers as well as our supply chain to ensure that we can deliver flawlessly. And so I would say across the board from a customer perspective, we're getting good visibility.

因此，當出現大規模成長時，與我們的客戶以及供應鏈合作以確保我們能夠完美交付就顯得至關重要。因此我想說，從客戶的角度來看，我們獲得了良好的知名度。

In the case where there is a new ramp, this is where things become more high-definition as you get closer to the actual ramp. There are some things that can happen where we've seen push-outs to a schedule. We've seen pull-ins. We've seen increases as we saw this quarter in the level of deployments as customers ramp.

如果有新的坡道，當你靠近實際坡道時，事物就會變得更加高清。有些事情可能會發生，例如我們看到計劃被推遲。我們已經看到了拉入現象。隨著客戶數量的增加，本季的部署水準有所提高。

But generally, as we look towards the future, we see several customers that are at different stages of deployments. And so if we look at the numbers and compare, say, Q2 to Q3, you saw a pretty big shift, pretty good move.

但整體而言，展望未來，我們將看到多個客戶處於不同的部署階段。因此，如果我們查看數字並進行比較，比如說，第二季度和第三季度，你會看到相當大的轉變，相當不錯的舉措。

And we indicated that, say, for our -- the first customer that we ramped with over the last several years, we talked about a period of time where supply and consumption, those curves will meet. Well, that happened in Q3.

我們指出，對於我們過去幾年來合作的第一位客戶，我們討論了供應和消費曲線將會相遇的一段時間。嗯，那是在第三季發生的。

We also talked about that customer returning to historic levels, which we expect to be over $100 million run rate, say, as we look at fiscal '26. We've got a new customer that we're ramping, and the size of allocations that they're getting from a GPU standpoint suggests that they're going to be a very large customer in our fiscal '26 as well.

我們也談到了客戶恢復到歷史水準的情況，我們預計到 26 財年，客戶營運率將超過 1 億美元。我們有一位新客戶，我們正在擴大規模，從 GPU 的角度來看，他們獲得的分配規模表明，他們也將成為我們 26 財年的一個大客戶。

I mentioned the two new customers. These would be customers that are a little less definition on the exact timing and size of the ramp. But both are promising in the sense that we're seeing the level of deployments they're talking about being sizable.

我提到了兩位新客戶。這些客戶對於產能提升的具體時間和規模並不清楚。但從某種意義上來說，兩者都很有前景，因為我們看到他們所談論的部署水平相當可觀。

So as you look at the next couple of quarters, I'm sure we're going to have a decreasing level of concentration. But it's still going to be something that you would say that's a large amount of concentration, but it's going to be balanced out over the second half of fiscal '26.

因此，展望未來幾個季度，我相信我們的集中度將會下降。但它仍然會是一件需要大量關注的事情，但它將在 26 財年下半年得到平衡。

Understood, and I'd rather see you guys have the Amazon revenue than not. So congratulations on the nice results.

明白，我寧願看到你們擁有亞馬遜的收入。恭喜你取得如此好的成績。

These are good issues.

這些都是好問題。

Tore Svanberg, Stifel.

托爾·思文伯格，Stifel。

Congrats on the record results. Bill, in your prepared remarks, you talked about several layers or multitiered innovation. And the one thing that obviously that you talk about is the system-level approach.

恭喜你取得創紀錄的成績。比爾，在你準備好的發言中，你談到了多層次或多層次的創新。顯然，您談到的一件事就是系統級方法。

That's obviously very obvious on the AEC part of your business. But as you venture into some of these new segments and especially on PCIe, will you take more of a system business model approach there as well or should we assume that's going to be primarily selling chips?

這在您業務的 AEC 部分顯然非常明顯。但是，當您進入一些新的領域，特別是 PCIe 領域時，您是否也會採用更多的系統業務模型方法，或者我們是否應該假設這將主要銷售晶片？

Well, I think the fact that at GTC in a few weeks, we're going to be demonstrating a full rack with multiple AECs that are active, that are live, we're going to be demonstrating our AEC capability. And it may actually be -- it might even be the first demo of its kind in the AEC space.

