If the central theme of AI computing power investment in 2024 was the GPU, then by 2026, the landscape is quietly shifting. As large-scale AI cluster construction expands from single data centers to cross-domain interconnections, the bottleneck of the "computing power highway" is moving from chip processing capabilities to network transmission bandwidth. Data center interconnect (DCI) demand, centered on optical communications, is rapidly becoming one of the fastest-growing segments in AI infrastructure spending.
Recent financial reports confirm this trend. On June 4, 2026, Ciena (NYSE: CIEN) released its FY2026 Q2 earnings, posting quarterly revenue of $1.57 billion—a 40% year-over-year increase and above the consensus estimate of $1.5 billion. More notably, adjusted earnings per share reached $1.64, nearly four times the prior year’s figure ($0.42), and far exceeding analyst forecasts of $1.46. Management attributed this robust performance to the continued acceleration of network investment by hyperscale cloud providers and the explosive growth in demand for optical interconnects driven by AI workloads.
For its annual revenue outlook, Ciena raised its FY2026 full-year guidance to a midpoint of $6.3 billion, up from the previous range of $5.9–$6.3 billion, representing a year-over-year growth rate of about 32%. The company also expects Q3 revenue to be between $1.575 billion and $1.675 billion, with a midpoint of $1.625 billion—again, higher than the market expectation of $1.555 billion. These upward revisions signal that AI-driven infrastructure spending has yet to show any signs of marginal slowdown.
Order Backlog and Capacity Shortages: The Core Dilemma of Demand Outpacing Supply
Behind Ciena’s outperformance lies a structural tension in AI computing power network construction that cannot be ignored. By the end of FY2026 Q2, the company’s backlog had increased by more than $600 million quarter-over-quarter, reaching a total of $7.7 billion, with around $6.4 billion in hardware orders. During the earnings call, management made it clear that if the supply chain allowed, customers would accept even more product deliveries. This supply-demand gap highlights that capacity constraints for key optical network components remain a bottleneck for industry growth.
From a customer perspective, hyperscale cloud providers now account for roughly one-third of Ciena’s revenue, with direct cloud customer income up 70% year-over-year. Against this backdrop, the company not only achieved strong revenue growth but also improved profitability—Q2 adjusted gross margin reached 44.9%, expanding by 390 basis points year-over-year. This indicates that with supply limited, Ciena has gained stronger pricing power—something rarely seen in the historical cycles of the semiconductor and optical communications industries.
Valuation Premiums Hit New Highs: What Is the Market Racing Toward, and What Is It Warning Against?
In sharp contrast to its strong fundamentals, Ciena’s share price actually dropped about 5.7% in pre-market trading after the earnings release. By June 5, the stock closed at $488.21, down $8.85 from the previous trading day.
The root cause of this pullback lies in extreme "front-running" of valuations. As of early June 2026, 20 analysts covering Ciena had set a 12-month average target price of $530.56, with the highest at $720 (Rosenblatt) and the lowest at $185. Notably, Morningstar issued a rare "1-star" rating, estimating that Ciena’s share price carries about a 115% premium over its fair value. GuruFocus’s valuation model suggests that, based on the current price of $500.60, Ciena is overvalued by roughly 440%.
Even so, "buy" ratings (13) still significantly outnumber "hold" ratings (7), and several institutions raised their target prices following the earnings report. Rosenblatt increased its target from $350 to $720, while Needham moved from $470 to $600. This phenomenon—stellar earnings but a declining stock price—essentially reflects a valuation system overhaul in the AI computing power infrastructure sector. The market’s pricing for AI-linked assets has shifted from "can growth be delivered" to "is the growth potential already fully priced in."
Beyond Ciena: Which AI Optical Network Industry Chain Players Remain Undervalued?
Ciena’s earnings send a clear signal: AI data centers are moving from isolated computing power competition to network interconnect competition, and the capacity barriers and pricing power of optical network infrastructure are reshaping the profit distribution across the AI industry chain. Within this framework, several sub-sectors deserve attention.
Optical Modules and Components: Lumentum and Coherent
Optical modules are the core hardware for optical interconnects in AI data centers, responsible for converting optical and electrical signals within data centers and across wide-area networks. Lumentum (NASDAQ: LITE) reported FY2026 Q1 revenue of $533.8 million, up 58% year-over-year, with non-GAAP EPS at $1.10—well above last year’s $0.18. More importantly, Lumentum’s data center order shipment ratio exceeded 4x, with lasers sold out through 2027. This metric directly reflects the urgent demand for advanced optical components.
Coherent (NYSE: COHR) is also benefiting from the expansion of AI infrastructure. FY2026 Q1 revenue reached $1.58 billion, up 17.3% year-over-year, with Q2 guidance midpoint at $1.63 billion, driven by continued penetration of DCI ZR/OCS/1.6T VCSEL and other technologies. JPMorgan recently rated Coherent "overweight" with a $380 price target. From a supply chain perspective, Lumentum and Coherent’s gross margins are also benefiting from the tightening supply logic, but their market valuations have yet to fully reflect the sustainable demand for optical interconnects.
