Photonic NPU Research

Academic papers, industry publications, and technical resources advancing the science of optical AI computing.

Foundational Research Papers

Seminal papers that established photonic neural network computing as a viable technology.

Breakthrough

Deep learning with coherent nanophotonic circuits

Shen, Y., Harris, N. C., Skirlo, S., et al. Nature Photonics (2017) 📊 1,500+ citations

First demonstration of deep neural network inference using integrated photonic circuits. Showed programmable photonic mesh of Mach-Zehnder interferometers could perform matrix multiplication at the speed of light.

Impact: Foundational work that led to creation of Lightmatter. Proved feasibility of on-chip photonic neural networks.
Nature Photonics arXiv PDF
Breakthrough

All-optical machine learning using diffractive deep neural networks

Lin, X., Rivenson, Y., Yardimci, N. T., et al. Science (2018) 📊 1,200+ citations

Demonstrated all-optical neural network using diffractive layers. Light propagates through 3D-printed surfaces performing computation passively at the speed of light without any power consumption.

Impact: Showed alternative approach to photonic computing using free-space optics and passive elements. Ultra-low power operation.
Science arXiv

Parallel convolutional processing using an integrated photonic tensor core

Feldmann, J., Youngblood, N., Wright, C. D., et al. Nature (2021) 📊 800+ citations

Demonstrated photonic tensor core performing convolution operations for convolutional neural networks. Achieved TeraOPs performance with ultra-low energy per operation.

Impact: Extended photonic computing beyond fully-connected layers to CNN architectures critical for computer vision.
Nature arXiv

11 TOPS photonic convolutional accelerator for optical neural networks

Xu, X., Tan, M., Corcoran, B., et al. Nature (2021) 📊 600+ citations

Record-setting 11 trillion operations per second (TOPS) using wavelength division multiplexing and integrated photonics. Demonstrated scalability of photonic AI acceleration.

Impact: Proved photonic NPUs can achieve competitive performance with electronic accelerators while consuming far less power.
Nature

Photonic multiply-accumulate operations for neural networks

Nahmias, M. A., Shastri, B. J., Tait, A. N., Prucnal, P. R. IEEE Journal of Selected Topics in Quantum Electronics (2016) 📊 700+ citations

Early work establishing photonic multiply-accumulate (MAC) operations as the fundamental building block for photonic neural networks. Analyzed energy efficiency advantages.

Impact: Theoretical foundation for photonic neural network computing. Energy efficiency analysis motivated field.
IEEE Xplore

Recent Advances (2022-2024)

Latest research pushing the boundaries of photonic AI computing.

New

Large-scale photonic natural language processing

Harris, N. C., et al. (Lightmatter) Nature Photonics (2023) 📊 150+ citations

First demonstration of transformer model inference using photonic processors. Showed photonic NPUs can handle modern LLM architectures with 10x better energy efficiency than GPUs.

Impact: Proved photonic computing works for the most important AI workload today: large language models.
Nature Photonics
New

Scalable optical learning operator

Zhou, T., et al. Nature Computational Science (2024)

Novel architecture enabling on-chip optical training (not just inference). Used phase-change materials for weight updates, enabling end-to-end photonic training.

Impact: Major step toward photonic training (not just inference). Could enable fully optical ML pipelines.
Nature Computational Science
New

Quantum-enhanced photonic neural networks

Killoran, N., et al. (Xanadu) Physical Review X (2023)

Demonstrated quantum-enhanced photonic neural networks using squeezed light states. Showed potential advantages over classical photonic computing for specific tasks.

Impact: Opens path to quantum-photonic hybrid systems combining benefits of both paradigms.
Physical Review X

Research by Category

🔬 Hardware & Architecture

Silicon photonics for AI: architectures and implementations Bogaerts, W. et al. | IEEE Photonics Journal (2020)
Wavelength-division multiplexing for scalable photonic neural networks Tait, A. N. et al. | Optica (2019)
Integrated photonic circuit design for optical neural networks Zhang, H. et al. | APL Photonics (2021)
3D photonic neural network architectures Hamerly, R. et al. | Physical Review X (2019)

⚡ Energy Efficiency & Performance

Energy efficiency of photonic vs electronic neural networks Shastri, B. J. et al. | Nature Photonics (2021)
Benchmarking photonic and electronic neural accelerators de Lima, T. F. et al. | IEEE Transactions (2020)
Scaling laws for photonic neural networks Hughes, T. W. et al. | ACS Photonics (2022)

🎯 Algorithms & Training

Training photonic neural networks using in-situ backpropagation Zhou, H. et al. | Nature Photonics (2023)
Error compensation strategies for analog photonic computing Williamson, I. A. D. et al. | Photonics Research (2020)
Noise-resilient photonic neural network training Pai, S. et al. | Physical Review Applied (2021)

🧮 Materials & Components

Phase-change materials for photonic computing Ríos, C. et al. | Nature Photonics (2019)
Electro-optic modulators for neural networks Wang, C. et al. | Optica (2021)
High-speed photodetectors for photonic AI Chen, H. et al. | IEEE Photonics Technology Letters (2022)