嗯，我認為在幾週後的 GTC 上，我們將展示一個完整的機架，其中有多個處於活動狀態的 AEC，我們將展示我們的 AEC 能力。事實上，它甚至可能是 AEC 領域中此類演示的首次。

So yes, absolutely. From day one, we've been planning on pursuing the same path with PCIe AECs as we have with Ethernet, and no change there. We're even planning -- as the scale of architecture could potentially go row-scale, we're also investing in the optical space as well.

是的，絕對是。從第一天起，我們就計劃在 PCIe AEC 上走與乙太網路相同的道路，並且不會做出任何改變。我們甚至正在計劃——由於建築規模可能會達到行規模，我們也在投資光學空間。

That's very interesting. And then as my follow-up, LRO, I haven't heard a whole lot about it since you obviously introduced it. So any update you can give us there? If you do have some stuff that you're going to talk about at OFC, I get it. But any update we can get on LRO would be really helpful.

這很有趣。然後作為我的後續，LRO，自從你介紹它以來，我還沒有聽到太多關於它的消息。您能提供我們任何最新消息嗎？如果您確實有一些事情想在 OFC 上談論，我明白。但我們能獲得的任何有關 LRO 的更新資訊都將非常有幫助。

Right. We continue to offer customers both full DSP solutions and LRO solutions. We're agnostic to their choice. We see customers continuing to purchase traditional full DSP solutions where they can fit within the power ceiling.

正確的。我們將繼續為客戶提供完整的 DSP 解決方案和 LRO 解決方案。我們不知道他們的選擇。我們看到客戶繼續購買能夠滿足功率上限的傳統全 DSP 解決方案。

In the case -- we do have cases where power is a deciding factor. And I think with all of the customers we're talking to, LRO really becomes that de facto solution compared to, say, LPO. I think we've all seen that the discussion around LPO has decreased pretty substantially.

在這種情況下──我們確實遇到權力是決定性因素的情況。我認為，對於我們交談過的所有客戶來說，與 LPO 相比，LRO 確實成為了事實上的解決方案。我想我們都已經看到，有關 LPO 的討論已經大幅減少。

The conversation with LRO, I think is -- it's a real opportunity to get the kind of signal integrity, bit error rates, and interoperability and being able to save that kind of power, those sub-10 watts for an 800-gig port. But it's ultimately up to our customers to decide.

我認為與 LRO 的對話是——這是一個真正的機會，可以獲得訊號完整性、誤碼率和互通性，並且能夠節省這種功率，即 800 千兆連接埠的 10 瓦以下功率。但最終還是由我們的客戶來決定。

Vijay Rakesh, Mizuho.

瑞穗的 Vijay Rakesh。

Great quarter, congratulations. Just on the AEC side, I know you mentioned two new customers, two new hyperscalers that you'll be working with. On your visibility, do you see them ramping to -- as you look out 12, 18 months, do you see that ramping pretty nicely to significant volumes? How do you see that ramp? And I have a follow-up.

非常棒的一個季度，恭喜。就在 AEC 方面，我知道您提到了兩個新客戶，兩個您將與之合作的新超大規模客戶。從您的視角來看，您是否認為它們會逐漸增加——展望 12 到 18 個月，您是否認為它們會逐漸增加到相當大的數量？您覺得那個斜坡怎麼樣？我還有一個後續問題。

The customer-facing team at Credo is hard at work, and that includes our applications engineering group. And I think it's a function of exactly when we intercept on the ramp versus if we do. Both of these customers, there are design engagements that are absolutely identified and committed to.

Credo 的客戶團隊正在努力工作，其中包括我們的應用工程團隊。我認為這取決於我們何時在坡道上攔截以及是否攔截。這兩個客戶都有絕對認同和致力於的設計約定。

So I think it's a function of exactly when the ramp starts. And that's when -- I make the comment that fiscal '26 is what we expect. I think if we wanted to be somewhat, I guess, on the conservative side, we'd say second half of fiscal '26. But -- and we're getting to the point where our confidence is quite high with the additional two customers that are in qualification.

