席艾納 (CIEN) 2026 會計年度第三季財報電話會議:在手訂單與 AI 需求帶動業績創新高
Ciena 2026財年第三季營收達16.7億美元,年增37%,調整後EPS為2.11美元,創下歷史新高。受惠於AI基礎設施與資料中心互連需求加速,在手訂單增至85億美元,預期年底將突破100億美元。公司預估2027財年營收將至少成長30%,達83億至84億美元。主要風險在於關鍵零組件供應限制可能持續至2028年,且客戶集中度較高。
Ciena (NYSE: CIEN) 公布創紀錄的 2026 財年第三季財報,主因 AI 相關網路投資加速帶動光學系統與互連產品需求。營收達 16.7 億美元,調整後營業利益率升至 22.5%,調整後 EPS 增加至 2.11 美元。
重點摘要
- 第三財季營收年增 37% 至創紀錄的 16.7 億美元。調整後 EPS 大增 215% 至 2.11 美元。
- 調整後營業利益率達公司創紀錄的 22.5%,較去年同期成長一倍以上。調整後毛利率為 46.4%,其中包括關稅退稅帶來的 70 個基點好處。
- 本季在手訂單增加 8 億美元,達到 85 億美元。管理層預期 Ciena 在 2026 財年結束時,在手訂單將突破 100 億美元。
- 包含互連產品在內的光學網路合計營收成長超過 45%。RLS 與 Waveserver 營收各自成長逾 55%,而互連產品營收則實現翻倍以上成長。
- 展望 2026 財年第四季,管理層預估營收為 17.5 億美元(加減 5,000 萬美元),調整後營業利益率約為 20%(加減 50 個基點)。
- Ciena 對 2027 財年的初步展望預計營收將至少成長 30%,達到 83 億至 84 億美元,若取得額外供貨能力則具備進一步上行空間。
核心財務數據
| 指標 | 2026 財年第三季 | 變動或背景 |
|---|---|---|
| 營收 | 16.7 億美元 | 年增 37%;創單季新高 |
| 調整後毛利率 | 46.4% | 年增 450 個基點;包含來自關稅退稅的 70 個基點 |
| 調整後營業費用 | 4 億美元 | 落在管理層指引區間的低端 |
| 調整後營業利益率 | 22.5% | 年增翻倍以上;創公司新高 |
| 調整後 EPS | 2.11 美元 | 年增 215%;創公司新高 |
| 自由現金流 | 1.16 億美元 | 扣除營運資金投資後 |
| 現金及現金等價物 | 28 億美元 | 截至第三財季末 |
| 在手訂單 | 85 億美元 | 本季增加 8 億美元 |
業務與營運表現
Ciena 表示,傳統廣域網路、AI WAN 基礎設施以及資料中心內部與周邊連接的需求正全面加速。公司目前約有 50% 的業務直接來自超大規模雲端業者,同時管理層也看到美國及國際市場新興「neoscalers」的需求持續成長。
包含互連產品在內的光學網路合計營收年增超過 45%。RLS 與 Waveserver 系統各自成長逾 55%,直接來自雲端提供商的營收更成長超過 80%。今年迄今,來自資料中心內部及周邊所產生的營收比重已成長至四倍。
WaveLogic 6 Extreme 的採用速度已超越先前 WaveLogic 5e 的放量速度。管理層指出,該平台在問世 18 個月後,仍是市場上唯一的 1.6 Terabit 高效能數據機。800ZR WaveLogic 5 Nano 可插拔光學元件在第三財季的出貨量呈季增翻倍以上。
Ciena 估計 RLS 在解耦光線路系統市場佔有率約為 70%。其下一代 RLS Hyper-Rail 平台仍按計畫進行,預計在 2026 日曆年末前取得首批客戶標準化認證。管理層預期 Hyper-Rail 營收將在 2027 財年放量增長至數億美元,不過具體速度將視零件供應情況而定。
公司還收到來自數家主要核心客戶對其開放式共封裝光學解決方案 Vesta 的樣品訂單。Ciena 預計其 CPX 業務的營收將於 2027 年開始產生,並於 2028 年加速放量。
管理層認為,隨著光互連擴展至 WAN 與資料中心架構,到 2029 年 Ciena 的總可定址市場 (TAM) 可能從約 250 億美元擴大至約 500 億美元。
管理層財務指引
| 期間與指標 | 管理層指引 |
|---|---|
| 2026 財年第四季營收 | 17.5 億美元,加減 5,000 萬美元 |
| 2026 財年第四季調整後毛利率 | 45%,加減 50 個基點 |
| 2026 財年第四季調整後營業費用 | 約 4.15 億美元,加減 1,000 萬美元 |
| 2026 財年第四季調整後營業利益率 | 約 20%,加減 50 個基點 |
| 2026 財年營收中位數 | 64.2 億美元 |
| 2026 財年調整後營業利益率 | 20%–21% |
| 2026 財年資本支出 | 預計落在 2.5 億至 2.75 億美元區間的高端 |
| 2027 財年營收 | 至少 83 億至 84 億美元,代表至少成長 30% |
| 2027 財年調整後毛利率 | 至少 45%–46% |
| 2027 財年調整後營業利益率 | 25%–27% |
2027 財年展望為初步預估。管理層將 83 億至 84 億美元視為基準線,並表示若 Ciena 取得額外的供應產能,營收有可能更高。
風險與觀察重點
供應仍是營收成長的主要限制。Ciena 已就部分關鍵零組件簽訂至 2029 年的長期協議,並正在認證額外的供應商與技術。管理層預計產業供需在 2028 年之前不會恢復平衡。
這些供應協議將需要額外的現金支出。由於公司需要為產能與供應承諾提供資金,預計第四財季的營運現金流將會下降。
第三財季毛利率獲得的 70 個基點關稅退稅效益屬於一次性會計影響。管理層指出,在當前框架下,關稅對利潤率大致維持中性,不過在採取緩解措施前,加拿大新的關稅制度可能會帶來每季約 1,000 萬美元的影響。
客戶集中度亦是一大重點:兩家客戶各佔第三財季營收的 10% 以上。管理層另外提到,受客戶集中影響,其客製化數據機消費模式的營收可能會出現波動。
分析師問答焦點
- 定價與合約條款:Ciena 預計經談判後的價格調幅將介於高個位數至 10% 後段或 20% 出頭的區間,具體取決於客戶與產品。部分漲價將選擇性地適用於在手訂單。討論內容亦涵蓋付款條件、履約率及互惠採購量承諾。
- 在手訂單品質:管理層將在手訂單的成長主要歸因於交貨期延長,而非投機性下單。預計財年結束時的大部分在手訂單,其客戶要求的交貨日期均落在 2027 財年,並涵蓋公司 2027 財年營收展望的大部分內容。
- AI 基礎設施持續性:管理層表示,目前可見的大部分需求皆與連接現有資料中心以及為 GPU 升級擴充網路容量有關。因此,Ciena 預計未來兩年受新資料中心建設放緩風險的影響有限。
- 橫向擴充 (Scale-across) 商機:Ciena 將資料中心間的橫向擴充互連視為強勁的成長驅動力,且目前仍處於早期階段。目前的部署主要集中在美國,初始用於支援分散式 AI 訓練。
- Hyper-Rail 經濟效益:管理層預計 Hyper-Rail 的毛利率將高於目前公司的整體平均,並隨著 2027 至 2029 財年營收規模擴大而帶動整體獲利增長。
- 電信營運商需求:受北美、印度、日本及中東部分地區的託管光纖網路活動支撐,Ciena 看好電信業者對光學基礎設施的持續投資。
電話會議完整逐字稿
完整財報電話會議逐字稿
管理層陳述
Operator
Hello, everyone. Thank you for joining us, and welcome to the Ciena Fiscal Q3 2026 Financial Results Call. [Operator Instructions]
I will now hand the conference over to Gregg Lampf, Vice President, Investor Relations. Gregg, please go ahead.
Gregg Lampf
Thank you, Jennifer. Good morning, and welcome to Ciena's 2026 Fiscal Third Quarter Conference Call. On the call today is Gary Smith, President and CEO; and Marc Graff, CFO. Scott McFeely, Executive Advisor is also with us for Q&A.
In addition to this call and the press release, we've posted to the Investors section of our website an accompanying investor presentation that reflects this discussion as well as certain highlighted items from the quarter. Our comments today speak to our recent performance, our views on current market dynamics and drivers of our business as well as a discussion of our financial outlook. Today's discussion includes certain adjusted or non-GAAP measures of Ciena's results of operations. A reconciliation of these non-GAAP measures to our GAAP results is included in today's release.
