Oxford spin-out Salience Labs raises $125 million for light-based AI data centre switches

Salience Labs, a British start-up spun out of the University of Oxford, said on Wednesday, October 14, 2026, that it has raised $125 million in an oversubscribed Series B round to scale up production of switches that move data between AI chips as light rather than electricity. The round was led by Bullhound Capital, with the venture arms of chipmakers AMD and Intel, AMD Ventures and Intel Capital, taking part alongside SBI Group, Hudson River Trading and the British Business Bank. Existing investors, including Applied Ventures, Oxford Science Enterprises, Cambridge Innovation Capital and Silicon Catalyst Ventures, also joined.

Training and running large AI models requires thousands of chips to work together, and the speed at which they can exchange data is becoming as important as the speed of the chips themselves. Salience's switches are designed to keep the signal as light throughout instead of converting it to electrical form along the way, an approach aimed at linking very large numbers of chips.

The investor list is telling. The presence of AMD and Intel's venture arms signals that major chipmakers see optical switching as part of the future of AI infrastructure. Backing from the British Business Bank also shows public support in the UK for deep-tech companies growing at home even as they expand in the United States.

For Indian founders and engineers, the deal highlights where some of the largest cheques in deep tech are going: hardware that makes AI data centres faster and more efficient. India is building its own AI computing capacity, and specialised hardware, from photonics to chip design, is a field where university research can turn into globally funded companies, as Salience's path from Oxford shows.

Short-term: Salience will move from demonstration units towards volume manufacturing and customer deployments, and expand its team in California, close to major AI and cloud companies.

Long-term: If optical switching proves reliable at scale, it could change how AI data centres are built, reducing energy use per computation. Commercial availability of larger switches still depends on further development and qualification.

Who is affected: AI data centre operators and cloud providers; chipmakers such as AMD and Intel; deep-tech investors; and engineers and researchers in photonics.

Questions and Answers