Hybrid quasiparticles cut AI energy costs at Penn
Researchers at the University of Pennsylvania built polaritons, hybrid light-matter particles, inside a specially engineered microcavity. They paired these polaritons with existing silicon photonic circuits to perform matrix multiplications at 10 times lower power than current GPUs. Tests showed inference speeds rose by roughly 40 percent while heat output dropped.
You learn that computation can move from electrons to photons. This shifts your workflow from buying bigger GPUs toward designing or renting photonic accelerators for repeated inference tasks.
The University of Pennsylvania Photonics Laboratory has already taped out a 64 by 64 polariton array chip and reported 2.3 pJ per operation on MNIST classification.
Step 1: Visit the Penn Photonics Lab site at photonics.seas.upenn.edu and download their open-source polariton simulation notebook. Step 2: Run the notebook on Google Colab to model a 16 by 16 matrix multiply using the provided polariton equations. Step 3: Compare the simulated energy per operation against a standard PyTorch GPU baseline to see the reported 10x reduction.