Photonic interconnect race targets the next AI data-center scaling bottleneck

The article describes an industry shift toward optical links for AI data centers as traditional copper connections become a tougher limit on scale. It cites Lightmatter chief executive Nick Harris in the context of competing approaches and emerging standards. This is adjacent to, rather than directly inside, the memory market, but it matters because AI clusters increasingly depend on balanced movement of data between compute, networking and memory tiers. Better optical interconnects can raise the practical ceiling for larger systems, which may in turn support demand for high-bandwidth memory, server DRAM and CXL-style expansion. The pricing effect is indirect, but the infrastructure signal is meaningful for the same AI buildout driving premium memory demand.