Before turning the call over to Gary, I'll remind you that during this call, we'll be making certain forward-looking statements. Such statements, including our quarterly and annual guidance, commentary on market dynamics and discussion of opportunities and strategy are based on current expectations, forecasts and assumptions regarding the company and its markets, which include risks and uncertainties that could cause actual results to differ materially from the statements discussed today. Assumptions relating to our outlook, whether mentioned on this call or included in the investor presentation that we posted earlier today are an important part of such forward-looking statements, and we encourage you to consider them. Our forward-looking statements should also be viewed in the context of the risk factors detailed in our most recent 10-K and our forthcoming 10-Q. Ciena assumes no obligation to update the information discussed in this conference call, whether as a result of new information, future events or otherwise.
As always, we'll offer as much Q&A as possible today, though we ask that you limit yourselves to one question and one follow-up.
Now I'll hand the call over to Gary.
Gary Smith
Thanks, Gregg, and good morning, everybody. Today, we reported record financial results across the board that we demonstrated outstanding third quarter performance, including revenues of $1.7 billion, another quarterly record and up 37% year-on-year. Adjusted operating margin of 22.5%, exceeding guidance was more than doubling year-on-year and the highest ever achieved for the company. Our adjusted earnings per share are up 215% year-on-year to a record $2.11.
We delivered results in the context of an extraordinary industry demand environment that continues to accelerate. We continue to see strong momentum in customer demand and order flow, with a Q3 book-to-bill ratio that was significantly greater than 1, which resulted in a substantial quarterly increase in our backlog. And we also expect backlog to grow at an even greater rate in Q4. In fact, just 1 quarter -- sorry, 1 month into this quarter, we are approaching a level of orders booked equal to the entirety of Q3. As a result, we are currently projecting to exit fiscal 2026 with over $10 billion in backlog. Overall, our outstanding Q3 performance reflects Ciena's essential role in the fundamental re-architecting of network infrastructure.
And looking at these industry dynamics, I would remind everybody that we remain in the very early stages of a multiyear, highly durable network investment era. This is springboarding and caused by the large and growing investments in data center infrastructure. AI is starting to build on the previous eras of communications, including those driven first by the Internet and then by the cloud, but it is doing so at a massive scale. As a result, AI is currently driving and will continue to drive significant increases in both bandwidth connectivity demand and network traffic growth.
In that context, high-speed, low-latency optical connectivity has become a critical enabler to not only operationalizing the AI-driven investments in the network and the data center, but also monetizing those investments over time. And because of the increasing demands for higher capacity, faster speed, greater density, improved reliability, reduced space and lower power and cost, optics have become the indispensable element for next-generation AI architectures. And this is manifesting across all 3 of our primary markets. First, you call it the traditional network or the wide area network, the WAN, encompasses the network backbone, network edge and network operations. It includes optical connectivity for long haul, subsea, metro regional applications that people are familiar with. And it has also been impacted by AI in a number of ways. From challenges of fiber availability in the backbone to quality of service demands at the edge to the requirements of automation to address the increasing complexity of network operations.
