FN Rides on Surging Data Center Demand: Can It Outpace AAOI & LITE?

Fabrinet's fiscal Q4 2026 data-center revenue rose 68% year over year and 13% sequentially to $669 million, reaching 51% of total revenue, while DCI's annualized revenue run rate exceeded $1 billion. The company plans to add $3.0-$3.5 billion of capacity through Building 10, lifting total capacity to $8.5-$9.3 billion as AI and hyperscaler optical demand ramps through 2027. Competitive intensity remains high: Lumentum's Systems revenue grew 123% to $357 million and Applied Optoelectronics' data-center revenue grew 140.4% to $107.7 million. FN shares are down 11.4% YTD despite consensus EPS of $4.19 implying 43.49% growth and a 20.79x forward P/E, below the sector's 21.36x.

Analysis

FN is transitioning from a diversified contract manufacturer into an AI-optics capacity story, but the equity’s key variable is conversion of customer visibility into committed, profitable utilization. The planned capacity step-up creates operating leverage if hyperscaler ramps hold, yet it also raises the risk that FN absorbs fixed costs ahead of demand; quarterly gross-margin progression and capex-to-revenue conversion matter more than headline capacity. A mix shift toward higher-complexity DCI/HPC assemblies could support both earnings revisions and a multiple re-rating over the next 1-3 quarters, provided concentration with its largest hyperscaler does not worsen.

Competitive positioning is asymmetric. AAOI offers the highest revenue-growth torque but is more exposed to transceiver pricing, yield, and working-capital execution as industry capacity expands; it is the likely loser if 800G supply normalizes faster than demand. LITE is better positioned for the architecture shift toward OCS, lasers and eventually NPO/CPO, where component IP can sustain margins better than conventional pluggables. FN benefits from volume regardless of which component vendor wins, but CPO adoption is a 6-18 month structural risk because it can shift value from external transceiver assembly toward switch-integrated photonics.

Consensus may be treating FN’s relative underperformance as a valuation anomaly rather than a proof-of-demand issue. The near-term rerating catalyst is not another broad AI announcement; it is evidence that next-generation programs ramp on schedule without margin dilution and that backlog/commitment language becomes firmer. Falsify a constructive FN thesis on a sequential utilization slowdown, gross-margin compression despite richer mix, or hyperscaler capex guidance cuts; for AAOI, watch inventory growth and pricing versus unit shipments.

Key Decisions for Investors

• Initiate a 3-6 month long FN / short AAOI pair, sized beta-neutral: FN has more diversified manufacturing exposure and less downside from transceiver ASP compression, while AAOI embeds a more demanding capacity-and-ramp outcome. Target 15-20% relative return; exit if FN reports sequential margin erosion or AAOI demonstrates sustained 1.6T mix expansion with controlled inventories.
• Add LITE on pullbacks ahead of the next two earnings cycles, preferably funded by reducing generic AI hardware exposure: OCS and laser content provide a cleaner 2027 architecture-upgrade option than commodity pluggables. Use a 12-18 month horizon; reassess if OCS revenue fails to scale sequentially or cloud customers defer network switching spend.
• Do not chase FN solely on stated capacity. Set an alert for the next earnings release: upgrade to an outright long only if management quantifies committed programs, guides utilization upward, and preserves or expands gross margin; otherwise the capacity build is a watch item rather than a catalyst.
• Avoid QBTS as a read-through trade. Quantum-computing valuation and commercialization drivers are unrelated to optical-networking demand, and inclusion alongside AI-optics names can create misleading thematic correlation.