Photonic computing has been converting skeptics for decades

The photonic computing industry is an ecosystem where dreams of future technology deployment come true.

The dreams of smart cities, smart autonomous mobility, smart personalized medicine and beyond are truly limitless. Moreover, despite the huge data processing involved, everything can be done in real time, with predictive and self-improving cybersecurity capabilities, all in a sustainable way and at scale. But many skeptics in the photonics industry have already been let down, whether by overstated claims, immature technologies, or a host of other issues that have kept bold ideas from reaching market.

With recent discoveries and new technologies emerging, this sector is emerging, bringing promise to companies looking to scale their real-time computing capabilities and reduce the costs associated with artificial intelligence in data centers and on the outskirts.

What are the promises of modern photonic computing?

Photonic computing is a tremendous high-speed technology that promises to replace traditional electronic computing components. The benefits are achieved by re-engineering today’s standard copper wires and silicon-based chips, which were designed to carry electrons. Instead, silicon-based chips, glass-based chips, and fiber-based devices can process information carried by light at a higher capacity.

With photonic computing, wires can be replaced either by fiber optic cables that can process signals before they reach their destination, or by special light-carrying waveguides in photonic chips. For denser functionality, glass and silicon-based ICs do even more work.

Today’s cutting-edge technology is speed-limited due to the huge computational load inherent in AI models, which keep growing. However, having a super-fast chip in a hybrid system where some of the computing is done in off-the-shelf electronics will, by definition, create information bottlenecks. These bottlenecks, such as PCI buses and memory bandwidth limitations, significantly reduce the overall performance benefits of the Photonics Processor.

Yet, it cannot be ignored that silicon chips reduce the space, time and energy needed to perform calculations. So far, the benefits of photonic chips, including reduced wear and overheating impacts, have to be balanced with scaling and yield challenges due to the nature of photons, with their much larger wavelength and therefore much larger component size (up to 10,000×).

Thanks to groundbreaking developments, the atmosphere of photonic computing today is similar to the mid-century transition from vacuum tubes to transistors. Much like what we saw in the mid-1950s, photonic computing is in its infancy, but already making waves. Realizing the unlocking potential of future technologies, experts are driven to reach the magic quadrant of efficient architecture and technology. Although there are some difficult obstacles to overcome along the way, there is no doubt that a world of IT efficiency and scalability awaits you on the horizon.

Why is it different this time

In 1975, light as an information carrier confirmed its clear supremacy over electrical wires when optical fibers replaced copper submarine communication cables and made it possible to speed up the flow of information via the global internet.

Since then, a series of remarkable engineering feats by companies such as Cisco, Juniper, NTT, Mellanox and others have enabled the bandwidth supported by optical communication to continually break its own records every year. reaching the staggering figure of 200 Gbps. If that weren’t enough to convince skeptics, speeds of 400 Gbps and 800 Gbps over a single fiber are on the horizon, including an experiment by NTT that demonstrated transmission of 1 petabit per second (1,000,000 Gbps ) on a single fiber.

The basic physics of controlling light interactions and its much longer wavelength, which dictates 1000× to 10,000× larger elements, raises concerns about mass production cost and yield. Nevertheless, attempts to invent, design, and build switches, routers, and other high-transmission products based on optical computing have been on the roadmap for decades for photonics companies and startups.

Breaking into the modern age, AI applications present a unique opportunity by leveraging photonic computing, as their outputs are measured in relative probabilities rather than bit-exact calculations. It doesn’t matter if an AI system deviates 5% in its output as long as it consistently gets the correct answer to the question — is that a face in the picture? A car? A dog? The correct answer is the goal, not the exact probability that led the system to give the answer.

This understanding, coupled with a strong motivation to solve the painful limitations of the current AI infrastructure, propelled the fledgling photonics computing industry onto an accelerated path with big investment, big promise and, hopefully, , great progress. Today, several technologies are vying for the Holy Grail: silicon photonics, photonic integrated circuits, glass-based photonic chips and fiber-based photonic devices.

Unlocking Future Capabilities with Photonic Computing

With incredibly low expected power consumption, efficiency exceeding 100×, and speeds that force us to reimagine what’s possible, it’s clear that the future is in pure photonics. Industry should aspire to a solution where data enters a system at the high speeds of optical communication and all calculations are done on the fly without any conversion to electronics or any memory read and write operations. dynamic. What can we achieve with such a non-von Neumann architecture? Imagine a cable of 100 fibers each carrying 800 Gbps, entering a photonic pod, broadcasting photons through a host of processors, and exiting as desired outputs at exactly the same input rate – 800 billion full calculations per second with a expected energy cost from 10 to 20 kW.

The current trend of reducing the carbon footprint is colliding with the growth of AI models. Pure photonics seems to be the only sustainable large-scale solution to reconcile these trends.

Read also:


Sherry J. Basler