Optical Computing: The 'Second Law of Thermodynamics' Moment for AI Infrastructure

If you compare the evolution of AI compute power to a river, Moore's Law was the channel we trusted—doubling every 18 months, wide and deep. But now, the channel is narrowing, and physical limits are like rocks on the riverbed, exposed by increasingly shallow water. Shen Yichen of Lightelligence proposed at WAIC 2026 that "light will drive the transformation of AI infrastructure," essentially saying: we need a new river, and light is that underground dark stream yet to be fully tapped.