Key Takeaways
POET Technologies Inc. (NASDAQ: POET) and Lumilens Inc. have entered a strategic supply agreement to commercialize a new class of wafer-level photonics for AI infrastructure, beginning with a $50 million purchase order for POET’s optical engines.
“Our focus has always been on redefining the integration paradigm in photonics,” said Dr. Suresh Venkatesan, Chairman & CEO of POET Technologies. “This new EOI platform will allow us to jointly bring semiconductor-style manufacturing discipline to optical engines — delivering precision, scalability, and cost structure advantages that are essential for AI infrastructure at scale.”
The agreement establishes a commercial framework for a joint development program, with the initial $50 million order from Lumilens representing the first phase of a supplier relationship that could exceed $500 million in total purchases over five years. To align the partnership, POET has granted Lumilens a nine-year warrant to purchase up to 22,921,408 common shares at an exercise price of $8.25 per share. Approximately 10 percent of the shares are immediately exercisable, with the remainder vesting as Lumilens meets payment milestones toward the $500 million target.
This partnership aims to solve a key bottleneck in scaling AI: the optical layer. The joint platform centers on a new Electrical-Optical Interposer (EOI) that combines wafer-level optical engine production with Lumilens’ next-generation chipsets. This approach replaces costly and slow active-alignment manufacturing with a capital-efficient, high-volume process, similar to the wafer-level integration that enabled modern GPUs and HBM.
“GPU interconnects are emerging as the defining bottleneck for scaling AI, and addressing it requires rethinking the full optical stack — silicon, photonics, and packaging — together,” said Ankur Singla, CEO and Founder of Lumilens.
Together, the companies will deploy the platform across a roadmap including 800G and 1.6T pluggable transceivers and future near-package and co-packaged optics. Engineering samples are expected in late 2026, with a production ramp scheduled to align with hyperscaler customer deployments in 2027.
This article is for informational purposes only and does not constitute investment advice.