Second is a market that we are referring to as AI WAN. It includes both data center interconnect or DCI for the WAN backbone and scale across currently used for distributed training across data centers and subsequently to be used for inferencing. Here, the fundamental challenges are related to power caused by the increasing GPU compute capacity and energy load required to train large language models at scale and the high volume, low power demands of deploying modems at much greater scale. The third is, of course, the data center themselves, which includes the fabric connectivity domains of scale up and scale out as well as data center operations. And as AI continues to drive up the data rates and bandwidth requirements inside the data center, new optical technologies and applications are required to provide the needed improvements in capacity and density for short-reach low-power connections.
Given the acceleration and projected increase in the compounding waves of spend on network infrastructure across these markets, we continue to believe that the total addressable market for our business will effectively double over the next 3 years, growing from approximately $25 billion today to approximately $50 billion by 2029. Moreover, given our growing competitive advantages, we expect our share of that TAM to continue to increase over that time frame. More specifically, Ciena's long-established technology leadership in optical networking positions us to capture a growing share of wallet as optical connectivity expands its role throughout the WAN and inside the data center. Across generations of Coherent technology, Ciena's first-to-market benchmarks have set the bar for the industry and continue to do so. Ciena was the first to commercialize coherent optics decades ago, and we continue to lead the industry in optical innovation, backed by very focused R&D deep expertise and proven deployment at scale.
Moving forward, performance gains will increasingly depend on precisely these capabilities. Because of our leadership position and value proposition, we've developed a high degree of competitive differentiation across our portfolio, with the clearest proof being the customer adoption that we're seeing across our portfolio in each of the primary market segments. So starting with both the traditional WAN market, as I outlined, and the AI WAN. Today's market dynamics are driving higher adoption rates for our WaveLogic 6 Extreme platform, which after 18 months is still the only 1.6 terabit, high-performance modem on the market today. Notably, its ramp has already exceeded that of our prior generation WaveLogic 5e.
Separately, customer adoption and scaling of our intelligent line systems remains exceptionally strong. RLS is basically the industry standard in disaggregated optical line systems where Ciena's first mover advantage has driven a leading installed base where roughly we have 70% market share. In addition to serving cloud providers and service providers in the network backbone and cloud providers for DCI in the AI WAN, RLS is the industry's first system deployed for scale-across applications. And the next generation of RLS Hyper-Rail is our second generation of RLS and represents our sixth generation of photonic line systems leadership. Co-created with the hyperscalers, it dramatically increases the density of existing optical amplifier infrastructure and as such, is purpose-built to address the needs to distribute AI training workloads in data centers across greater distances. With customer orders ramping, we remain on track for initial customer standardization for RLS Hyper-Rail by the end of 2026, and scaling to material revenue as we move throughout 2027.
Turning to our interconnects portfolio. We're applying our optical leadership to a growing portfolio of connectivity solutions that address surging bandwidth demands inside and around the data center and the performance limitations, of course, of today's short-reach technologies. Starting with our WaveLogic 5 Nano pluggable optics, we are seeing strong market adoption as we continue to ramp into production volume. In fact, in Q3, we shipped more than twice the volume of 800 ZR plugs than in the previous quarter. In addition, during the quarter, we made strong progress with the components portion of our interconnect portfolio. We are seeing strong market receptivity to Nitro, a linear redriver for active copper cable solutions. And I'm pleased to report that we received sample orders from several anchor customers in the ecosystem for Vesta, our open co-packaged optical or CPX solution. This represents another important step towards the commercialization of our open ecosystem approach towards short-reach data center optics. And we believe this is gaining meaningful industry momentum. Most importantly, with potential customers. As in any new growth sector, our CPX business will continue to strengthen over time with revenue expected to begin in 2027 and ramping into 2028.
And finally, it's worth noting that last quarter, we announced a significant win with a major hyperscaler that integrates our WaveLogic 6e coherent technology into their own platform. This solution goes well beyond the modem and combines our DSP, drivers, TIAs and Coherent expertise into a complete module that will be deployed broadly across the customer's global optical network via their own optical platform. I think this win demonstrates our ability to deliver for our customers across multiple consumption models with our best-in-class portfolio, and this represents a significant takeaway from a component competitor.
At the highest level, the current and future waves of AI-driven demands on bandwidth and network traffic will require industry-leading high-speed optical connectivity. We remain focused on managing the business with this long-term view, supported by durable demand, a broad set of co-creation opportunities and customer design wins robust orders and a backlog that extends well into fiscal 2028. Looking forward, the strength of our market position and the breadth of our portfolio provides us with growing confidence and visibility into a multiyear runway of growth, operating leverage and increasing profitability. As a result and to add to this level of confidence, we recently secured a significant increase in customer commitments that extend through 2029.
At the same time, as Marc will discuss in a few moments, we've also secured incremental supply capacity for critical component optical components to service that multiyear demand. So in summary, Ciena's unmatched combination of leading optical technologies, incumbency, portfolio breadth and deep expertise across systems, components, software and services gives us a powerful and sustainable competitive advantage. And really as the only pure-play optical systems and interconnect vendor operating at scale, we are uniquely positioned to convert AI-driven demand into durable top line growth with increasing operating leverage and earnings power over multiple years, delivering differentiated value for our customers and our shareholders.
With that, I'll hand the call over to Marc for an update on our financials and our outlook.
Marc Graff
Thank you, Gary, and good morning, everyone. As Gary just discussed, our focus is on delivering strong financial performance while ensuring security of supply and manufacturing capacity necessary to support the significant multiyear demand in front of us. Within that context, we continue to make excellent progress against our 3 financial priorities.
