Pat Gelsinger, the former CEO of Intel, is pursuing a new approach to sustain the pace of computing progress. He aims to use light instead of electricity to power future chips.
Moore’s Law, the observation that transistor counts double roughly every two years, has slowed in recent years. Gelsinger believes optical interconnects can revive this trajectory.
Tiny beams of light could move data faster and with less energy than traditional copper wires. This advancement is critical for powering the next generation of artificial intelligence systems.
Current AI models demand enormous computational resources. Data transfer between chips has become a major bottleneck, limiting performance and efficiency.
Optical technology offers a solution by transmitting information at the speed of light. Gelsinger’s vision involves integrating photonics directly into silicon chips.
This shift would allow AI systems to process data more quickly while consuming less power. It could unlock new capabilities in machine learning and data analysis.
Industry experts have long explored optical computing, but practical challenges remain. Manufacturing and cost barriers have prevented widespread adoption.
Gelsinger’s plan builds on decades of research, aiming to make optical interconnects commercially viable. Success would mark a significant step toward extending Moore’s Law.
The initiative targets high-performance computing and data centers first. These environments stand to benefit the most from reduced latency and lower energy costs.
If realized, this technology could transform how AI systems are built and deployed. The era of light-based computing may be closer than many anticipate.