所以我認為這與坡道何時開始有關。那時——我評論說 26 財年符合我們的預期。我想，如果我們想要保守一點，我們會說是 26 財年下半年。但是——隨著另外兩個客戶獲得資格，我們的信心已經相當高了。

Got it. And then on the PCIe scale-up side, very great to see you guys ramp so fast since you -- so quickly after you introduced the cables. You mentioned one server ODM. How does that pipeline look? And do you have to qualify the server ODM or the GPU, ASIC guy? Just trying to understand.

知道了。然後在 PCIe 擴展方面，很高興看到你們在推出電纜後發展得如此之快。您提到了一台伺服器 ODM。管道看起來怎麼樣？您是否必須對伺服器 ODM 或 GPU、ASIC 人員進行資格審查？只是想了解一下。

So understand the timing, the timing of our ramp for PCIe will -- it will still take some time. We're really targeting the Gen6 market, which is -- that's really a calendar year '25 design cycle and a calendar '26 production ramp cycle.

因此，了解時間安排，我們對 PCIe 的提升時間安排將 - 它仍然需要一些時間。我們真正瞄準的是 Gen6 市場，也就是 25 年的設計週期和 26 年的生產爬坡週期。

And so it's really somewhat out in time. Even though we're demonstrating we're passing interoperability tests, we're absolutely ready to compete, it's going to be some time.

所以這確實有點不合時宜。儘管我們證明我們通過了互通性測試，我們已經做好了參與競爭的準備，但這還需要一些時間。

So another question would be, are there going to be Gen5 opportunities that we're able to pursue? And of course, as our solution is brought to market, we're open for business. And we'll bring advantages to Gen5 and easy upgrade to Gen6.

所以另一個問題是，我們是否能夠追求 Gen5 機會？當然，隨著我們的解決方案進入市場，我們就開始營業了。我們將為 Gen5 帶來優勢並輕鬆升級到 Gen6。

And so what we're seeing from a customer perspective is that there is independence in making decisions, and there is a need for a more broad supply chain than exists for Gen5.

因此，從客戶的角度來看，他們在做決策時具有獨立性，並且需要比 Gen5 更廣泛的供應鏈。

Sean O'Loughlin, TD Cowen.

肖恩·奧洛林 (Sean O'Loughlin)，TD Cowen。

Congrats on a nice set of results and guide. I'd like to trend a little bit and ask about front-end networking. The original -- obviously, the original AEC deployment was on a front-end solution.

恭喜您取得了一系列出色的成果和指導。我想稍微了解一下趨勢並詢問有關前端網路的問題。最初的——顯然，最初的 AEC 部署是基於前端解決方案。

And I wonder whether or how you and your customers are thinking about that front-end opportunity that still potentially exists. Is it just that traditional front-end networking is not where the innovation is happening in the space, and so customers are focused on the back end?

我想知道您和您的客戶是否或如何看待仍然存在的前端機會。是因為傳統的前端網路並不是該領域創新的領域，所以客戶只專注在後端？

Or is it a matter of time or some hybrid of the both? And then to squeeze an extra question on top of that, is there really a difference in the product that you ship to front end or go-to-market for the front end versus back end?

或者這只是時間問題，還是兩者的結合？然後在此基礎上再問一個問題，運送到前端或進入前端市場的產品與進入後端的產品真的有差別嗎？

So from a front-end network perspective, it applies to general compute and it applies to AI clusters. So the front-end connections are pretty similar between the two.

因此，從前端網路的角度來看，它適用於通用運算，也適用於人工智慧叢集。因此兩者的前端連接非常相似。

Now with general compute, that's the only connection to the network, the front-end network. And if we think about the traditional general compute space, it's really a question about really the x86 road map. And that has typically moved at a slower pace than what's happening within AI clusters.

現在，對於通用運算來說，這是與網路、前端網路的唯一連接。如果我們考慮傳統的通用運算空間，這實際上是一個關於 x86 路線圖的問題。而且這一進程通常比人工智慧叢集內部的進程還要慢。

And so it's very common still today to see 25-gig per lane front-end network connections. In fact, the one that we've shipped in high volume is of four lanes of 25 gig. So it's a 100-gig port connecting to that front-end network.