First, let me discuss our progress on gross margin. We achieved 46.4% adjusted gross margin this quarter, which was positively impacted by the treatment of tariff refunds by about 70 basis points. Even without the tariff impact, gross margin achieved the top end of guidance, reflecting disciplined cost execution, favorable mix and pricing discipline. Our midterm goal is to structurally position the company to achieve mid-40s gross margins. With our results over the past several quarters, we are confident that we have reset our margin baseline to this mid-40s goal. But as we've said over the past year, the mid-40s goal was a waypoint, not the final destination. We have the cost structure and leadership portfolio to expand gross margins further over the next few years.
Second, as we balance the investments to support the growth of our business, working capital remains a focus. While our cash conversion has taken a step back quarter-on-quarter, the overall trend is positive relative to the year ago results. We've invested working capital to support slightly higher inventory levels and to increase revenue through the quarter. Even with these investments, we've generated $116 million in free cash flow. Third is capital allocation. We continue to take meaningful steps to improve the operational and financial efficiency of the business. Our June convertible debt issuance achieved 2 specific goals. First, it lowered our cost of capital with a 5-year 0 coupon instrument at an economic conversion premium of 114% from which we retired our 5.5% interest term loan.
Second, it provided the capital to help secure supply over the next 3 years. We maintain strategic flexibility as a result and have $2.8 billion in cash and equivalents at the end of Q3. Our capital investments this year have increased capacity sufficient to support RLS, plug and Waveserver revenue growth, all over 60% year-to-date. Additionally, we continue to return capital to our shareholders in Q3. We repurchased 356,000 shares for an aggregate price of $172 million, reflecting the acceleration associated with the convert deal. Lastly, we expect to be on the high end of our capital expenditure range of $250 million to $275 million.
Now let's move to the quarterly results in more detail. As Gary noted in his opening remarks, revenue achieved $1.67 billion at the top end of our guidance, an increase of 37% year-on-year and another quarterly record. Our total combined optical networks revenue, including interconnects, grew over 45% year-on-year, supported by over 55% growth for both our RLS and Waveserver systems. Our interconnects more than doubled year-on-year, while our direct cloud provider revenue grew over 80%. Our in and around the data center percent of revenue has quadrupled year-to-date, well ahead of our committed 3x growth from the beginning of the year. We had 2 customers that each contributed more than 10% of revenue. And lastly, we exited Q3 with an $800 million increase in backlog to $8.5 billion.
Orders continue to accelerate, as Gary noted, 1 month into the quarter, we have booked nearly as much demand as all of Q3 and expect to end the year with more than $10 billion in backlog. As I noted earlier, adjusted gross margin was 46.4% exceeding the top end of our guidance by 90 basis points and up 450 basis points year-on-year. Q3 adjusted operating expense was $400 million, coming in at the low end of our guide and driving a record adjusted operating margin of 22.5%, 250 basis points over our guide and more than doubling the year ago results. Adjusted EPS reached $2.11 more than triple the year-ago figure and achieving a new record level for the company.
Now let's move to guidance for the last quarter of the year. In Q4 '26, we expect to deliver revenue of $1.75 billion, plus or minus $50 million, raising the full year midpoint to $6.42 billion, up $120 million from last quarter. With this revenue guide, we expect to increase our market share in the combined optical systems and plug market by about 4 points to approximately 30%. We expect adjusted gross margins of 45%, plus or minus 50 basis points, bringing the year to a similar range, a raise of 50 basis points from last quarter. Adjusted operating expense will be roughly $415 million, plus or minus $10 million, with our annual OpEx at 1.6, slightly down from the June guide. All told, we expect to drive an adjusted operating margin of approximately 20%, plus or minus 50 basis points, bringing the full year to between 20% and 21% and exceeding the 20% annual figure for the first time in the company's history.
Now let me address the topic of supply more closely. Gary described an unprecedented durable demand environment with backlog now into 2028. To ensure our ability to service this demand, we've taken decisive steps to strengthen our supply security and to increase our output. In recent weeks, we finalized long-term agreements that secure supply of certain key components through 2029, including incremental capacity for those key components that will enable us to support growing customer demand. This extends the investments we've made for this year to drive 35% revenue growth. We continue to invest upstream to drive security of supply to meet and eventually bring into balance the demand backlog. As a result, we expect to see a reduction in cash from cash from operations in Q4 as investments are disbursed to support these agreements.
At the same time, we continue to make progress in the value exchange discussions with our customers. In addition to price discussions, these conversations are increasingly focused on the alignment of various demand terms and conditions with the capacity required to support them. With this ongoing momentum and improved visibility into future demand, we believe it's prudent to provide early direction for fiscal 2027. As we see it today, we expect to deliver another record year with revenue growing a minimum of 30% year-on-year yielding at least $8.3 billion to $8.4 billion in revenue, with supply-driven upsides. Our investments in capacity and supply allow us to accelerate absolute revenue growth from '25 to '26 and now into '27.
At these levels, we expect to again increase our market share in optical systems and plugs in fiscal 2027. We expect gross margins to be at least between 45% and 46%. And we expect to achieve fiscal 2027 adjusted operating margin between 25% and 27%, posting yet another record in profitability and serving as yet another proof point for the earnings potential of Ciena's model. Again, this is our preliminary view of 2027 and we'll provide an update when we report our Q4 results in December. In the interim, we look forward to continuing the dialogue in a few weeks' time in Ottawa at our Investor Forum, the content from which will be posted on our investor website afterwards.
To close out, Q3 was a testament to the strength of Ciena's technology leadership, customer engagements and supply resiliency in the face of unprecedented multiyear demand. The execution of our business model has driven an acceleration of our earnings in Q3 in 2026, and we now believe into '27 and beyond.
With that, operator, we'll now take questions from our sell-side analysts.
Operator
Thank you. We will now begin the question-and-answer session. [Operator Instructions]
Your first question comes from the line of George Notter with Wolfe Research. Please go ahead.
