The rapid evolution of artificial intelligence models from centralized computing training to distributed computing clusters and edge inference applications is triggering a historic upgrade cycle for global optical network hardware infrastructure. JPMorgan recently held an in-depth post-earnings discussion with the investor relations team of leading optical network company Ciena (CIEN.US), including head Gregg Lampf and Patti Trautwein. In a subsequent research report, JPMorgan highlighted that Ciena, with its industry-unique 1.6T wide area network solution WaveLogic 6, has successfully penetrated the core supply chains of hyperscale data centers, with a massive $7.7 billion order backlog demonstrating significant operational control.

Service Provider Growth Drivers Accelerating: MOFN Expands Domestically

Ciena management noted that the current robust growth among service providers is primarily driven by three core factors. First, there is a need to address infrastructure gaps. In recent years, traditional telecom and network service providers heavily prioritized 5G capital expenditures, leading to chronic underinvestment in underlying fiber infrastructure. Operators are now working hard to close the foundational network construction gap that emerged post-pandemic. Second, there is explosive growth in Metro Optical Fiber Networks (MOFN). To support massive data center interconnect (DCI) demands driven by AI, service providers are actively deploying MOFN, which now accounts for roughly 15% of Ciena's service provider-related revenue. Third, providers are positioning for AI inference traffic. Operators are extensively renovating network architectures to ensure their networks do not become a bottleneck for AI monetization when inference traffic surges. Notably, Ciena emphasized that while MOFN-driven capital expenditure was initially concentrated in international regions, this demand is now rapidly expanding within the United States, indicating the AI-induced fiber network reconstruction is fully underway domestically. Ciena collaborates deeply with partners like Lumen in this area, and Lumen has signed $8.5 billion in AI fiber agreements with Microsoft, AWS, Google, and Meta. Recently, Verizon selected Ciena's WaveLogic 6 Extreme solution to complete a 1.6 Tbps data transmission trial on a live fiber network, further validating the feasibility of AI-optimized fiber.

Order Visibility Extends to 2027: "Co-Creation" Model Dispels Double-Ordering Concerns

In traditional hardware cycles, a major market concern is customer "double-ordering" during supply chain crunches, which inflates demand. However, JPMorgan notes Ciena's current order book is exceptionally healthy with minimal signs of double-ordering. This is attributed to the "co-creation" partnership model Ciena has established with hyperscale data center operators. In this deeply integrated model, Ciena participates directly in designing the client's next-generation network architecture, while clients share their future product roadmaps openly. Management's confidence in the $7.7 billion backlog stems from this: these orders represent equipment destined for immediate deployment, not pandemic-era panic buying. If Ciena can deliver early, clients order more—there is almost no evidence of double orders. This deep collaboration allows Ciena to manage project-based order flow with exceptional clarity, with visibility extending directly into 2027. The total backlog stands at $7.7 billion and is expected to grow through fiscal 2026. Conversion rates are impressive, with approximately 80% of the $6.4 billion hardware backlog expected to be delivered within the next 12 months, indicating high order realization efficiency.

WaveLogic 6 Gains Ground: Capturing DSP Business, RLS System Secures Early Wins

On the technology front, Ciena is leveraging its generational advantage to outperform competitors. Using its latest WaveLogic 6 (WL6) coherent optics technology, Ciena successfully captured the core digital signal processor (DSP) business from an existing supplier for a leading global hyperscale data center operator. Previously, this client used industry-standard components for optical transport. WaveLogic 6, as the only 1.6T WAN solution on the market, helped the client transition to a next-generation network for AI workloads. Competition in the 2026 800G coherent pluggable market will be intense, but Ciena holds a leading position with WL6's symbol rate and high-performance DSP design differentiation. Ciena has also secured the industry's first "multi-rail" system order (the RLS Hyper-Rail solution) from a top hyperscale cloud provider for remote AI training. Furthermore, in the high-profile remote AI training scenario, Ciena secured the first order for its Ultra-High-Speed Rail System (RLS) earlier than anticipated. Company advisors noted that for transmission distances exceeding roughly 1,000-2,000 kilometers, the RLS system will use full modems rather than pluggable modules, increasing the proportion of photonic components and revenue opportunities. The RLS Hyper-Rail platform is expected to be standardized by late 2026, with volume deployment in 2027. Ciena is already in discussions with most major hyperscalers regarding similar deployments. The platform uses a unique long-haul optical amplification architecture, promising significantly improved space and power efficiency for interconnecting AI training clusters across campuses.

Financial Perspective: Vertical Integration Boosts Margins, Variable Compensation Lifts Short-Term OpEx

Driven by a robust product mix, Ciena's fiscal Q2 2026 results and Q3 gross margin guidance significantly exceeded Wall Street expectations. The discussion clarified that Q2 gross margin expansion to 44.9% was driven by three factors: continued product mix optimization with higher-value products, design-side cost reduction initiatives, and improved pricing power (due to supply constraints), with price increases taking effect about six months earlier than previously expected. Operating expense (OpEx) growth was almost entirely driven by variable compensation—as orders and revenue exceeded expectations, variable compensation payouts increased Q2 expense recognition. Excluding compensation volatility, management views the operating model as highly efficient. Significant operating leverage exists as R&D resources are shared across WaveLogic product forms and solutions. Beyond the high-value premium from WL6, the company is accelerating vertical integration. For instance, it has brought ITLA pump laser production in-house and plans to replicate this vertical integration in line systems, aiming to tightly control supply chain costs and further support gross margin expansion.

Strong Results but Stock Pullback: Market Anticipated Gains, But Fundamentals Remain Solid

In stark contrast to the strong fundamentals, Ciena's stock fell approximately 13.66% following last week's earnings report, with intraday volatility reaching 14%. The core reason for this pullback is extreme valuation anticipation—prior to earnings, Ciena's stock had surged over 600% in the past year and nearly 200% year-to-date, with a forward P/E approaching 90x. The stock currently trades around $445. Even after the correction, the current trailing P/E remains elevated. However, market sentiment is recovering. As of June 10th, the overall analyst rating remains "Buy" (13 Buys, 6 Holds), with an average price target around $559. Multiple firms, including UBS ($508), Morgan Stanley ($490), and Rosenblatt ($720), raised their price targets post-earnings. The "strong results but stock retreat" phenomenon essentially reflects a shift in the market's pricing logic for AI infrastructure stocks from "can they deliver growth?" to "is the growth potential already fully priced in?"—a trend not unique to optical communications stocks. The strongest signal from Ciena's earnings is not just the quarterly numbers but a structural shift in AI investment: as computing clusters scale, demands for data transmission speed, density, and latency increase—the bottleneck for large-scale cluster construction is gradually shifting from chip computing power to network transmission bandwidth. Hyperscale cloud provider capital expenditure is projected to exceed $720 billion by 2026, almost entirely driven by AI infrastructure demand. Management forecasts the total market size will roughly double to about $50 billion by 2029. This growth encompasses not only the traditional WAN market but also high-growth markets within and around data centers—such as scale-out and scale-up interconnects inside data centers. As copper-based solutions approach their limits for short-reach interconnects, optical interconnects are poised to play a larger role in connections within and between racks.