Optical Module Contract Manufacturing and Packaging: Fabrinet
Fabrinet (NYSE: FN), a leading contract manufacturer of optical modules and components, is also showing notable growth within the supply chain. FY2026 Q1 revenue reached $978.1 million, up 21.6% year-over-year, beating the upper end of guidance. Particularly impressive, its data center interconnect business grew 92% year-over-year and 29% quarter-over-quarter. About 35% of Fabrinet’s revenue comes from data center products and 49% from telecom equipment, both highly dependent on hyperscale cloud provider capital expenditures. With ongoing capacity tightness in the optical network industry, contract manufacturing’s order overflow effect is often one of the most direct beneficiaries.
Optical Fiber and Cable: Corning
The geographic spread of AI clusters is driving new demand for optical fiber infrastructure. In May 2026, NVIDIA signed an expansion agreement with Corning (NYSE: GLW) potentially worth up to $3.2 billion, with plans to use equity as a basis to expand US fiber capacity to meet AI cluster interconnect needs. Corning’s share price surged about 18% in a single week in May 2026, reflecting a market revaluation of fiber assets in the AI computing power chain. Compared to optical modules and switching devices, fiber capacity construction cycles are longer and more capital-intensive, meaning that once supply and demand become imbalanced, pricing elasticity may persist longer.
Network Switching Equipment: Arista Networks
While Ciena mainly addresses wide-area interconnects between data centers, Arista Networks (NYSE: ANET) focuses on internal switch architecture within data centers. Needham analysts note that about 54% of Arista’s revenue comes from hyperscale cloud customers. The company has raised its 2026 AI-related revenue target from $2.75 billion to $3.25 billion, reflecting ongoing expansion of AI cluster scale. With accelerated adoption of 800GbE switches, Arista’s market share advantage in high-speed switching is translating into sustained performance growth.
Network Testing Equipment: Viavi Solutions
The increasing complexity of AI optical networks is driving simultaneous growth in network testing demand. Every new generation of network speed requires updated testing equipment. Viavi Solutions (NASDAQ: VIAV) holds a solid market share in this segment, with testing demand closely synchronized to capital expenditure cycles. As the industry evolves from 400G to 800G and even 1.6T, the market for testing equipment is also expanding.
Looking across the industry chain, profits in AI computing power infrastructure are cascading from "GPU—optical module—optical switch—optical fiber and cable—testing equipment." Ciena’s valuation premium signals to the market that AI-driven optical network demand is resetting the sector’s valuation benchmarks. Meanwhile, upstream segments like optical components and modules have yet to see their valuations fully reflect their sustained benefits, making them potential areas of structural undervaluation in the current computing power narrative.
Gate Stock Trading: One-Stop Cross-Market Allocation for AI Computing Power Infrastructure Assets
For investors looking to track these industry chain dynamics, traditional cross-border investing often faces numerous hurdles, including account systems, funding channels, and trading hours. As digital asset platforms and traditional finance converge at an accelerated pace, these barriers are gradually being dismantled.
In June 2026, Gate officially launched real stock trading services, allowing users to trade US mainstream securities market stocks and ETFs directly with USDT on the platform. This covers more than 10,000 US stocks and ETFs. Gate has formed a strategic partnership with Alpaca, a licensed US brokerage, connecting directly to major exchanges like NYSE and NASDAQ to provide real stock price discovery and settlement—not on-chain tokenized products.
Gate’s stock trading feature supports direct trading of all core assets mentioned, including Ciena, Lumentum, Coherent, Fabrinet, Arista Networks, and Corning. Users can access the "Stocks" section via the Gate app or web, search for the relevant ticker, select the quantity, and submit orders to buy—no need to switch between multiple platforms or open separate overseas brokerage accounts. Gate spot stock trading involves no funding rates or overnight holding fees, making it more suitable for long-term allocation strategies compared to traditional CFD products. In the future, the platform will gradually support margin trading and one-click stock asset transfers between brokerages, further enhancing cross-market allocation flexibility and asset flow efficiency.
Conclusion
Ciena’s FY2026 Q2 performance is not just a single company’s earnings beat—it’s a bellwether for the shift in AI infrastructure investment logic from "centralized computing power" to "distributed interconnection." Optical networks are moving from the periphery to the core of the AI industry chain, while upstream segments like optical components, modules, fiber, and testing equipment may be relatively undervalued in the current structural rally. For investors looking to position ahead of the curve in AI computing power infrastructure, Gate stock trading offers a convenient pathway from the crypto asset ecosystem to direct participation in these sectors—eliminating the complexity of traditional brokerage processes and enabling multi-asset cross-market allocation within a unified account.