因此，如今每通道 25 千兆的前端網路連接仍然很常見。事實上，我們大批量出貨的是四條 25 千兆的通道。因此，它是一個連接到前端網路的 100 千兆連接埠。

The unique feature that we brought was that our lead customer implemented a dual-TOR design to achieve levels of availability, the SLAs of [5/9s]. So it was a way for them to really achieve a goal that had been in place for 5 to 10 years even. So that unique functionality is what turned on that business for the front-end network connection at 100-gig ports.

我們帶來的獨特功能是，我們的主要客戶實施了雙 TOR 設計，以實現可用性級別，即[5/9秒]。因此，這是他們真正實現已堅持 5 年甚至 10 年的目標的一種方式。因此，這種獨特的功能為 100 千兆埠的前端網路連線帶來了業務上的突破。

As we look at that general compute, there's a natural move to 50-gig per lane and 100-gig per lane. But in contrast, if we look at AI, just the nature of the application is causing customers to move to the fastest connection, so 100-gig per lane today, and really putting pressure on delivering 200-gig per lane solutions in the future.

當我們考慮一般計算時，自然會轉向每通道 50 千兆和每通道 100 千兆。但相較之下，如果我們看一下人工智慧，應用程式的性質正在促使客戶轉向最快的連接，即今天的每條通道 100 千兆，並且確實對未來提供每條通道 200 千兆的解決方案施加了壓力。

And within AI clusters, those front-end connections exist. And at best case, they'll be at the same line rates as we see in the back end. But probably most likely, they'll be at lower line rates. And so our largest customer is using our Credo AEC solutions for both back-end and front-end connections.

在人工智慧叢集中，這些前端連接是存在的。在最好的情況下，它們的線路速率將與我們在後端看到的相同。但最有可能的是，它們的線路速率會較低。因此，我們最大的客戶正在使用我們的 Credo AEC 解決方案進行後端和前端連接。

Great. Super helpful. And then just as a quick follow-up, you mentioned the three hyperscalers that are achieving volume shipments today. I think last quarter, you talked about the two hyperscale obviously that we all know about and then the third emerging hyperscale. I just want to clarify that that is not the third hyperscaler that you talked about today but is, in fact, a different customer.

偉大的。超有幫助。然後，作為快速的跟進，您提到了今天實現批量出貨的三家超大規模企業。我想上個季度，您談到了我們都知道的兩個超大規模，然後是第三個新興的超大規模。我只是想澄清一下，這不是您今天談到的第三個超大規模企業，事實上，這是一個不同的客戶。

So as it relates to this concept of emerging hyperscalers, it really boils down to the amount of spend. And if you look at the allocation that is being given to this -- the lead, quote-unquote, emerging hyperscaler, we really are classifying them now as a hyperscaler. So we're not drawing the line between the two.

因此，當它與新興超大規模的概念相關時，它實際上歸結為支出的數量。如果你看一下對此的分配——領先的、所謂的新興超大規模企業，我們現在確實將其歸類為超大規模企業。所以我們不會在兩者之間劃出界線。

Christopher Rolland, Susquehanna.

克里斯多福羅蘭 (Christopher Rolland)，薩斯奎漢納。

I guess yeah, as it concerns your largest customer here and maybe any of these large ramps going forward, I think a lot of these things are project-based. We have very large deployments, sometimes 200,000 GPUs in just a couple of months.

我想是的，因為它關係到你們這裡最大的客戶，也許還有未來的任何大型項目，我認為這些事情很多都是基於專案的。我們的部署規模非常大，有時短短幾個月內就部署了 20 萬個 GPU。

So as we think about these large, lumpy deployments, should we be expecting these customers, for example, in this case, a pretty strong handoff in the next few quarters to other customers? Like Bill, maybe if you could talk to how project-based these revenue ramps are, that would be great.

