The Lightmatter team has significantly advanced optical computing toward a breakthrough. Their innovative application of their state-of-the-art machine to execute artificial intelligence (AI) models was a great accomplishment. This groundbreaking technology allowed the machine to play the popular arcade game Pac-Man. It made it possible for the machine to create written text and categorize movie reviews as positive or negative. With the ever-growing demands for AI, this new innovation has the potential to be a game-changer in computational efficiency and energy usage.
The Lightmatter machine orients hundreds of thousands of optical components alongside traditional electronic parts. Compact but powerful, the chip is roughly the size of a smart phone. The technology needs to be scaled up manifold. This clarification is important to make it more useful and applicable to AI companies more broadly.
Advancements in Optical Computing
Geoff Webb, a member of the Lightmatter team, emphasized the pressing energy demands of AI, stating, “AI just requires so much energy to do the amount of computing that’s required.” The energy costs associated with running AI are substantial, with Webb adding, “It costs billions of dollars for the energy needed for AI, and if you can reduce the cost of that, it’s very significant.”
The PACE machine, created by the Lightelligence team, represents ceramic and hardware cutting-edge developments in optical computing by achieving truly amazing computational speed reductions. With that, it was able to reach a new record low latency of only 5 nanoseconds, a reduction from 2,300ns. This unprecedented pace sets optical computing up to serve as a promising replacement to traditional architectures that have hit scalability barriers.
In particular, we discovered the system is able to run certain algorithms orders of magnitude faster than an equivalent traditional computing architecture,” said the Lightelligence crew. These kinds of capabilities imply that optical computing may fundamentally change the way many multifactor algorithms are computed.
The Future of AI and Optical Computing
Nick Harris, a key figure in the development of this technology, remarked on the current state of electronic computing: “We’re at an inflection point where traditional electronic computing … is hitting fundamental scaling limits — chips aren’t getting significantly faster or more energy-efficient each year, and costs are rising dramatically.” This observation points to the need for non-traditional factors computing approaches such as optical systems.
We asked Dr. Youssry about the specific advantages that photonic computers might have compared to their traditional electronic counterparts. He stated, “[Photonic computers] do have advantages over electronic computers in terms of energy efficiency, and possibly in terms of speed.” The implications are far-reaching, especially as AI gets more advanced and resource-intensive.
Dr. Youssry cautioned against the notion that optical computing will fully replace electronic systems. “I do not think they will replace electronic computers 100 per cent. They will be more like specialised hardware, or integrated with electronic supercomputers to perform some specialised task including AI.”
Implications for Energy Efficiency
With AI continuing to permeate every industry, energy-efficient computing solutions are key. Legacy systems just can’t hack it. The workloads AI, machine learning, and analytics require tremendous horsepower to process ever-growing datasets. Reducing the energy cost of an AI query would go a long way to alleviating most of these issues. As those professionals in the field tell us, those remarkable achievements don’t always result in dramatic cuts to consumers’ energy bills.
The Lightmatter team’s discoveries represent another step down the optical computing path to a new, powerful, faster, energy-efficient and cheaper alternative. They are hard at work on augmenting their technology. This development bodes well for emerging AI applications down the line which will need even lower power and faster processing speeds.
Leave a Reply