#laserprocessingunit

Laser processor

An Advance in Optical Computing by LightSolver Allows Direct Simulation of Physical Events

Today, LightSolver, a leader in laser-based computing, announced that its Laser Processing Unit (LPU) can directly map and solve partial differential equations. This enhancement expands the LPU's capabilities beyond optimisation difficulties, making it a powerful platform for computational physics, engineering, and science.

LPUs were traditionally used for completely linked optimisation. New discovery employs 2D grid of connected lasers to solve complex PDEs like heat equation and Poisson equation. It could be expanded to wave dynamics and Schrödinger's equation. The Laser Processing Unit is versatile for industrial and research use due to this achievement.

Digital Computing Limits Overcome

Electricity and silicon processors power conventional computing, but optics' energy economy and parallel processing appeal as performance increases. Data transit bottlenecks and considerable heat generation might emerge from digital computers' need to divide activities into electrical operations. Traditional computers also need to digitise the analogue environment to replicate natural events, which increases run time and energy consumption.

Different technique is utilised by LightSolver. Natural events and light physics are used to compute it. Ruti Ben-Shlomi, LightSolver co-founder and CEO, said physical machines can solve large-scale physics simulations faster than HPCs or quantum systems. Due to their electromagnetic nature, lasers can mimic natural phenomena.

LightSolver relies on a programmable degenerate optical resonator. Instead of passing light through a system once, the Laser Processing Unit compels it to circulate in a closed loop for thousands of spins. Since all degrees of freedom are updated simultaneously, each round trip is a clock cycle that takes nanoseconds, regardless of problem size. This unique idea eliminates memory transfer bottlenecks by employing lasers as the computer and storage media.

New complex equation solution roadmap

Many simulations use partial differential equations, including fluid dynamics, heat transport, electromagnetism, climate prediction, and financial market behaviour. Digital electronics restrictions make large-scale equation solutions challenging. Even quantum computers with enormous speedups must deal with I/O bottlenecks, memory constraints, and the need for repeated statistical trials to provide reliable results.

LightSolver's Laser Processing Unit completes constant-time iteration steps in nanoseconds, regardless of issue size. A 100-fold speedup over GPU-based solvers is achievable. The problem is encoded once and the system doesn't need external memory or significant statistical sampling, so every computation is a real physical trial, giving rapid, scalable, and energy-efficient answers.

This invention speeds up LightSolver's product roadmap and introduces new modelling capabilities by scaling to 100,000 variables by 2027 and one million by 2029. The LPU features a Python frontend and compiler that instantly transforms PDEs into direct laser commands, making this cutting-edge technology accessible without laser physics knowledge.

Market Development and Access

Along with its technological advances, LightSolver has grown in the market. The company is partnering with Ansys and national labs and HPC centres to position the technology for real-world impact.

Within weeks, LightSolver will offer early access to its Laser Processing Unit Lab, which includes a digital emulator and Alpha hardware. This will provide scientists and businesspeople a preview of the future commercial-grade processor. Peer-reviewed “Solving Partial Differential Equations on an Analogue, Optical Platform” was presented at ACM Computing Frontiers 2025 by business researchers.