分析師問答
George Notter
Congrats on the terrific results here. I guess I wanted to start just by, you mentioned value exchange on the call. Certainly, I think you mentioned prices as well as alignment of demand terms and conditions. Could you just talk a little bit more about what's going on there? Any sense for what price increases might look like? Any sense for what terms and conditions might be looking like in terms of the context of value chain exchange?
Marc Graff
George, it's Marc. Thanks for the question. I'll take it kind of in 2 parts. On the pricing piece, we've had conversations with customers across different product lines. And we've gotten to a space where we would expect, depending on the customer and the product line anywhere between, call it, high single digits types of price increases to something in the range of high teens, low 20s type of price increases. And what's remarkable, and I think you'll appreciate this, George, is some of that will selectively hit backlog, right? So I think we've made really good progress there.
The second pillar in terms of conditions is really a 2-way discussion. The first is we're on the hook to make sure that we deliver what we say we're going to deliver, but we expect the reciprocity of that from the customer side as well. And so we've covered those aspects. We've talked a little bit about payment terms. We've talked a little bit about fill rates and things like that. So we're trying to make it a pretty holistic conversation in terms of that value exchange. and not just have a conversation about price. Because just like we're looking for supply security. Our customers are looking for supply security from us as well. And it's something that we feel pretty confident that with the supply agreements that we can fulfill.
George Notter
Got it. Super. And then I know that there were some price increases. I think earlier in the year last year, just around tariffs. Is that something that's flowing into the model now? I know that in the past, you guys weren't -- you were not repricing backlog certainly, but is that something that's helping the gross margin now? Any sense there?
Marc Graff
Yes. It's -- this is Marc again, George. It's relatively neutral. We're not putting margin on top of tariffs, right? If we get $10 of tariffs, we kind of pass on that $10 of tariffs. What we saw in Q1 was kind of -- or I'm sorry, in Q3 was kind of a onetime accounting adjustment for those tariff refunds that we don't expect to continue moving forward. And that gave us about 70 points -- 70 basis points of uplift. But moving forward, I would say the tariff impact, again, under today's current regime, is relatively neutral. We're monitoring pretty closely some of the impacts that are coming out of the new Canadian tariff regime. That could have an impact of, call it, $10-ish million a quarter. But again, we're still trying to work through the mitigation actions that we've got associated with that.
Operator
Your next question comes from the line of Tal Liani with Bank of America. Please go ahead.
Tal Liani
Gary, if I told you 3 years ago that you're going to grow 30% with 26% margin, you would have asked to drink the same thing I'm drinking. So the question I have is about backlog. So your backlog is doubling this year, and it grows even faster than revenues. Your revenues are growing fast, and it grows even faster than revenues. I'm trying to understand the early ordering portion of the backlog, the maybe customers are buying ahead just because of supply constraints. I'm not -- it's not a concern. I just want to understand kind of get understanding of how backlog could behave in 2027. That's the reason for my question.
Gary Smith
Okay. I think it's almost entirely just driven by a function of lead times. The demand is absolutely there. And just to sort of illustrate that. Marc gave an early indication of what we think our guidance is for the euro just a really directional indication for next year, it would be greater than that if supply was greater. I mean that sort of, I think, summarizes it.
We've got -- we think at least a $10 billion backlog as we leave this year. And in the midpoint of what Marc was talking about, you're looking at revenues of 8.3, 8.4 as sort of baseline for us for next year, it would be greater than that if we had more supply. And so the demand is absolutely there. You look at our installation services, they're up 35% for the year. And as soon as we can ship it, it's installed and carrying traffic.
Marc Graff
Yes. Tal, maybe I'll just add maybe a little bit more context here. If you look back all the way back to 2024, our orders, call that demand doubled from '24 to '25. From '25 to '26, we're expecting another 50% increase. And as you rightly noted, backlog is doubling across all 3 of those years from '24 to '25 was a double from '25 to '26 is a double. And so what we're really constrained with is the industry needs to add a significant amount of capacity to keep up with that demand. And so we think it's going to be a multiyear journey before we see that supply and demand get back into balance and multiyear. So we don't see that happening before '28 at all. And so I think you'll see a very similar constrained dynamic going into '27 and likely into '28.
Tal Liani
Got it. And any -- if I can just ask a follow-up. Any color on customer composition, meaning hyperscalers, I understand. What about smaller hyperscalers, meaning new clouds and new hyperscalers like Oracle? So without names of customers, but can you discuss your ability to kind of grow the customer list over time? And where is the demand?
Gary Smith
Yes. About 50% of our business is now hyperscalers directly. But increasingly, I think to your point, we're seeing the sort of neoscalers, umbrella of neoscalers, which covers multitude of different business models, et cetera. We are very focused on that space. They are leaning very much into networking now and they are securing networks on MOFN deals. They're beginning to put their own fiber in when they can get it. and we are taking more than our fair share of that market as it grows. So we are very focused on addressing that market, both in the U.S. and globally. We're seeing that in certain parts of the world where these neoscalers are investing in the networking. So I think as we go through '27 and '28, that will become an increasingly important part of our business.
Operator
Your next question comes from the line of Meta Marshall with Morgan Stanley.
Meta Marshall
Great. A couple of questions. Maybe just following up on George's question. Just in terms of some of these new arrangements that you guys are having with -- or discussions that you're having with customers, is some of those -- are some of those pricing adjustments dependent on time line of delivery like in terms of if you can deliver 6 months earlier, you can capture high single digits versus a mid-single-digit price adjustment? Just trying to get a sense of whether there's any kind of escalators in there? And then second question, just as you guys look to assure more supply, have you qualified additional suppliers at this point? Or is this largely reaching long-term agreements with existing suppliers?
Marc Graff
Yes. Thanks, Meta. It's Marc. So on your first question in terms of escalators, we really haven't built those in, like the price increases that we've talked about aren't necessarily performance-based per se. They will cut in as more and more backlog from those orders becomes a bigger part of our revenue. So I wouldn't say that it's performance related. Once we agree to those price increases, it's really around when we deliver it, they'll pay for it.
