Diffractive Deep Neural Network

Cross-cuts: CommunicationsComputeLife & Frontier
last updated 2026-05-27
Spatial Light Modulators (SLMs)Spatial Light Modul…Diffractive Optical ElementDiffractive Optical…Optical Circuit Switching (OCS)Optical Circuit Swi…Multicast RoutingMulticast RoutingMoe All to All CollectivesMoe All to All Coll…Diffracti…

What it is

A D2NN is a stack of diffractive optical elements (DOEs) where each layer applies a programmable phase transformation to an incoming light wavefront. The composition of multiple layers computes complex optical transformations, originally for classification (image → class) and more recently for arbitrary information routing.

Architecture in words: light passes through DOE1 → DOE2 → DOE3 →… → output. Each DOE is implemented as a spatial light modulator (SLM): a 2D pixel array where each pixel applies a controllable phase delay. Reprogram the phase pattern, change the optical transformation.

Trade-off: SLM-class switching speed (single-µs to ms depending on technology — liquid-crystal SLMs slowest, LCOS/DMD faster) bounds the architecture’s applicability. Sub-ns electrical switching is not in reach.

Lineage

The lineage is genuinely a 6-year research arc with peer-reviewed milestones. The pedigree is real. The commercial productisation question is separate.

What it could enable in AI infrastructure

The Zigzag application repurposes D2NN as an optical interconnect fabric for MoE all-to-all collectives:

  1. Native optical multicast without splitter losses. Conventional optical multicast uses a 1×N power splitter (12dB loss minimum for 1×16). D2NN does multicast via constructive interference at multiple output ports directly. No splitter, no fixed pattern, programmable.

  2. Combined wavelength + spatial routing in one device. Conventional OCS does spatial routing only; wavelength routing requires AWG/WSS in series. D2NN handles both in one stack.

  3. Per-layer fabric reconfig if SLMs hit single-µs. Today’s mechanical OCS (Google Apollo) sets topology once per workload. D2NN could potentially reconfigure mid-batch to match current expert-routing patterns.

  4. No foundry queue. Free-space optical elements don’t need a wafer process. Both a moat (no foundry capacity constraint) and a manufacturability risk (alignment + thermal stability + scaling).

Competitive landscape

D2NN-based OCS is one of four credible architectures competing for the MoE all-to-all interconnect prize:

Reading list

Related concepts

Frontier questions