Junwei Cheng

Wuhan National Laboratory for Optoelectronics

Papers

Total Citations

H-Index

About

Junwei Cheng is a rising leader in the field of optical computing, where his work targets the fundamental energy and latency bottlenecks of electronic processors. His primary research focus is the development of photonic integrated circuits for neuromorphic and recurrent computing, aiming to realize high-speed, low-power hardware accelerators. His most notable contribution, the "Monolithically integrated asynchronous optical recurrent accelerator" (2025), demonstrates a fully on-chip optical system that shifts computational bottlenecks from the electrical to the optical domain, achieving unprecedented efficiency. Though early in its citation impact (17 citations), this work represents a significant step toward practical, scalable optical neural networks. Cheng’s research is distinguished by its emphasis on monolithic integration—a critical engineering challenge for real-world deployment. By tackling both the architectural design and the physical integration of photonic components, he is helping to pave the way for next-generation computing systems that could dramatically outperform traditional electronics in speed and energy efficiency. His work positions him as a key innovator in the race to build the first practical optical accelerators for AI.