POET's new supply and joint development deal with Lumilens, a $50 million initial order for optical engines and potentially $500 million over five years, places it at the next AI bottleneck. As GPU clusters scale to hundreds of thousands of chips, moving data between them at the speed compute requires becomes the constraint. Copper doesn't get there and photonics does, and POET's optical engine platform sits exactly at that intersection. x.comÂ