AUO targets major revenue contribution from optical engines, glass core substrates within 2 to 3 years

AUO's pivot is fundamentally a dimensional reuse of panel-grade precision manufacturing—repurposing large-area thin-film lithography and etching for optical waveguides and through-glass vias (TGV), hedging against the structural decline of its display business. On the technical chain, glass core substrates directly challenge the physical ceiling of organic BT-resin in AI accelerator packaging. Dies exceeding 100mm² demand warpage tolerances resin can no longer guarantee; glass CTE matching and TGV density make it the inevitable next step in CoWoS-L architecture. Upstream, Schott, AGC, and LPKF laser-drilling equipment see near-term demand pull. Downstream, ASE and Amkor face yield-curve recalibration. The optical engine play targets CPO penetration in 1.6T switch silicon, where AUO's thin-film patterning slots into silicon-photonics waveguide fabrication alongside Broadcom and Marvell designs. Competitively, Intel anchors glass substrate mass production at 2030; Samsung and Absolics push customer validation in 2025. AUO's two-to-three-year revenue window is aggressive, betting that panel-scale area processing compresses unit cost. The risk is acute: TGV yield will likely sit at 70-80% within 12-18 months, while AI customer qualification cycles exceed 18 months—a razor-thin timeline. Layered on top: equipment export restrictions and low-cost glass substrate capacity from mainland China squeeze the cost structure. The 12-24 month litmus test: does AUO secure second-source glass substrate certification in NVIDIA Rubin or AMD MI400 packaging, and does its optical engine bind to a hyperscaler CPO switch design? If neither materializes, "significant revenue contribution" collapses into an R&D-expense narrative.