In terms of qualifying new suppliers, yes, that is part of our supply resiliency strategy that we're driving. We've got our typical providers that are in the table that you know very well, but we are looking at expanding both the type -- the numbers of suppliers that we have as well as we're constantly looking at new technologies to satisfy the same type of functionality. So we're taking both a quantity as well as a technology perspective to our supply chain.
Operator
Your next question comes from the line of Joseph Cardoso with JPMorgan.
Joseph Cardoso
I'll share my congrats as well on the results and guidance here. Maybe another backlog question and more on the composition of it. As we think about the portfolio offerings that you guys have and maybe the several irons in the fire that you guys are trying to address, where are you seeing the strong demand inflection as we were entering the back half of the fiscal year? And as a second part to that, it's great to hear that you're seeing visibility now into '28, but any color you can provide on the weighting of orders coming in for '27 versus '28, essentially, just trying to get a better understanding of how much of '27 is already covered versus what is building for '28 now? And then I have a follow-up.
Gary Smith
So the first part of that question, Joe, thank you, is really we're seeing broad demand across the portfolio and you'd say characterized as being line systems, both in terms of the existing RLS and Hyper-rail. We've got a number of new wins for hyper rail that we're beginning to -- will begin to ramp up during '27. So there are a lot of infrastructure going in for that, I mean, and think scale across, not entirely, but predominantly are the deployments for that, that's driving it.
And then on the modem side, we shared some of the statistics that you're seeing for WaveLogic 6. We doubled output, it's already at this point, exceeding 5e in terms of its adoption. And I think that, again, just talks to the need for high-speed distance for these kinds of applications. So we're seeing it on the modem side. And of course, we're seeing it on the infrastructure of line systems. We're also seeing that both in terms of MOFN deals globally as well to support this expansion, particularly markets like India and the Middle East. And certain parts of Asia where the hyperscalers are leaning into provisioning of extension of their networks. Submarine as well, massive build-outs going on across the global submarine market, where we have #1 market share in the world. So we're seeing that across it, Joe.
And in terms of the profile of the backlog, we've got -- as you said, we'll probably have about $10 billion plus backlog as we go into 2027. We cannot satisfy basically all of the requirements that they would take to deliver all of that in '27.
Scott McFeely
Joseph, the vast majority of that $10 billion comes with the customer request date, that's actually in '27, meaning that they would take it if we could give it to them. So you -- 2 questions. The backlog covers most of the '27 guide.
Operator
Your next question comes from the line of Ruben Roy with Stifel.
Ruben Roy
Yes. Gary, for the first question, I wanted to maybe drill into the performance optics discussion and sort of the consumption model that compared to the systems model. Can you -- is that a bespoke arrangement with 1 customer? Are you productizing this consumption model as you go forward? And I guess, as you think about that longer term, and how that sits in the interconnect family, if you could talk a little bit about the margin structure as that consumption model starts to build.
Scott McFeely
Ruben, it's Scott. So first of all, I kind of ask the question in 2 different angles about how I heard it from you. First of all, the performance modem portfolio, the WaveLogic Extreme family, if you like, WaveLogic 5 and then WaveLogic 6. It's obviously very broadly deployed solution within our systems business. So we have a lot of deployments out there on extreme volumes. The specific opportunity you're referring to, though, of taking that and offering it up in a different consumption model is bespoke relationships with individual customers. We have 2 examples of that today. One that was a recent announcement last quarter.
It's certainly something that we don't shy away from. We made the technology available. However, our customers want to consume it. But in those examples, those are very, very unique in terms of how those customers want to deploy them, so they're kind of custom development for them. And the relationship we have with those customers reflects that.
Ruben Roy
Okay. And then as a quick follow-up for Marc, sorry if I missed this, Marc, but with the 25% to 27% operating margin guidance for '27 or first look at '27, that implies, I think, roughly flat to maybe up a little bit, operating expenses. If you could just walk us through sort of the mechanics around operating expense as you look out into fiscal '27, that would be helpful.
Marc Graff
Yes. No problem, Joe. We haven't really closed in yet. We're kind of in the middle of our annual planning process. But the puts and takes that you should kind of be thinking about is this year, we'll spend roughly, call it, $1.6 billion. Keep in mind that $1.6 billion includes a bunch of onetime variable compensation that a year ago, we were telling you it was going to be about $1.5 billion. We're at $1.6 billion, mostly because of that variable comp. We're going to reinvest that onetime. So that's -- when you say it's about flat, all the folks sitting around the table here are looking at $100 million more of investment that they get regardless of the performance of the company. So we are reinvesting those onetime things.
And I think you'll also see there'll be a little bit more investment in some of the activities, particularly around line systems as we continue to grow that business and invest in our interconnect portfolio.
Operator
Your next question comes from the line of Ryan Koontz with Needham & Co.
Ryan Koontz
Great. In light of some of the politics around data center construction and the like, which I know weighs on investor minds a lot and it's a lot of the broader sector around. How do you feel about the pace of catch-up of your WAN projects relative to data center construction? Do you feel like you've got visibility independent of pacing of data centers in that light? Maybe you can comment on that, Gary.
Gary Smith
Yes. No, listen, it's a great topical question. I would say that as we talk to the hyperscalers and we talk about durability of demand and their long-term view and the rest of it in getting long-term agreements with them and commitments. So part of that, one of the comments that was made to me was basically that if they stopped building data centers tomorrow, Gary, you probably wouldn't notice for 2 years. Meaning, they've already got these data centers out there, and they need connectivity and they're not going to strand the assets. And secondly, they've got data centers that they must increase the network capacity to.
