Could optical computing solve AI’s power demands?

Primarily, optical processors are aimed at AI acceleration and inference tasks, as well as specialised high-end applications. Meanwhile, optical quantum computers, which leverage qubits and other quantum effects, are targeting use cases such as simulation, optimisation, and AI/ML. Photonic-based quantum computing is projected to grow significantly by 2030, with companies like Quandela, QUIX, and Pasqal leading the charge. The market for this technology is expected to be worth hundreds of millions of USD by 2034.

“The first optical processor shipments will appear around 2027/28, initially for custom systems implementing parts of the technology, with revenue largely coming from NRE services. By 2028, general-purpose optical processors would enter the market, with early adopters and OEMs gradually integrating them from 2029 onward. By 2034, at Yole Group, we expect optical processor shipments to reach nearly 1 million units, representing a multi-billion-dollar market,” said Eric Mounier, Ph.D., Chief Analyst, Photonics & Sensing at Yole Group.

In this context, Yole Group has launched its new report, Optical Computing 2024, offering an in-depth analysis of the current state and future potential of optical computing. The report covers both analog/digital optical systems and quantum optics approaches, examining how these technologies can meet the rising computational demands of AI, ML, and other applications that are limited by traditional electronic systems. This new report is part of Yole Group's broader photonic analysis collection, which includes Silicon Photonics 2023 (with a 2024 edition forthcoming) and Co-Packaged Optics for Datacentres.

While optical computing itself is not a new concept, the implementation of optical logic gates through photonic ICs and quantum optics has shown the most promise. However, practical large-scale optical logic gates still face significant hurdles. To compete with electronic gates, optical gates must achieve scalability, cascadability, and efficient recovery from optical losses. Currently, research remains focused on single gates or simple circuits, and large-scale optical computers are still in early development.

Silicon photonics is considered a key enabler for optical computing due to its potential for scalability. However, integration remains a challenge in photonic technologies. Progress in integrated optics, using materials such as SOI, SiN, TFLN, graphene, BTO, and polymers, is expected to drive the development of practical optical processors based on photonic ICs. These advancements will also aid quantum optics, paving the way for quantum optical computers with more qubits in compact form factors.

There are several approaches to building an optical processor, whether analog or digital, using different optical media like photonic ICs, free-space optics (FSO), or fibre optics. In the field of optical quantum computing, three primary methods are being explored: one utilising photon qubits and two others using photonics to control non-photonic qubits, such as trapped ions and cold atoms. Additionally, some companies are working on optical quantum computers that do not rely on qubits, instead harnessing optical quantum effects and nonlinearity. Emerging materials, such as metasurfaces and SiC, are also under investigation for optical processors, although they remain in early research stages.

For more insights into the evolving field of optical computing, Yole Group invites you to explore the report How to Compute with Photons – An Interview with Akhetonics, which delves deeper into the advancements and challenges facing this exciting frontier.