Enlightra has emerged from stealth with a $15 million seed round to commercialise a multiwavelength laser platform designed to reduce the energy footprint of AI data centre interconnects. The startup is positioning optical links as the next step for scaling GPU clusters without proportionally growing power and cost.
Tackling AI’s Power Bottleneck
Modern AI training and inference increasingly rely on high-bandwidth GPU interconnects. Traditional copper links are hitting physical limits: bandwidth density cannot keep pace without sharp rises in power consumption and heat. That constraint slows cluster scaling and forces data centre operators to trade off performance against energy budgets.
Optical Solution for Next-Gen AI
Enlightra replaces many power-hungry electrical links with compact optical communication. Its comb-laser platform produces multiple stable wavelengths from a single source, consolidating what would otherwise be dozens or hundreds of discrete lasers. Each wavelength serves as an independent data channel, enabling dense wavelength-division multiplexing over a single fibre or short-reach optical link. The result is lower power per bit, smaller footprint, and reduced BOM costs compared with one-to-one laser approaches.
Funding and Future Scale
The $15 million round includes participation from Y Combinator and Runa Capital among others. That capital will support pilots with AI cluster vendors and ramp design for large-scale manufacturing using silicon photonics. Silicon photonics lets Enlightra move from lab prototypes to wafer-scale production, which is key to deploying the technology across data centres, subsea systems, and chip-to-memory interconnects.
By replacing energy-inefficient electrical links with multiwavelength optical solutions built for mass production, Enlightra aims to decouple performance growth from energy and cost increases. For operators and investors focused on sustainable scaling of AI, the company presents a viable path to maintain accelerated model throughput without runaway energy demands.
