Photonics should compute, not just connect

Photonics is increasingly recognised as a critical enabling technology for the future of digital infrastructure as AI and other intensive workloads boom.

It is central to the pursuit of higher-bandwidth communications, more efficient interconnects and better performance across increasingly demanding systems. However, treating photonics only as transport technology drastically underestimates the opportunities it presents.

The next frontier for photonics is not only to connect compute more effectively, but for it to participate in computation itself.

In AI, secure computing, advanced communications and other data-intensive domains, the main challenge is no longer simply the availability of processing power.

It is the memory wall and the cost of moving data through the system before useful work can be completed. As workloads grow, that movement creates increasing pressure on bandwidth, power and latency, and exposes the limitations of architectures that still treat communication and computation as fundamentally separate processes.

This data movement bottleneck is where photonics has the potential to move from enabler to foundational architecture.

For years, optical technologies have been valued for their role in carrying data efficiently. That remains essential. But there is now a broader strategic question worth asking: can photonic systems do more than transport signals between electronic processing points? Can they also contribute directly to the transformations that modern workloads increasingly require?

What does this mean for photonics in the UK?

The UK already has significant strength in photonics research, design capability and deep-tech innovation. What it now needs is a stronger framework for turning that scientific and technical strength into global infrastructure leadership and commercial success.

The case for doing so is growing stronger; modern workloads are becoming more structured, more distributed and more sensitive to the cost of moving intermediate data between stages of processing. In AI, that shows up in memory pressure, interconnect overhead and the cost of synchronising large distributed systems. In secure computing, it shows up in the burden of repeated transformations on large, encrypted data structures.

In each of these cases, system performance and costs are increasingly shaped not only by computation, but by the memory requirements for moving the data itself.

Photonics offers a compelling route around that problem because it operates at the physical layer where bandwidth is naturally high and certain classes of mathematics can be aligned with signal propagation itself.

Electronics will remain essential for general-purpose control, programmability and nonlinear processing while integrated photonics will define a new class of high-throughput, structured operations that reduce system-level bottlenecks in ways conventional architectures struggle to match.

Rather than transporting data between memory and processors, transformations are embedded within the data path itself at transceiver line rates.

When compute moves with the data, latency decreases, energy efficiency improves and infrastructure can scale more sustainably.

Photonics at the forefront of frontier compute

The UK has an opportunity to help shape that category. It can do so not by treating photonics as an isolated sector, but by linking it to the infrastructure challenges that industry and Government are already trying to solve.

The Government’s UK AI Hardware Plan explicitly recognises photonics as part of the emerging AI hardware stack. Alongside advanced semiconductors, secure hardware and heterogeneous compute, the plan makes clear that sovereignty will depend on more than access to processors alone.

It will require infrastructure that is efficient enough to scale and secure enough to trust. Photonics speaks to both sides of that challenge: improving bandwidth, latency and energy efficiency across AI systems, while supporting the demanding mathematical workloads behind post-quantum cryptography, confidential AI and privacy-preserving computation.

We must put more emphasis on real workloads, more focus on deployment pathways and encourage stronger collaboration between academia, photonics companies, compute infrastructure players, systems integrators and public-sector stakeholders.

It also means telling a more ambitious story. The value of photonics is not only that it can improve communications efficiency, but it can help redefine where useful computation happens in modern systems.

For policymakers, this should influence how capability is supported. Photonics should be included not only in discussions about connectivity and semiconductors, but also in discussions about AI infrastructure, resilient digital systems and advanced compute architectures. For industry, it should encourage more active exploration of where photonic approaches can address immediate bottlenecks in performance, energy and scalability.

And for the wider ecosystem, it should prompt a more joined-up view of frontier technology. The systems that define the next era of computing will be hybrid, layered and increasingly shaped by how seamlessly we can integrate different physical technologies.

The UK is ideally placed to contribute to that future and cement technological leadership, but it will only do so if it treats photonics as more than connective tissue and explores the transformative potential of photonic Compute-in-Transit.

Technology and Innovation programme activities

techUK bring members, industry stakeholders, and UK Government together to champion emerging technologies as an integral part of the UK economy. We help to create an environment where innovation can flourish, helping our members to build relationships, showcase their technology, and grow their business. Visit the programme page here.