SEMI and Global Net Corp. published a glass core substrate market report putting the segment at 67.2% CAGR from 2028 to 2040, with initial production beginning in selected high-performance applications around 2028. The numbers are industry forecasts, not bookings, but the directional claim is load-bearing: organic ABF cores, the dominant substrate technology for advanced flip-chip and 2.5D packages today, are running out of room on three axes simultaneously: flatness tolerance, electrical loss at high frequencies, and thermal expansion mismatch with silicon at large form factors.
Glass fixes all three. Glass substrates have a coefficient of thermal expansion closer to silicon than ABF, which matters for chiplet assemblies where die stacking forces and substrate warpage interact over the product lifetime. Glass also supports finer lines and spaces at lower dielectric loss, which is directly relevant to the interconnect density required for 1.6T Ethernet optics and CPO integration. The reason glass has not shipped at volume is yield: glass is brittle, polishing glass to semiconductor flatness tolerances at scale is hard, and the handling equipment supply chain is not built yet.
2028 production start means the equipment ecosystem has roughly 18 months to finish qualifying the tooling. For hardware teams designing systems that will tape out in 2025 to 2026 for a 2027 to 2028 silicon delivery, glass substrates are not a current option; they are a design-point choice for the generation after next. The decision that matters now is whether your next-generation package architecture locks you into ABF assumptions that will be structurally disadvantaged by 2030, or leaves room to migrate. The SEMI report is a planning anchor, not a procurement signal.