Silicon photonics can be the missing link in the UK’s Compute Roadmap

On the 17 of July 2025, the UK government published the UK Compute Roadmap, which is a £2 billion government plan to enhance the UK's computing infrastructure. This roadmap outlines a bold vision to deliver world-class AI infrastructure by 2030. It positions computing not just as a technical utility, but also as a sovereign capability.

But even with this significant amount of investment, the truth is that the UK cannot realistically outspend the US or China in the AI arms race; nor compete in frontier AI model development.

The opportunity lies somewhere else – it is in specialised hardware innovation.

By backing silicon photonics, the UK has a chance to strengthen the foundations of its AI infrastructure and advance its vision for sustainable and sovereign AI compute.

This is not only because silicon photonics offers the UK a pathway to innovate in AI compute infrastructure by dramatically improving efficiency, speed, and energy consumption, but also because it allows us to build domestic silicon photonics chips that are less dependent on overseas disruptions.

What is silicon photonics?

But silicon photonics takes this technology to the next level by combining light-based components onto silicon chips. Some analysts even argue that it has the potential to redefine what constitutes a leading-edge chip. By merging light and electronics, silicon photonics makes it possible to build fast, energy-efficient, and affordable systems for sending and processing data, which are useful in places like data centres.

Silicon photonics is not a new technological invention. The commercial application of silicon photonics has already been proven in conventional data centers. For example, major players such as Intel and Cisco already deploy silicon photonic transceivers in hyperscale data centres to handle explosive data demand.

The global rollout of silicon photonic technologies by major companies proves the market readiness of this technology. Meanwhile, the UK’s opportunity lies in customising and improving it, turning proven concepts into home-grown solutions for sustainable and sovereign AI compute.

Looking further ahead, there is an emerging and promising research area with optical neural networks (which are networks that use light instead of electricity to perform computations).

In the domain of AI compute, the research explores how light can be not only a carrier of information (moving data from one point to another) but also a medium for computation itself. These early designs rely on or complement silicon photonics platforms. This opens doors for new hardware design innovations in AI, which can be faster, greener and more energy efficient.

It already has a strong academic research base and expertise. It is a home to one of the pioneering research groups in silicon photonics, having been active in the field since 1989, and with a plethora of high impact research groups active in the area spanning all 4 UK nations. The University of Southampton holds the world record for the fastest data rate from an integrated silicon transmitter of a particular common design configuration, a faster and greener version of a device that Intel has been deploying in data centres for the past decade. 

What is the next step?

The UK Compute Roadmap makes a promise to support UK tech firms by advancing secure, efficient, and sustainable compute infrastructure with specialised chips.

While the Compute Roadmap provides the policy direction, the next step should be to provide a physical infrastructure – to turn research excellence, ideas into industrial capability.

There are many UK-based companies already leveraging silicon photonics chips, but what they lack is access to infrastructure for testing and scaling up new ideas and products. Before they can produce silicon chips in massive amounts and access big foundries, they need access to prototyping facilities to prove that their product works and that it can be manufactured at scale.

One solution is to have a UK-based silicon photonics pilot line, which can be defined as a semi-industrial facility designed to bridge the gap between lab-scale prototypes and full-scale commercial production. It allows companies to test, refine, and scale up photonic circuits in a reliable and repeatable environment before committing to mass manufacturing.

And the benefits of having a UK-based silicon photonics pilot line are clear. Firstly, such a pilot line lowers the risk of overseas supply chain disruptions. Secondly, it reduces the risks of sending UK intellectual property, know-how outside the UK jurisdiction. It can encourage a knowledge spillover into the UK’s supply chain, offering opportunities for packaging, testing, and materials companies. Not to mention, it creates jobs and economic growth within the UK borders.

The UK will never match the AI budgets of the US or China. The good news is that it does not have to. It should strategically invest in silicon photonics technology and its necessary infrastructure to secure innovative, domestics solutions in the UK for future sustainable and sovereign AI compute.

Author

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.