- LightSolver has launched a Laser Processing Unit that uses lasers instead of silicon chips to solve equations.
- The system avoids processor-memory bottlenecks by using built-in optical memory.
- It can map physics equations directly onto the laser system for fast and energy-efficient simulations.
- The company plans to scale its technology to handle up to one million variables by 2029.
A young startup from Tel Aviv has revealed a breakthrough technology that might reshape the future of high-performance computing. LightSolver has introduced what it calls a Laser Processing Unit, or LPU.
Unlike traditional CPUs, GPUs, or even emerging quantum computers, this new approach uses lasers as the foundation for solving complex problems. It is not designed to run Windows or traditional software environments, but instead focuses on solving physics equations directly at lightning speed.
The LPU is built to tackle one of the most important challenges in science and engineering: partial differential equations. These equations are the backbone of physics and are used to describe processes like heat transfer, fluid flow, and even the fundamental mechanics of quantum systems.
LightSolver claims that its technology can map such equations directly onto its laser system, skipping many of the delays and inefficiencies that plague digital computers.
Why Lasers Instead of Silicon?
Today’s computers, whether CPUs or GPUs, rely on digital logic and the constant transfer of data between processor and memory. This back-and-forth shuttling of information creates bottlenecks that slow performance and consume vast amounts of energy.
LightSolver’s LPU avoids this issue by using optical memory that is built directly into the laser system itself. In practice, this means that information stays inside the laser resonator and does not need to move externally for every calculation step.
This design allows the machine to operate with constant-time iterations, measured in nanoseconds. The speed of each step does not depend on the size of the problem, which is a dramatic departure from the way digital systems work today.
According to LightSolver, the result is a platform that can handle massive physics simulations in a fraction of the time required by high-performance clusters or quantum machines.
The Difference from Photonic Chips
Some might confuse the LPU with photonic processors, which also rely on light. However, LightSolver points out that its approach is very different. Current photonic processors are built on flat, two-dimensional chips. LightSolver’s design, on the other hand, is three dimensional. This distinction, the company argues, will allow its technology to scale much further.
By arranging lasers in a three-dimensional grid that can work together in unison, the LPU creates a computing environment that goes beyond the current boundaries of chip design. This gives it the potential to solve larger and more complex problems as the system evolves.
Roadmap and Early Access
LightSolver has already laid out an ambitious roadmap. By 2027, the company aims for its LPU to handle 100,000 variables. By 2029, it plans to scale this up to one million variables. These milestones are intended to make the system powerful enough to address the toughest computational challenges faced by science and industry.
For now, the company is opening access to early users. Through what it calls its LPU Lab, LightSolver is making both a prototype processor and a digital emulator available to researchers. This move allows scientists and engineers to begin testing the technology and exploring how it might be applied to their own fields.
A Vision for Physics-Based Computing
Ruti Ben-Shlomi, CEO and co-founder of LightSolver, has explained the philosophy behind the technology. In her words, traditional computers digitize the analog world of nature, and that conversion comes with a cost.
It leads to longer run times and significant energy waste. By running simulations directly on a physical machine that operates according to the principles of light, she argues, LightSolver can deliver faster and more energy-efficient solutions than anything available in high-performance or quantum computing today.
The company’s research has already been shared in academic papers and presented at conferences, including the ACM Computing Frontiers 2025 event. LightSolver is also working with simulation software developers, high-performance computing centers, and national laboratories to expand adoption of its approach.
What makes this effort particularly exciting is not only the promise of raw performance but also the possibility of reducing energy consumption in computing. As the demand for simulations in areas like climate modeling, materials science, and quantum research continues to grow, so does the need for machines that can deliver results faster and more sustainably.
Follow TechBSB For More Updates
