FICG eyes 4x optical communications capacity by 2027

FICG's simultaneous capacity buildout in Johor and mainland China is not a routine expansion—it is the inevitable downstream transmission of AI's compute arms race into the optical interconnect layer. When GPU clusters scale from thousands to tens of thousands of units, transceiver port density escalates exponentially, and the 800G-to-1.6T generational shift lands precisely on FICG's 2027 ramp window. The company is betting on the structural optical bottleneck of AI infrastructure, not a single product cycle. Upstream, quadrupling output will pull EML laser chips, silicon photonics integration, and fiber preform orders. Downstream, it accelerates CPO from lab to volume. Johor's location is deliberate: the Southeast Asian data center cluster is in its 2025-2027 construction peak, and local production eliminates cross-border yield loss and delivery latency. On compliance, the dual-site strategy is a defensive hedge against escalating export controls. Yet if Malaysian lines incorporate China-originated photonic chips, de minimis rule scrutiny remains a live risk that demands proactive mitigation. Competitively, InnoLight and Eoptolink are locked in an 800G share war, while Coherent and Lumentum build silicon-photonics moats. If FICG cannot anchor to 1.6T or CPO platforms, a pure 800G volume play invites a price slaughter by H2 2026. The 12-24 month swing factors: CPO standard convergence, Malaysia's semiconductor incentive renewal, and AI inference's demand for low-power optical architectures. If all four-fold capacity lands on legacy pluggable modules, structural oversupply looms by 2027.