Slow-And-Wide Transceivers Shouldn’t Be Color Blind

ACM SIGCOMM Workshop on Hot Topics in Optical Technologies and Applications in Networking (HotOptics 2026)

Abstract

Modern AI interconnects face hard trade-offs: copper links are efficient but have a short reach, optical transceivers are long-reach but more costly and energy-hungry. Similar trade-offs occur in the switching domain: packet switching is flexible but power-hungry, while circuit switching is efficient but coarse-grained. Recently, slow-and-wide transceivers have closed the gap between copper and traditional optics by exposing many low-rate optical channels, but limit them to a single logical point-to-point link. To address this, we propose ChromaTile, a potential extension of slow-and-wide transceivers with multicolor transmission and passive color steering. It would enable channels to be independently allocated across destinations simultaneously, creating a novel switching abstraction, i.e., optical micro-circuits. We outline the optical components that could be leveraged for the data plane, present one possible control-plane algorithm for collision-free channel assignment, and discuss use cases. Our analysis suggests that ChromaTile can extend the scale-up domain beyond tightly packed racks and reduce scale-out network infrastructure requirements — using ChromaTile with 8 colors reduces the number of switches by ~40% while improving the network latency by one hop. As a position paper, we introduce optical micro-circuits as a new switching abstraction and hope to spark broader collaboration across academia and industry to evaluate and realize this vision.

People

BibTex

@inproceedings{röllin2026slow-and-wide,
  author    = {R{\"{o}}llin, Lukas and Das, Sushovan and Hoffman, Benjamin and Vanbever, Laurent},
  title     = {{Slow-And-Wide Transceivers Shouldn't Be Color Blind}},
  booktitle = {ACM SIGCOMM Workshop on Hot Topics in Optical Technologies and Applications in Networking (HotOptics 2026)},
  address   = {Denver, CO, USA},
  year      = 2026,
  month     = aug,
  doi       = {10.1145/3789240.3822612},
  url       = {https://doi.org/10.1145/3789240.3822612}
}

Research Collection: 20.500.11850/802340