Polaris Electro-Optics discloses $49.5M financing for AI data-center optics

Why it matters

Polaris has disclosed enough capital to fund the difficult move from prototype modulators toward wafer-scale output. Its progress will test whether ferroelectric nematic materials can ease AI data centers' bandwidth and power constraints while fitting existing silicon-photonics manufacturing.

Cory Pecinovsky's Polaris Electro-Optics, a developer of optical modulators for data centers, disclosed a much larger financing. Polaris sold $49.5 million of a planned $50 million offering to 10 investors, according to a Form D filed with the SEC on September 4th.

The first sale occurred on August 21st, making this a two-week-old financing rather than a retrospective disclosure. The filing leaves $500,000 available. The filing does not name the investors or disclose a valuation. It also does not establish whether the offering is a Series B. The Form D reports equity and rights to acquire another security, but does not explicitly identify the securities as a Simple Agreement for Future Equity.

The financing is substantially larger than Polaris's previous rounds. The CU Boulder spinout previously disclosed a $3 million seed round in 2023 and a $10 million Series A in 2025. Adding the new securities sold brings Polaris's publicly announced and filed gross financing to about $62.5 million. The $49.5 million alone is nearly four times the capital disclosed across those two earlier rounds.

The SEC filing lists Polaris's principal place of business in Carlsbad, California. The capital arrives as Pecinovsky tries to move a new optical material from small prototype runs into components that can be manufactured by the million. His original bet was that ferroelectric nematic liquid crystals could leave a university lab, fit into established silicon-photonics production and carry growing volumes of traffic inside AI data centers.

Pecinovsky left stock options for a new material

Pecinovsky earned a doctorate in chemistry at the University of Colorado Boulder and spent his industrial career working on liquid crystals and electro-optical materials. CU Boulder physicists Noel Clark, Joe MacLennan and Matt Glaser were among the researchers behind the ferroelectric nematic phase, a class of liquid crystals with properties suited to converting electrical signals into optical ones.

The inventors knew Pecinovsky and approached him about building a company around the material, according to UC San Diego's account of Polaris's origins. Pecinovsky was new to entrepreneurship and felt compelled to commercialize the research. He left his job and stock options to build Polaris with the researchers. CU Boulder identifies Clark, MacLennan, Glaser, Pecinovsky and earlier CEO Jason Sickler as co-founders or founding leaders.

Polaris calls its material platform Polaris Glass, or ferroelectric nematic glass. An optical modulator encodes electrical data onto light, allowing information to travel across optical links. Polaris says its material can deliver sub-volt operation and compact devices while being added to silicon-photonics wafers after the foundry process. Those are company performance and manufacturing claims. Lower voltage and smaller components could reduce the power and space consumed by interconnects surrounding large GPU clusters if the devices retain those characteristics in volume production.

Polaris reported a 102 Gbps PAM-4 demonstration in September 2024. Polaris presents 400 Gbps per lane and beyond as a target. Its public material establishes a technical demonstration and prototype evaluation rather than commercial production at that speed.

The distance between those stages is substantial. In July 2025, Pecinovsky described Polaris as producing devices in "onesies and twosies" while needing wafer-scale output by the million. Polaris was also shipping samples to large cloud infrastructure operators for evaluation and purchasing an optical wafer prober for installation at UC San Diego's Qualcomm Institute, where it had used shared cleanroom and photonics-testing equipment. Those activities document customer evaluation and manufacturing preparation, rather than verified commercial shipments or production volume.

A new CEO for the production phase

Polaris appointed Morten Nissov as CEO alongside its Series A announcement in August 2025. Nissov previously held executive roles at Lumentum and Oclaro and was vice president of sales and business development at Genuine Optics. His arrival shifted Pecinovsky into the founder and CTO role while putting an executive with more than 25 years in semiconductors and photonics in charge of commercialization.

Koch Disruptive Technologies led the $10 million Series A. Rhapsody Venture Partners, M Ventures and Buff Gold Ventures returned from earlier financings, while Blue Sky Capital joined the round. Koch Disruptive Technologies Managing Director Isaac Sigron sits on the Polaris board and is listed as a director in the new filing.

The September filing does not establish whether those investors participated again. It lists 10 buyers without identifying them, leaving the ownership terms and lead investor outside the public record.

Polaris previously said its Series A would support engineering and operations hiring, engagements with hyperscalers and system partners, and the move toward volume production. The new financing gives Nissov and Pecinovsky considerably more room to pursue that plan, although customer qualification and manufacturing yield will determine whether the material reaches production systems.

AI's networking bottleneck gives Polaris its market

GPU clusters depend on links between processors, making networking a growing constraint on performance and power. As data rates rise, optical connections replace copper across progressively shorter distances. Modulators sit at the electrical-to-optical boundary, turning them into a contested part of the AI infrastructure stack.

Polaris is pursuing a materials-led answer. Lightwave Logic develops electro-optic polymers, while NLM Photonics works on silicon-organic-hybrid modulators. Polariton uses plasmonics, and OpenLight combines indium-phosphide components with silicon photonics. Polaris's pitch is that ferroelectric nematic glass can improve speed, voltage and footprint without forcing manufacturers to abandon standard silicon-photonics processes.

Pecinovsky has already turned a scientific discovery into a working device that outside investors were willing to finance. The $49.5 million filing raises the standard of proof. Polaris must now show that its material can survive qualification, move from prototype samples to repeatable wafer-scale output and earn a place in systems where component failures carry data-center-sized consequences.