Startups Bring Optical Metamaterials to AI Data Centers

2026-03-19 · Source: IEEE Spectrum · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Emerging Technologies & Innovation · Depth: Advanced, short

Summary

Two startups, Lumotive and Neurophos, are leveraging metamaterials to enhance data center bandwidth and accelerate artificial intelligence processing. Lumotive, which debuted a new microchip on March 19, has developed metamaterials with electronically programmable liquid crystal elements to precisely steer, lens, shape, and split light beams. This technology aims to improve optical circuit switches, offering a no-moving-parts solution that can scale up to 10,000 by 10,000 ports, with initial products expected by late 2026. Neurophos is developing metamaterial-based optical modulators, the optical equivalent of transistors, that are 1/10,000th the size of current designs. These chips, built using standard CMOS processes, are projected to offer 50 times greater compute density and energy efficiency than Nvidia's Blackwell-generation GPUs, with proof-of-concept chips for hyperscalers this year and systems targeting early 2028.

Key takeaway

For CTOs and infrastructure architects evaluating next-generation data center and AI compute solutions, Lumotive's and Neurophos's metamaterial-based technologies present a compelling alternative to conventional electronic and existing optical systems. Your teams should monitor the performance and scalability of these emerging optical switches and AI processors, especially Lumotive's planned 2026 launch and Neurophos's 2028 production target, to assess their potential for significant improvements in energy efficiency and compute density.

Key insights

Metamaterials offer a path to overcome bandwidth and power limitations in data centers and AI computing.

Principles

• Programmable metamaterials enable dynamic light manipulation.
• Miniaturization of optical components boosts compute density.

Method

Lumotive uses electronically programmable liquid crystal elements within copper structures to steer and shape light. Neurophos employs metamaterials to create ultra-small optical modulators for AI processing.

In practice

• Replace MEMS/silicon photonics in optical switches.
• Develop high-density, low-power optical AI processors.

Topics

• Metamaterials
• Optical Computing
• Data Centers
• AI Processors
• Optical Switches

Best for: Investor, Entrepreneur, CTO, AI Engineer, MLOps Engineer, Research Scientist

Related on AIssential

• This Breakthrough Could Kill the GPU Race — Optical computing with active metasurfaces offers a path to dramatically reduce AI's energy footprint by performing computation directly with light.
• Meta Memo Outlines Ambitious Hardware Plans, Including New AI Pendant — Meta is planning new AI hardware, including a pendant, as part of its ambitious device strategy.
• Penn physicists use light-matter particles to boost AI chip speeds — Exciton-polaritons enable all-light signal switching in semiconductors, potentially enhancing AI chip speed and efficiency.
• This Will Power Everything — Optical interconnects are critical for scaling AI data centers beyond copper's physical and thermal limitations.
• Optical Implementation of Equilibrium Propagation Using Spatial Photonic Ising Machines — A Spatial Photonic Ising Machine enables energy-efficient optical-digital Equilibrium Propagation for training energy-based networks.
• Scaling Power-Efficient AI Factories with NVIDIA Spectrum-X Ethernet Photonics — Co-packaged optics in Ethernet switches significantly enhance AI factory scalability, power efficiency, and network resilience.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by IEEE Spectrum.