And so largely, what we've got in backlog here and what we've got visibility to going forward is really the data centers that are already there. And particularly, you've got a lot of international expansion as well. And it's really the refresh of the GPUs in these data centers that are already there that require massive amounts of additional connectivity. And bear in mind, we have a unique insight into this because we've got #1 market share in data center connectivity before all of this AI expansion began. So we have got the connection to most of the data centers around the world. A lot of what we're seeing is the expansion and increasing of that capacity and connectivity to enable the refreshing of the GPUs, et cetera. And you've also got all of the inference and agentic stuff in front of us.
So Ryan, certainly, for the next couple of years, we think we're largely immune from what may or may happen with the pacing of new data centers.
Ryan Koontz
That's great. And maybe as a follow-up, any commentary on the product mix here as you've seen like in the most recent quarter or maybe recent bookings in terms of shift of line systems versus pluggables and transponders, any commentary there?
Marc Graff
Yes, maybe I'll jump in and others can add color, Ryan. So as I think through what we've seen, particularly over the last 12 months, I think we've seen our plugs and our line systems, particularly RLS, really grow at much higher than corporate average growth rates, right? I think I mentioned plugs and RLS together as part of our optical piece growing 45%. So you kind of see how that's becoming a bigger piece of the pie. One of the things that from a margin perspective that we're seeing is, as our DCOM solution really starts to increase over the last year. That's really driven pretty accretive dynamic for us moving forward. We expect that to continue. And then obviously, as we add Hyper-Rail, that's going to be another accretive motion for us.
And so I think what you're seeing is the optical piece of our portfolio really driving a bunch of the growth for the company. And then obviously, we've got the DCOM piece, which is shown in route and switch, really in the early part of its ramp as well.
Scott McFeely
I think, Ryan, I mean, a dynamic that's been going on since 2024. We're just seeing more and more demand for line systems, meaning more fibers are getting it. And those are getting lit with coherent optics of all flavors, whether it be plugs or performance optics consumed in Waveserver. And all those, to Marc's point, are up well north of the 35% or 37% that we're reporting as a corporate average. And that's going to continue, we think, going into the foreseeable future. The DCOM piece is a great adder, but it's a bit lumpy because of the concentration of the customers. So from quarter-to-quarter DCOM will come and go. But it's a net new add for us.
Operator
Your next question comes from the line of Tim Long with Barclays.
Timothy Long
Appreciate it. Two for me as well. Maybe first, if we could dig a little deeper into Hyper-Rail, I mentioned it a few times here on the value-add side and ramp. Just kind of update us on -- it sounds like a few customers, but where are we in the demand profile? And how quickly could we see the ramp of this product? And kind of just to remind us on the economics versus like more the RLS prior generation. And then the follow-up would be on just the pure telco business, maybe ex [indiscernible], if you could just talk a little bit about the durability of that business in the past, that's been a little bit more cyclical. So just curious of the outlook on just the pure telco piece.
Marc Graff
Yes. I'll start on Hyper-Rail, Tim, and then others can jump in. We're on track for getting that product to standardization by the end of this calendar year, and you'll see the ramp starting in and that ramp in '27 will be to several hundred million dollars, right? So we're looking at that as a pretty meaningful ramp for us. Yes. The back story on that is it probably could be faster if we get more components, right? So obviously, we're working day and night on that.
From an economics perspective, relative to RLS, I think the team has done a fantastic job of improving the margins over the last 4 to 8 quarters on RLS to get us to a pretty decent margin profile. Hyper-Rail will be a step function on top of that. And with the size of the ramp and the opportunity that we think is coming through with hyper rail, and the economics of that, it's going to be accretive to the company as a whole once we get into '27, '28, '29. So we're really looking forward to getting Hyper-Rail out there. And I think our customers are placing quite a few orders that's represented in that $10 billion of backlog that we expect at the end of the year.
Gary Smith
And on service provider growth, it's actually quite difficult to separate it from a lot of the MOFN activity that's going on. And we know the MOFN activity is high. But I'd say there's 2 things going on with the service provider piece. One, it is growing anyway because I think there's been underinvestment in optical infrastructure in the last 5 years. And you've got the service providers returning to drive out infrastructure for optical infrastructure. and you've also got this MOFN piece. And you've seen that phenomenon now, certainly in North America.
If you go back about 18 months, it was very much an international phenomenon. But now with training and the rest of it, we're seeing that very much so in North America. And that is driving a lot of the -- particularly the wholesale market in the U.S. and the wholesale carriers that specialize in that, we're seeing very strong growth in that space. and we expect that to continue. Markets like India, particularly for MOFN, we're seeing explosive growth in provisioning of MOFN networks for multiple Hyper-Railers in places like India, Japan, I would also highlight and then certain parts of the Middle East. So we expect to see good, steady service provider growth continue over the next few years, irrespective of the MOFN phenomenon.
Operator
Your next question comes from the line of Simon Leopold with Raymond James.
Jeffrey Koche
Jeff Koche in for Simon. I really wanted to ask on -- first question on the software business. It doesn't really appear like the web-scale RLS deployments are a driver here. Is that like kind of the right interpretation? Is that because they have their own solutions? And to that end, like how do you win the RLS deals if it's not like a management platform take play? I have a follow-up.
Scott McFeely
Scott here. Yes. I mean, your hypothesis that there's less off-box software components in a web-scale deal in general, not just an AI deal, is valid. That's a fair statement. However, having said that, to your second question, don't take that comment to mean that the only thing these guys are buying is merchant hardware from us because the value that they get is much broader than that. and whether it's submarine networks, their existing backbone, their DCI networks are their scale across networks, the statement is true across the piece they're getting -- yes, the hardware platform, but some very sophisticated on-box software capabilities that is embedded into their back office system that has an awful lot of intelligence in it. protection mechanism, et cetera, to allow them to deliver to their SLAs.
