Cross-cutting photonic component appearing in 26+ concept pages.
Physics / mechanism
Devices that imprint a spatial pattern onto a coherent or incoherent light beam by modulating amplitude, phase, or polarization pixel-by-pixel. Three dominant technologies today:
Performance trade-offs: LCOS gives best phase resolution but is slow (≤120 Hz) and limited by liquid-crystal birefringence dynamics; DMD is fast but binary; MEMS is fast and continuous but small array. The frontier — bespoke metamaterial / III-V SLMs — promises GHz rates with continuous phase control on production-scale arrays, but manufacturing has historically been impossible at volume.
Competitive landscape
Tom Walton-Pocock (Geometry, 2026-05-06) summarises the current frontier:
“Prior efforts to do III/V SLM all failed because manufacture was impossible … you have to make these tiny refractive units and III/V units have been notoriously tricky to make to wire in. The SLM standard now is everything is on a backplane. Building sufficiently many quantum wells that work all on the same wafer is combinatorially extremely hard. Microsoft tried this a lot — they have several papers.”
Implication: the dominant architectural pattern is active optics layer + silicon-backplane electronics (the LCOS pattern, but extending to non-LCOS materials). Lawrence’s hypothesis: III/V on silicon is the eventual stack. BTO is on the wrong side of the SLM/modulator divide — it’s a modulator material, not an SLM material.
Companies using
Connected ideas
Sources
Frontier (open questions)
- Microsoft’s III/V SLM papers — Tom referenced multiple; pull via
/kb-research. What pixel pitch, refresh rate, yield achieved? - Neurophos copper-metamaterial spec — pixel pitch, switching speed, addressing scheme. Where Geometry’s diligence landed.
- OLIX architecture — Tom + Geometry don’t have material clarity. Watch for disclosures.
- GF III/V-on-Si capability — does any GF process node host III/V quantum wells with backplane integration?
- Foundry path for active metasurfaces — Neurophos-class metamaterials need bespoke fab; is there a path to standard foundry?
- 2PP parallel-foci ceiling — current published SOTA 49 foci (Hahn/Wegener 2024); what unlocks 500+?
- BTO-as-SLM — Tom’s view is “modulators only.” Worth verifying with Altro: is there any device geometry where BTO functions as SLM?