AI requires new hardware. On-chip integrated optics can provide orders of magnitude performance improvement. Salience Labs has a world-class solution backed by a decade of research from top research groups to address AI inferencing: a market expected to exceed $50B by 2025. Salience is developing an ultra-high throughput, low latency Photonic Tensor Processing Unit. The work has its origins in labs at the University of Oxford and Münster University.
Salience Labs uses a proprietary amplitude-based approach to photonics, resulting in small, dense photonic chips clocking at 10’s of GHz. This, combined with massively parallel performance (up to 64 beams parallel operations can be packed into a single beam of light), results in a high-density processing unit.