Skip to navigationSkip to main contentSkip to right columnADVERTISEMENTMicah Zimmerman, The Motley FoolSun, July 19, 2026 at 10:57 PM GMT+2 4 min readMuch of the AI conversation centers on chips, but there is a quieter bottleneck forming inside data centers, and it is all about how those chips talk to one another. As artificial intelligence clusters swell to tens and even hundreds of thousands of processors, the humble copper wiring that has connected computers for decades is running out of room.The fix is silicon photonics, and investment in it is ramping quickly. Two companies look especially well placed to benefit.Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »Image source: Getty Images.Why copper is hitting a wall in AI clustersTo train a modern AI model, thousands of chips must act like one enormous brain, constantly shuttling vast amounts of data among them. The more chips you add, the more traffic flows across the wires linking them, and that is where copper starts to break down. At the blistering signaling speeds these systems now demand, a passive copper cable can only carry a clean signal for less than a meter before it degrades. Push it further, and you have to pump in more and more electrical power just to keep the data intact, which generates heat and drives up the energy bill.In a small server, that was never a problem. In a warehouse-sized cluster stretching across rows of racks, it becomes a hard physical ceiling that engineers now call the copper wall. When your bottleneck is measured in centimeters and watts, you cannot simply add more copper and hope for the best.How silicon photonics breaks throughSilicon photonics solves the problem by sending information as pulses of light through fiber instead of electrons through metal. Light travels farther, carries far more data, and uses less power over distance, which is exactly what a giant AI cluster needs. The cutting edge of this shift is co-packaged optics, where the optical components are built right next to the switch chip rather than plugged in at the edge of the box. That tight integration slashes the power lost in translation and packs far more bandwidth into the same space.Copper is not disappearing, but its job is shrinking to the shortest hops inside a package, while optics take over everything from board to rack scale. The money following this transition is real. The optical interconnect market for AI data centers is expected to grow several times over this decade, and the broader optical transceiver market is projected to jump about 60% in a single year to roughly $26 billion in 2026.Terms and Privacy PolicyEU DSA contactPrivacy & Cookie SettingsMore Info