How photonics could give the UK a competitive AI advantage

OpenAI recently announced that it would be pausing its ambitious Stargate UK project. The company cited the cost of British industrial electricity and an unresolved regulatory environment as the reason it was walking away from a commitment to deploy 8,000 GPUs that were supposed to be live by Q1 2026.

Meanwhile, a recent investigation found that the billions in announced AI investments underpinning the government's strategy are largely phantom figures: a flagship supercomputer site in Essex was still a scaffolding yard when visited in February, and the government admitted it had no mechanism to verify the commitments it was counting as delivered.

The pattern is becoming hard to ignore. The promise to make the UK the fastest AI adopter in the G7 will ring hollow if policy continues to rely on headline deals with US hyperscalers rather than building on what the UK actually has. What is being left out of that conversation is the country's most distinctive opportunity: a technology that does not just power AI more efficiently but fundamentally reimagines how AI infrastructure works. The technological solution is photonics.

From fibre to the future

Photonics has long underpinned telecommunications, powering fibre optic networks that transmit data globally. But once that data reaches a computing device, it has to be converted from light into electricity so it can be processed, switched, and routed. That conversion introduces latency and consumes significant power.

Photonics is now moving beyond its traditional role in fibre and transceivers to encroach into parts of the compute and network infrastructure traditionally dominated by electronics. It brings the promise of increased speeds, elimination of latency, and reduced power consumption at a systemic level.

This approach is where the UK has a unique industrial opening.

Why photonics changes the game

For decades, electrons have been the carriers of data inside devices, performing all computation and routing. But electrons come with limits: heat generation, power inefficiency, and bandwidth bottlenecks. Every time a signal switches between optical and electronic domains, latency and energy cost increase.

Photonics replaces some of these electrical pathways with optical ones, allowing data to travel at the speed of light while reducing energy consumption dramatically. Emerging architectures such as co-packaged optics bring optical data transceivers directly next to compute chips, cutting power and latency by shortening conversion distances.

Beyond this, technologies like optical interposers and fully all-optical computing systems, where photons handle processing without conversion, are rapidly progressing.

All optical network switches can reduce switch power consumption by up to 90% dramatically lowering latency, creating a huge opportunity for modern data centres, especially those supporting AI workloads, where bandwidth and energy demand grow exponentially.

Have you read?

AI-powered grid assistant goes open source

Why repowering wind is key to meeting the AI power surge

A manufacturing opportunity hiding in plain sight

Advanced electronic semiconductor fabrication is a mature industry dominated by overseas behemoths. For the UK, the most realistic strategic position in electronics is to remain strong in design while playing a smaller role in global manufacturing. By contrast, photonic integrated circuits (PICs) is a fast-growing emerging market in which the UK could hold a large stake.

Unlike electronics pushing to 2–5 nanometre nodes requiring vast mega-fabs, photonic integrated circuits operate typically at 200 nm features today, requiring less complex manufacturing and lower capital investment.

Photonics manufacturing prioritises materials science and integration over extreme miniaturisation, an area where the UK already holds considerable strengths. Facilities such as Cornerstone in Southampton, the CSA Catapult in Newport, and emerging R&D infrastructures in Glasgow and Sheffield offer an advanced network, ready to scale.

Photonics relies on a diversity of materials such as silicon nitride for low-loss waveguides, indium phosphide for lasers and detectors, thin-film lithium niobate for modulators - creating a landscape less dominated by one production model, and more amenable to flexible, innovation-led ecosystems.

The UK’s blend of academic excellence, industrial design, and agile manufacturing therefore, makes it a prime contender in high-value, low-volume photonics production.

For the energy sector specifically, this matters beyond industrial policy. As utilities and grid operators accelerate the digitalisation of network infrastructure, the underlying optical components that enable high-speed, low-latency data transmission will become a critical procurement. A domestic photonics manufacturing base means supply chain resilience for the very infrastructure the energy transition depends on, rather than importing components from markets with competing strategic interests.

Data centres, AI, and the photonics power shift

AI’s progress has outpaced the energy infrastructure it relies on, leading to a 12 to 24-month gap between data centres’ power demands and grid capacity. Even hotspots like Northern Virginia and Dublin, are struggling to keep up with the energy load of hyperscale computing.

The UK faces the same challenges: power-hungry electronics in an era of tightening energy grids and rising prices.

Ofgem recently reported that around 140 data centres are currently queuing to connect to Britain's power grid, and their combined energy requirements are estimated to exceed the country's entire peak electricity demand. That demand won’t be solved without reducing what each facility actually draws upon. If photonics-based infrastructure can cut switching power consumption by up to 90 per cent, the effective load entering that queue shrinks.

This new approach to photonics offers one of the few available solutions that scales sustainably. By cutting switching losses, minimising heat, and reducing the energy per bit transmitted, photonics directly addresses both cost and carbon intensity in AI infrastructure.

Leading chipmakers are already embracing this shift. NVIDIA’s co-packaged optical interconnects bring light closer to silicon to save power and bandwidth, while hyperscalers are beginning to redesign networks around light-based platforms for precisely these reasons.

Let there be light

This opportunity won’t stay open forever. Across Europe, companies are receiving major funding through the EU Chips Act for glass-based photonics substrates. The US, Japan, and Canada have similarly accelerated public-private investment around photonic integration and packaging. If the UK hesitates, it risks falling behind in the next great manufacturing wave.

To secure leadership, the UK needs a coordinated national strategy where industry, government, and academia collaborate from prototype to production. That means targeted backing for high-value materials like lithium niobate and indium phosphide, increased support for scale-up centres such as Cornerstone, and strong ties with international hubs like Canada’s C2MI.

Photonics will be the physical foundation of the next digital economy. AI, quantum computing, and cloud networking all depend on the ability to move and process data faster, cooler, and smarter. In the ongoing race to optimisation, light wins every time.

About the author: Mark Rushworth is founder and CEO of the all-optical network switch company, Finchetto. Mark holds certificates in integrated photonics chip design and is a member of the UK Government's Optical Communications and Photonics Expert Working Group.