They're getting planning tools and deployment tools, they're getting link engineering tools. They're getting a global across the world service capability to turn these things on, preposition them, preconfigure them and turn them on a set of skills that we've developed with them and their relationships for more than a decade now. So just because we're not selling as much off-box software components to service providers don't conclude, therefore, it's just like a commodity hardware sale. It's apart from that.
Jeffrey Koche
Great answer. So maybe just with that in mind, can you maybe give a little bit of color on how the gross margins are for that business and maybe how they're changing just maybe even just relative to the average.
Marc Graff
You're talking about the software, the off-box software?
Jeffrey Koche
For the line system. For the line of systems.
Marc Graff
Yes. So as I said previously with Tim, we've seen really good improvements in the RLS gross margins over the last 2, 3 years. And I expect that to continue, and those are approaching what I would call the corporate average. As we move into the next-generation Hyper-Rail, the economics get significantly better, right? And so those will be above the current corporate average. And I would expect with the size of opportunity that we have with Hyper-Rail over the next couple of years, that will be accretive.
Jeffrey Koche
Terrific. Terrific. And then if I could just do another follow-up on the interconnect business and really just inside the data center. We know that Google you hear like Google is looking to deploy 2.4 terabit Coherent-Lite solution for CPUs. Maybe talk about -- are you in those deals? Are you being evaluated? What's your take there? And what's your take on optical circuit switching? Is that a product that Ciena would explore?
Scott McFeely
Yes. So a couple of things. There's a couple of questions there. The Coherent moving inside the data center, we said for a long time now that we think that's a trend that is inevitable, and it's going to happen, and we're committed to that. We absolutely believe in the Coherent-Lite market. We think the right intercept for that for the general market is at 3.2 terabits. And we think we'll be in a great position to be a leader in that in that market.
In terms of OCS. We love OCS because it is part of the continuation of more optics inside the data center. And it will drive actually adoption of coherent inside that data center faster than without OCS in our belief system. So that's all good news. That's a separate answer whether or not we're going to jump into the OCS market ourselves, and we're not going to comment on plans, make any product announcements or that on the call today.
Gregg Lampf
We're going to move on to one last question.
Operator
Your last question comes from the line of Tim Savageaux with Northland Capital Markets.
Timothy Savageaux
Congrats on the results and especially the guide, and that's kind of the focus of my question, which is along several lines, the case for accelerating revenue growth in fiscal '27 looks particularly strong. Whether we're talking about anecdotal commentary, backlog, hyper rail scale across, which will be my focus on my follow-up. And I know you sort of termed this as an initial guide, but I'd be interested in your commentary on the prospects for delivering accelerating revenue growth in fiscal '27. I know you mentioned supply as a constraint. And maybe what things might look like if that constraint were relaxed a bit.
Marc Graff
Yes. Tim, it's Marc. I think you kind of answered your own question. As we look at it -- and you're right, we are early, right? And typically, we wouldn't do this. But as we look at the demand -- or the dynamics that we're seeing in the market, we thought it was prudent to give our owners and the investment community, at least some initial thoughts on what we think the floor will be going into 2027.
But as you rightly pointed out, all of our focus right now is on how do we get more supply to get to get more of that demand. And as Gary said and Scott said as well, if we could get more demand, we would unwind that $10 billion of backlog faster right? And that 8.3 to 8.4 that we talked about is the floor would be higher. And so from an absolute dollar terms, we think we are accelerating the growth from '24 to '25 to '26 and into '27, excuse me, but it's really going to be dependent on that supply. And a year ago, when we did this for the first time, we said we thought '26 would grow 17% and here we are at 35%. Now I'm not suggesting that my 30% in the year is going to be 60%, right, because we're in obviously a different supply environment. But we want to make sure that we give you guys a floor and make sure that we can achieve at least that number that we give you while we continue to work on supply.
Timothy Savageaux
Great. And as a quick follow-up on scale across, I mean, to what extent is that maybe even the primary driver of growth in '27? And I'd be interested in your reaction so some pretty extraordinary comments from suppliers and competitors about dynamics and scale across, I think Cisco talking about 14x the port count versus traditional DCI and some pretty spectacular comments from Lumentum as well. Maybe we can sharpen the focus on the scale across opportunity, how you see that TAM having maybe increased in recent quarters?
Gary Smith
Yes, Tim, I think it's a major driver of demand. And we were the first out there with the first scale across piece that came out of the data center. So we've got good visibility to it. I concur with most of the industry comments that's gone on to it. I think, it's excuse the pun, it is at a massive scale, and it's just beginning. That's the point I would make is we're just beginning to roll out the first connectivity between these data centers. It is almost entirely North American, U.S. based, and we're just beginning to link the first few data centers for a couple of hyperscalers to it. And that's all in front of us. So yes, it's a massive driver to it.
But I also -- we're also seeing just a general increase in connectivity around the data centers as well. The agentic stuff is beginning to flow, particularly on the submarine cables. And the inference traffic, we also think is a big step function. It's mainly in front of us. So everywhere you look, basically, Tim, you're looking at compounding waves of applications and traffic growth that we'll just build on top of each other because even the scale across which is really predominantly now on training, started off with synchronous training. You're going to get asynchronous training as well. You're also going to get large amounts of inference cascading into that as well. So massive amounts of connectivity between these data centers in front of us. And we're only just at the early innings of that. And we are incredibly well positioned to it, having the leading platform for Hyper-Rail, RLS was an industry standard. We have about 70% of that market share, and we expect that to continue with Hyper-Rail and this next generation and the leading modem technology we can basically move bits faster and longer than anybody else in the world. And that's a super valued critical element that will enable this.
Gregg Lampf
Thanks, Tim, for the question. Thanks, Gary. We look forward to seeing everyone over the next several weeks at a very busy schedule. Thanks for your time this morning.
Operator
This concludes today's call. Thank you for attending. You may now disconnect.











