A new chip-connection system could help overcome the “memory wall,” a major bottleneck that slows computing speed and restricts AI model growth. Instead of relying on traditional electrical wires, this technology transfers data through reconfigurable light pathways, allowing for much faster and more efficient communication.
The project, led by the University of Michigan, is backed by a $2 million grant from the National Science Foundation’s Future of Semiconductors program. It brings together researchers from the University of Washington, the University of Pennsylvania, and Lawrence Berkeley National Laboratory, with guidance from industry leaders including Google, Hewlett Packard Enterprise, Microsoft, and Nvidia.
To read more, click here.