Microsoft unveils light based computer using vintage tech that could make AI 100x faster

A computer that runs on beams of light

Imagine a computer that does not rely on electricity flowing through silicon chips but instead harnesses beams of light. Microsoft has unveiled a new prototype that could change how machines handle complex problems.

The system is called an analog optical computer, or AOC, and it embodies a completely new way of processing. By shifting from traditional transistors to light, researchers are exploring a path toward faster and more efficient computing than anything available today.

Old technology inspires a new vision

While the AOC may sound futuristic, the concept of using light for computing actually dates back decades. Microsoft researchers revisited this approach with modern tools to make it practical for today’s challenges.

By combining micro LED lights, optical lenses, and smartphone camera sensors, the team has created a working model using commercially available parts. That choice keeps costs down and opens the door for scaling up through existing manufacturing supply chains.

Aiming for one hundred times efficiency

One of the most striking claims about this new machine is its energy performance. Researchers estimate the AOC could deliver around a hundred times more efficiency compared with current hardware.

That improvement is possible because the system avoids many energy‑wasting conversions between digital and analog domains for the targeted workloads.

Instead, it uses light directly to perform the calculations, cutting losses and bypassing bottlenecks that limit the speed of traditional processors.

Solving problems with steady states

The way the AOC works is different from the usual on and off switches of digital chips. It sends light through sensors at different intensities, allowing it to add and multiply numbers in a loop.

This process continues until the system finds a stable solution, something researchers call a steady state. That unique ability makes the AOC well-suited for handling optimization challenges, where the best possible outcome has to be chosen from countless options.

From a small test to bigger ambitions

The current prototype can only manage a limited number of parameters, about 256 weights, which restricts the size of the problems it can handle. Even so, the progress compared with earlier versions has been significant.

Researchers predict that, with further miniaturization and additional micro‑LED modules, future designs could scale to millions or even billions of variables. With that kind of scale, the machine could potentially outperform today’s fastest processors in specialized tasks.

A digital twin extends the power

Alongside the physical device, Microsoft also built a computer model that mimics the AOC’s behavior. This digital twin allows researchers to test larger and more complex problems than the prototype can currently manage.

It acts as a bridge between today’s small hardware and tomorrow’s scaled-up machines. By experimenting in the digital environment, scientists can refine algorithms, predict performance, and share results with outside collaborators who want

A boost for healthcare imaging

Medical scans often take a long time to complete, sometimes requiring patients to remain still for half an hour or more. Researchers used the digital twin of the AOC to simulate the reconstruction of the same image with reduced data.

The experiment successfully recreated a 320×320 pixel brain scan using just over 60% of the original information. This finding hints at a future where MRI scans might be shortened dramatically, improving access to diagnostics and reducing waiting times.

Tackling financial complexity

Microsoft worked with Barclays Bank to test how the AOC could help in the world of finance. Clearinghouses that handle thousands of trades daily face enormous optimization challenges when settling transactions between banks.

Using the prototype and its digital twin, researchers demonstrated how the system could improve delivery versus payment securities exchanges. Even at a reduced scale, the experiment showed that the optical computer could make financial processes more efficient while minimizing risks.

Practical parts make it possible

Unlike exotic quantum machines that require extreme conditions, the AOC is built with parts you might find in everyday devices. Micro LEDs, projectors, optical lenses, and smartphone camera sensors form its core.

Because these components are widely available, the prototype operates at room temperature without the need for specialized environments. This practicality makes it easier to imagine future versions being manufactured and adopted at larger scales.

A team with diverse expertise

Building the AOC required input from multiple fields, from mathematics and physics to biomedical signal processing. The Microsoft team in Cambridge, UK, brought together specialists with different backgrounds to make the idea a reality.

Researchers like Francesca Parmigiani led the project, while others, such as Jannes Gladrow, contributed expertise in artificial intelligence. Their combined efforts helped the machine not only solve optimization problems but also explore new frontiers like medical imaging and machine learning.

Early steps into machine learning

Although not designed as a general-purpose computer, the AOC has already shown it can handle certain machine learning tasks. Researchers mapped algorithms to the system to test how it would perform with basic classification problems.

The results were encouraging, with the optical machine performing about as well as a digital computer on simple tests. This early success suggests that larger versions could one day process more demanding workloads with far greater energy efficiency.

Room temperature durability

One of the practical strengths of the AOC is that it runs at normal room temperature. Unlike some advanced computing systems that need extreme cooling or controlled environments, this machine avoids such complications.

That means future models could operate in everyday settings without massive infrastructure costs. The design not only saves energy during processing but also reduces the need for additional power-hungry cooling systems, adding another layer of sustainability.

Opening doors for collaboration

Microsoft has publicly shared its optimization solver algorithm and the digital twin to encourage other researchers to join the effort. By doing this, the company hopes to speed up progress and explore new applications.

Collaboration is seen as crucial because the potential uses for analog optical computing stretch far beyond what one team can investigate. From logistics to science, other groups can now test ideas and propose new problems for the system to tackle.

A future role in artificial intelligence

Large language models and other AI systems consume huge amounts of power today. The AOC’s design could handle reasoning tasks like state tracking in a much more energy-efficient way.

State tracking involves keeping track of complex rules and decisions over time, similar to playing a long game of chess. If scaled up, the optical approach could allow AI models to run at lower cost while still managing intricate reasoning challenges.

Scaling up every two years

The research team envisions releasing new generations of the AOC roughly every two years. By gradually doubling the number of parameters and improving component designs, they plan to expand the system’s computational capacity over time.

With each step, the technology moves closer to handling real-world problems at full scale. The long-term ambition is for the AOC to become a significant part of computing’s future, transforming industries in ways not possible with digital processors alone.

Suppose you’re looking for a change that could save users both time and frustration. Check out how Microsoft brings AI-driven file search to Windows 11 through Copilot.

A sustainable vision for tomorrow

The drive behind this project is not just about speed but about sustainability. As AI and computing demand more energy worldwide, new solutions are urgently needed.

By harnessing light, the AOC could help cut power use dramatically while solving problems in finance, healthcare, and beyond. Microsoft researchers believe this breakthrough is only the beginning of a shift toward a cleaner, more efficient computing era.

Discover how Cloudflare and Microsoft joined forces to reshape search for an AI-driven web. Stay ahead of the curve and explore the future of online searching today

Do you think this new approach could change the way we use technology forever? Or do you see it as just another passing trend? Share your thoughts in the comments and let us know what excites you most about this breakthrough.

Read More From This Brand:

• Elon Musk plans a Microsoft built entirely by AI
• Microsoft is reportedly cutting 10 percent of its workforce in France
• Microsoft engineer’s death at tech campus reignites burnout discussions

Don’t forget to follow us for more exclusive content right here on MSN.

If you like this story, you’ll LOVE our Free email newsletter. Join today and be the first to receive stories like these.

This slideshow was made with AI assistance and human editing.