因此，當我們考慮這些大規模、不均勻的部署時，我們是否應該預期這些客戶，例如，在這種情況下，在未來幾季內向其他客戶進行相當強勁的移交？就像比爾一樣，如果你能談談這些收入成長是如何基於專案的，那就太好了。

I think that when -- I guess when I think about your question and we talk about project-based, of course, all of these programs that we're involved with are projects that have been planned for quite some time. And going through the development internally within hyperscaler customers, it's a long process to architect and ultimately bring all of the gear up to a point where they're ready for qualification.

我認為，當我思考你的問題並且我們談論基於專案時，當然，我們參與的所有這些專案都是已經計劃了相當長一段時間的專案。在超大規模客戶內部進行開發是一個漫長的過程，需要建置並最終將所有設備提升到可以進行認證的程度。

Qualification takes a significant amount of time, and then planning the ramp takes time as well. And so we're at the point where we've gotten through that process with -- or nearly through that process with many of the customers.

資格審查需要花費大量時間，規劃坡道也需要時間。因此，我們現在已經與許多客戶完成了流程，或幾乎完成了流程。

I don't really see a handoff so much as I see a diversification in the shipments that we'll be seeing throughout fiscal '26. This is just a function of where we are with each customer. But I don't think we'll see a customer get to a point where they are done with a project.

我實際上並沒有看到太多的交接，而是看到了我們將在整個 26 財年看到的出貨量多樣化。這只是我們與每位客戶關係的功能。但我不認為我們會見到客戶完成整個專案的階段。

There will be a transition from a, say, 50-gig per lane technology to a 100-gig per lane technology. And that, as we've talked about in the past, is -- we are absolutely queued up and ready to ramp. We're the short pole in the tent, so to speak, that -- all of that technology that needed to be developed internally at Credo is done, and we're ready to ramp as soon as our customers are.

將會有一個從每通道 50 千兆位元的技術到每通道 100 千兆位元的技術的轉變。正如我們過去所談到的，我們已經排好隊，準備全力以赴。可以這麼說，我們就像帳篷裡的短桿一樣——所有需要在 Credo 內部開發的技術都已完成，並且我們已準備好隨著客戶的需求而不斷改進。

And so I don't see us getting to the point where we say that, hey, that project is over, and now there's some an air pocket. The key is making sure that we're first, making sure that we're first to deliver next-generation samples, we're first to get qualified, and we're first to ramp production, which really is our strength at this point given the fact that we're taking accountability for the entire system solution.

因此，我認為我們不會走到這一步，說「嘿，那個項目已經結束了，現在出現了一些空洞」。關鍵是要確保我們是第一個，確保我們是第一個提供下一代樣品，我們是第一個獲得資格，我們是第一個提高產量，考慮到我們對整個系統解決方案負責，這確實是我們目前的優勢。

Great. And maybe as a second question, analog copper cable solutions seem to have missed perhaps the B300 cycle. Is that an opportunity for you guys? Do you think you're gaining share and design wins just as that technology did not ramp? Does that increase your TAM and your ramp expectation?

偉大的。也許第二個問題是，模擬銅纜解決方案似乎錯過了 B300 週期。這對你們來說是一個機會嗎？您是否認為，由於該技術尚未發展，所以您就獲得了市場份額和設計勝利？這是否會增加您的 TAM 和坡道預期？

I think that TAM really existed within the NVIDIA ecosystem. The TAM that we're looking at, we've not seen really any customer think seriously about using amplified solutions. So this referring to the analog amplified solution.

我認為 TAM 確實存在於 NVIDIA 生態系中。我們正在關注的 TAM，我們還沒有看到任何客戶認真考慮使用放大解決方案。這指的是模擬放大解決方案。

We haven't really been competing head-to-head with any of those solutions really at any other customer. We really see it within the NVIDIA ecosystem.

我們實際上還沒有與任何其他客戶的任何解決方案進行正面競爭。我們確實在 NVIDIA 生態系統中看到了這一點。

And what role do we play within that ecosystem? That is one that we're pretty conservative in talking about. When there is an opportunity for us to add value, we're absolutely ready, but it's not something that's built into any kind of forecast that we've got.

