Polaris Raises $50M for AI Photonics
Polaris Electro-Optics has raised $50 million to commercialise photonic components aimed at reducing networking bottlenecks inside large AI data centres.

AI infrastructure bottlenecks are spreading beyond GPUs. Polaris Electro-Optics has raised $50 million to bring a new optical-modulator technology into production as data centres struggle to move information fast enough between accelerators.
What happened
The Series B was led by Walden Catalyst Ventures. Polaris is developing FenGlass, an electro-optic material and device platform aimed at high-bandwidth optical communication.
Its first commercial products are targeted for 2027 and are intended for data-centre interconnects where conventional electrical links become increasingly power-hungry and difficult to scale.
Why it matters
Large AI clusters depend on thousands of accelerators working together. If the connections between them cannot move data quickly and efficiently, expensive compute sits idle.
Photonics offers a way to move more information using less power, making optical networking a strategic part of the AI hardware stack rather than a niche telecom component.
The bigger picture
The AI infrastructure boom is opening investment opportunities across an entire supply chain: memory, power systems, cooling, networking and photonics.
Polaris is part of a broader wave of companies trying to solve the less visible constraints around AI compute. Winning commercial adoption will require manufacturing reliability and integration with major networking platforms, but the market pressure for more efficient interconnects is only increasing.
