Analog optical computer for AI inference and combinatorial optimization

Kirill P. Kalinin (), Jannes Gladrow, Jiaqi Chu, James H. Clegg, Daniel Cletheroe, Douglas J. Kelly, Babak Rahmani, Grace Brennan, Burcu Canakci, Fabian Falck, Michael Hansen, Jim Kleewein, Heiner Kremer, Greg O’Shea, Lucinda Pickup, Saravan Rajmohan, Ant Rowstron, Victor Ruhle, Lee Braine, Shrirang Khedekar, Natalia G. Berloff, Christos Gkantsidis, Francesca Parmigiani () and Hitesh Ballani ()
Additional contact information
Kirill P. Kalinin: Microsoft Research
Jannes Gladrow: Microsoft Research
Jiaqi Chu: Microsoft Research
James H. Clegg: Microsoft Research
Daniel Cletheroe: Microsoft Research
Douglas J. Kelly: Microsoft Research
Babak Rahmani: Microsoft Research
Grace Brennan: Microsoft Research
Burcu Canakci: Microsoft Research
Fabian Falck: Microsoft Research
Michael Hansen: Microsoft
Jim Kleewein: Microsoft
Heiner Kremer: Microsoft Research
Greg O’Shea: Microsoft Research
Lucinda Pickup: Microsoft Research
Saravan Rajmohan: Microsoft
Ant Rowstron: Microsoft Research
Victor Ruhle: Microsoft
Lee Braine: Barclays
Shrirang Khedekar: Barclays
Natalia G. Berloff: University of Cambridge
Christos Gkantsidis: Microsoft Research
Francesca Parmigiani: Microsoft Research
Hitesh Ballani: Microsoft Research

Nature, 2025, vol. 645, issue 8080, 354-361

Abstract: Abstract Artificial intelligence (AI) and combinatorial optimization drive applications across science and industry, but their increasing energy demands challenge the sustainability of digital computing. Most unconventional computing systems1–7 target either AI or optimization workloads and rely on frequent, energy-intensive digital conversions, limiting efficiency. These systems also face application-hardware mismatches, whether handling memory-bottlenecked neural models, mapping real-world optimization problems or contending with inherent analog noise. Here we introduce an analog optical computer (AOC) that combines analog electronics and three-dimensional optics to accelerate AI inference and combinatorial optimization in a single platform. This dual-domain capability is enabled by a rapid fixed-point search, which avoids digital conversions and enhances noise robustness. With this fixed-point abstraction, the AOC implements emerging compute-bound neural models with recursive reasoning potential and realizes an advanced gradient-descent approach for expressive optimization. We demonstrate the benefits of co-designing the hardware and abstraction, echoing the co-evolution of digital accelerators and deep learning models, through four case studies: image classification, nonlinear regression, medical image reconstruction and financial transaction settlement. Built with scalable, consumer-grade technologies, the AOC paves a promising path for faster and sustainable computing. Its native support for iterative, compute-intensive models offers a scalable analog platform for fostering future innovation in AI and optimization.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.nature.com/articles/s41586-025-09430-z Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:nat:nature:v:645:y:2025:i:8080:d:10.1038_s41586-025-09430-z

Ordering information: This journal article can be ordered from
https://www.nature.com/

DOI: 10.1038/s41586-025-09430-z

Access Statistics for this article

Nature is currently edited by Magdalena Skipper

More articles in Nature from Nature
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().