The photonic-active cousin of Glass Interposers. Glass-interposers = glass carrying electrical routing; glass-photonics = glass carrying light.
Stack context: glass photonics is one option in Layer 1 (fibre-to-chip optical coupling) of the PIC packaging stack, competing with polymer Photonic Wire Bonding and edge/grating coupling. It is not a Silicon Photonics replacement — SiPh is the active platform; glass is the coupling/passive-routing medium around it. Full landscape: 2026 05 30 Pic Packaging Approaches Landscape.
Physics / mechanism
Glass photonics writes optical waveguides inside a glass substrate (typically fused silica or borosilicate) using femtosecond laser direct-write: ultrafast pulses locally raise the refractive index along a 3D path, defining a waveguide buried in the bulk. This is fundamentally different from silicon photonics, which patterns waveguides lithographically on the surface of a silicon-on-insulator wafer.
The load-bearing limit: curvature radius. Laser-written glass waveguides cannot take tight bends (like an optical fibre, bend too hard and light leaks). So no compact ring resonators, no dense routing. Glass is structurally confined to passive routing + the fibre interface + wide-bend point-to-point links. It does not win modulators (TFLN/BTO territory). For meter-scale chiplet-interconnect substrates this matters less; for dense PICs it is disqualifying (Photonics Material Class War).
Competitive landscape — the glass-photonics cohort
Adjacent but different category: Akhetonics (Germany, all-optical general-purpose compute chip, ~€8.3M raised) appears as an “Optoscribe competitor” in databases but is optical compute, not glass interconnect/packaging — though it consumes a laser-etched glass interposer + photonic wire bonding (Fraunhofer-IZM SPOC project).
Contrast with the electrical glass-substrate primes (Corning, Absolics/SKC, AGC, Samsung, KCC, LG) tracked under Glass Interposers — those carry copper, not light.