POET Technologies Inc. has signed a multi-year supply and development agreement with Lumilens Inc. to provide AI-focused photonic components [1].

The deal signals a significant expansion for the NASDAQ-listed company as it targets the growing demand for optical networks in hyperscale data centers [2]. This move positions POET to integrate its wafer-level photonic technology directly into the infrastructure powering next-generation artificial intelligence [3].

The five-year supply agreement is valued at $500 million [3]. As part of the initial rollout, the companies have established a first purchase order totaling $50 million [4]. This partnership focuses on advancing photonic integration, which allows data to move more efficiently through AI hardware [3].

Alongside the partnership announcement, POET Technologies reported its financial results for the first quarter of 2026 [5]. The company said its revenue for the period beat previous expectations [1]. This financial growth coincides with the company's effort to expand its footprint within the hyperscale market [1].

Market reaction to the news was immediate. Reports on the stock's performance varied, with some sources stating shares climbed 16% [4], while others reported a rally of 24% [3]. The discrepancy reflects the volatility of the stock following the announcement of the Lumilens deal [3, 4].

The collaboration involves the development of photonic components that aim to reduce power consumption and increase bandwidth in data centers [3]. By utilizing wafer-level integration, POET and Lumilens intend to streamline the production of optical interconnects used in AI clusters [3].

“The five-year supply agreement is valued at $500 million.”

The agreement reflects a broader industry shift toward optical interconnects to solve the 'bandwidth bottleneck' in AI computing. By securing a $500 million commitment, POET Technologies moves from a developmental phase to a large-scale supply role, validating its photonic integration technology as a viable component for hyperscale data center architecture.