我們在該生態系中扮演什麼角色？我們在談論這個話題時相當保守。當我們有機會增加價值時，我們已經做好了充分的準備，但這並不是我們所做的任何預測中所包含的內容。

Richard Shannon, Craig-Hallum.

理查德·香農，克雷格·哈勒姆。

Let's see. Maybe I'll just ask a simple question here that I think you get most quarters here, Bill. But I just want to get your latest thoughts on competition in the AEC space. Obviously, you've got a very high share, you're doing exceptionally well here.

讓我們來看看。也許我只想在這裡問一個簡單的問題，我認為你在這裡得到了最多的答案，比爾。但我只是想了解您對 AEC 領域競爭的最新看法。顯然，你們的份額非常高，你們在這方面做得非常好。

It certainly makes sense for some of your large customers to attempt to do dual sourcing. Have you seen any evidence of that, either attempts or see that coming here anytime soon? Just kind of your latest thoughts on that, please.

對於您的一些大客戶來說，嘗試進行雙重採購當然是有道理的。您是否看到任何證據，無論是嘗試還是看到這種情況很快就會發生？請問您對此的最新想法。

I think the fact that the AEC product category is now really de facto for in-rack connectivity, naturally, there is a desire for multiple sources in the market. But the way that we think about competition, it's really -- our objective is always to be the best possible partner to our customers.

我認為 AEC 產品類別現在實際上已成為機架內連接的事實，自然而然地，市場需要多種來源。但我們看待競爭的方式其實是——我們的目標始終是成為客戶最好的合作夥伴。

That means delivering on the innovations first, delivering on the most reliable solutions first, passing qualification first, and getting our customers into production predictably. And so that's -- our focus with our customers is really on creating a relationship with them where they're counting on us to help them ramp.

這意味著首先實現創新，首先提供最可靠的解決方案，首先通過資格認證，並讓我們的客戶可預測地投入生產。所以，我們與客戶的重點實際上是與他們建立一種關係，他們依靠我們來幫助他們發展。

And having the second source thing, something that's a trailing-edge effort. So that's really the way that we think about competition.

並且擁有第二個來源的東西，這是一種後發努力。這就是我們看待競爭的真正方式。

Now I will mention again that being accountable for the system solution allows us to deliver in a faster, more predictable way. Owning every level from the SerDes to the ICs and ultimately to the cable system, it gives us a big advantage from our perspective.

現在我要再次提到，對系統解決方案負責使我們能夠以更快、更可預測的方式交付。擁有從 SerDes 到 IC 再到電纜系統的各個層面，從我們的角度來看，這給了我們很大的優勢。

That's what we're seeing across the board with customers. With that said, we have a lot of respect for competition, but we don't see any significant changes on the competitive front.

這就是我們在客戶身上看到的情況。話雖如此，我們非常尊重競爭，但我們並沒有看到競爭方面有任何重大變化。

Okay, great. My second question, really just looking at the large AEC opportunities here. And clearly, the largest customer is really driving a lot of content here, both, as you said, both in the back end and the front end.

好的，太好了。我的第二個問題實際上只是看看這裡的大型 AEC 機會。顯然，最大的客戶確實在這裡推動了很多內容，正如您所說，無論是在後端還是前端。

I think maybe just thinking about from a back-end perspective, are you seeing any other hyperscalers getting close to adopting and even qualifying your AECs in the back end? And thought process on timing here. I'm assuming it's somewhere in fiscal '26, if not after, but maybe just kind of your latest thoughts there as well.

我認為也許只是從後端的角度來思考，您是否看到其他超大規模企業即將在後端採用甚至認證您的 AEC？這裡是關於時間的思考過程。我假設它是在 26 財年的某個時候，如果不是之後的話，但也許只是你最近的想法。

With our new customers, we see it playing out the same way that it's played out with our existing customers. And it's always starting with the first -- starting with a first project or a first SKU that we're working on, starting in a given rack, and then really expanding from there.