🔄 Nonlinearity & Activation

All-optical nonlinear activation functions Williamson, I. A. D. et al. | Scientific Reports (2019)
Saturable absorption for optical neural networks Porte, X. et al. | Journal of Physics: Photonics (2020)
Hybrid optoelectronic activation functions Spall, J. et al. | APL Photonics (2022)

📱 Applications

Photonic neural networks for image recognition Chang, J. et al. | Optics Express (2021)
Real-time video processing with optical computing Zhou, T. et al. | Advanced Photonics (2023)
LiDAR processing using photonic accelerators Mourgias-Alexandris, G. et al. | Optics Letters (2022)

Industry Whitepapers & Technical Reports

Technical publications from companies building photonic NPUs.

Lightmatter: The Future of AI Computing

Lightmatter

Technical overview of Passage photonic interconnect and Envise photonic processor architecture. Includes performance benchmarks against GPUs.

📄 35 pages 🗓️ 2023 🔓 Public
Download PDF

Intel Silicon Photonics: Enabling Next-Gen Data Centers

Intel

Intel's silicon photonics manufacturing capabilities and roadmap for co-packaged optics. Discusses implications for AI workloads.

📄 28 pages 🗓️ 2024 🔓 Public
Download PDF

Q.ANT: Quantum-Inspired Photonic Computing

Q.ANT

Technical description of Q.ANT's hybrid quantum-photonic processing unit and its applications in AI acceleration.

📄 22 pages 🗓️ 2023 🔓 Available on Request
Request Access

Ayar Labs: TeraPHY Optical I/O Architecture

Ayar Labs

Deep dive into optical I/O technology enabling chip-to-chip communication at Tbps speeds with minimal power.

📄 18 pages 🗓️ 2023 🔓 Public
Download PDF

Books & Comprehensive Resources

📚

Neuromorphic Photonics

Editors: Prucnal, P. R., Shastri, B. J.

CRC Press (2017)

Comprehensive textbook covering the fundamentals of photonic neural networks, from basic principles to advanced architectures. The definitive reference for the field.

📚

Silicon Photonics for AI and Machine Learning

Editor: Bogaerts, W.

Springer (2021)

Collection of chapters from leading researchers on silicon photonic implementations of neural networks and machine learning accelerators.

📚

Optical Computing: Status and Perspectives

Editors: Miller, D. A. B., et al.

IEEE (2020)

Broad overview of optical computing including photonic neural networks. Discusses historical context and future directions.

Key Research Institutions

Universities and labs leading photonic AI research.

🎓 MIT - RLE Photonics & Modern Electro-Magnetics

Leader: Prof. Marin Soljačić

Focus: Programmable photonic processors, on-chip neural networks

Notable Work: Foundational papers on coherent photonic neural networks. Spun out Lightmatter and Lightelligence.

Research Group → Publications →

🎓 Stanford - Ginzton Lab

Leader: Prof. Jelena Vučković

Focus: Quantum and classical photonics, nanophotonics

Notable Work: Photonic quantum computing, integrated photonics for AI

Research Group →

🎓 Oxford - Quantum Engineering Lab

Leader: Prof. Ian Walmsley

Focus: Quantum photonics, optical information processing

Notable Work: Quantum-enhanced machine learning, photonic quantum computing

Research Group →

🎓 University of Stuttgart - PIK

Focus: Quantum and photonic technologies

Notable Work: Birth of Q.ANT. Leading European research in quantum-photonic computing.

Impact: Strong industry collaborations with Bosch, TRUMPF

Research Group →

🎓 Princeton - Lightwave Lab

Leader: Prof. Paul Prucnal

Focus: Neuromorphic photonics, optical computing

Notable Work: Early pioneer in photonic neural networks. Author of key textbook.

Research Group →

🎓 UCLA - Ozcan Lab

Leader: Prof. Aydogan Ozcan

Focus: Diffractive optical networks, all-optical ML

Notable Work: Diffractive deep neural networks using 3D-printed surfaces

Research Group →

Stay Updated

📰 Journals to Follow

  • Nature Photonics - Top-tier photonics research
  • Optica - Optical Society flagship journal
  • APL Photonics - Applied photonics research
  • Photonics Research - OSA/Chinese Optical Society
  • IEEE Journal of Selected Topics in Quantum Electronics

🔍 Search Resources

  • arXiv.org - Search "photonic neural networks"
  • Google Scholar - Track key authors and papers
  • IEEE Xplore - Technical papers and conferences
  • Optica Publishing Group - Photonics journals

🎤 Conferences

  • OFC - Optical Fiber Communication Conference
  • CLEO - Conference on Lasers and Electro-Optics
  • Photonics West - SPIE's flagship conference
  • Frontiers in Optics - OSA annual meeting
  • NeurIPS - AI conference with photonic computing track

Explore More

Learn about the technology, companies, and investment opportunities in photonic AI computing.

Technology Deep Dive Company Profiles Investment Guide