NLM Photonics adds key investors in quest to reduce the power needed to move data between chips
$13M for the power problem nobody can ignore: moving data between chips costs more than computing it.
Why it matters
As AI workloads scale, interconnect power is becoming the bottleneck. Photonic I/O—encoding data on light instead of electricity—could reshape data-center economics and unlock denser GPU clusters. NLM's funding signals the venture market sees this as infrastructure-critical.
The key facts
11 to knowNLM Photonics raised $13M from Mitsubishi Chemical Ventures and Pangaea Ventures
Materials-on-silicon approach: photonic data encoding layered on standard silicon chips
Use case: inter-chip communication with lower power consumption than electrical interconnects
Relevance: AI compute buildout is power-constrained; photonic I/O is a candidate solution to interconnect scaling
Timing: September 2026—venture backing for compute infrastructure plays remains strong
$13M funding round
Investors: Mitsubishi Chemical venture arm, Pangaea Ventures
Technology: photonic materials on silicon for optical data encoding
Problem solved: power efficiency of inter-chip communication
Seattle-based startup
Application: reducing power draw for data movement between chips (critical AI infrastructure bottleneck)
Go to the source
GeekWiregeekwire.com
Publisher excerpt: The Seattle startup's materials sit on top of silicon chips and imprint data onto a beam of light using less power. Mitsubishi Chemical's venture arm and Pangaea Ventures joined a $13 million round.