對於我們的新客戶，我們看到它與我們現有客戶的表現是一樣的。它總是從第一個開始 - 從我們正在進行的第一個專案或第一個 SKU 開始，從給定的機架開始，然後從那裡真正擴展。

And so one of the customers that's new, we started with a switch rack, so a disaggregated chassis, a relatively low hurdle 50-gig per lane type of project. But it quickly turned into two additional SKUs that are targeted for AI clusters that they'll be building in the future. And those are going to 100-gig per lane speeds.

對於其中一位新客戶，我們從交換器機架開始，也就是分解式機箱，這是一個門檻相對較低的每通道 50 千兆的專案類型。但它很快就變成了兩個額外的 SKU，用於他們未來將要建立的 AI 叢集。這些將達到每聲道 100 千兆的速度。

And so we see there's always got to be a first program. But the execution that we have on that first program will always lead to a deeper relationship. And we see that playing out the same way with our fourth and fifth customers as we did with the first three.

所以我們看到總是必須有一個第一個程式。但我們在第一個項目上的執行將會帶來更深層的關係。我們發現，第四和第五個客戶的情況與前三個客戶的情況相同。

Suji Desilva, ROTH Capital.

羅仕資本（ROTH Capital）的蘇吉·德席爾瓦（Suji Desilva）。

Congrats on the progress here. I just want to understand from the lead customer and the future 10% customers, just to clarify whether the ramp is today, a single program and expected to be a single program across those 10% customers or whether you get into the point where you have multiple projects across these customers in the pipeline that might diversify you for each customer.

恭喜你在這裡的進展。我只是想從主要客戶和未來 10% 的客戶角度來理解，只是為了澄清今天的成長是否是一個單一的計劃，並且預計會是針對這 10% 客戶的單一計劃，或者您是否已經進入了針對這些客戶的多個項目的階段，這些項目可能會使您針對每個客戶實現多樣化。

I would say that if you look at any of our customers, there's multiple projects. For the first three that we're ramping, there's multiple projects. There's multiple SKUs. And I expect that to be the same with the additional customers that we add, although we always are going to start with the first project.

我想說的是，如果你看看我們的任何一位客戶，你會發現他們有多個項目。對於我們正在推進的前三個項目，有多個項目。有多個 SKU。我希望我們增加的其他客戶也能如此，儘管我們總是從第一個專案開始。

Okay, that's helpful, Bill. And then the opportunity and specifically in back end scale up, I just want to be clear here, is that an opportunity only on in-house ASICs or is that opportunity as well on merchant GPUs in the marketplace?

好的，這很有幫助，比爾。然後是機會，特別是在後端擴展方面，我只是想在這裡澄清一下，這是否只是內部 ASIC 的機會，還是也是市場上商用 GPU 的機會？

It's an opportunity across the board. And so I mean we can talk about the NVIDIA ecosystem, and that's probably part of the TAM that we're not really thinking about too aggressively as we start. But even for the deployments that use NVIDIA GPUs and use -- using the open protocols in the market, PCIe will be an opportunity there as well. So hopefully, that gives you an answer to your question.

這是一個全方位的機會。所以我的意思是我們可以談論 NVIDIA 生態系統，這可能是 TAM 的一部分，我們在開始時並沒有太積極地考慮它。但即使對於使用 NVIDIA GPU 並使用市場上的開放協定的部署，PCIe 也將是一個機會。希望這能解答你的問題。

There are no further questions at this time. And Mr. Brennan, I will turn the call back over to you.

目前沒有其他問題。布倫南先生，我會把電話轉回給您。

Well, thanks for joining us today. Really appreciate the interest and support. As we look ahead, we're absolutely laser-focused on executing on our strategy, delivering value to our customers, and driving long-term growth and profitability. We look forward to the callbacks. Thanks.

好吧，感謝您今天加入我們。非常感謝您的關注與支持。展望未來，我們絕對專注於執行我們的策略，為我們的客戶提供價值，並推動長期成長和獲利。我們期待回電。謝謝。

And ladies and gentlemen, this concludes today's call, and we thank you for your participation. You may now disconnect.

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