As the AI data centre arms race shifts from single-chip performance to cluster scale, the “roads” connecting GPUs are becoming the new bottleneck. Oxfordshire-based Salience Labs is paving those roads with light—and just secured a $125 million bet.
In October 2026, photonic chip startup Salience Labs announced the close of an oversubscribed $125 million Series B round. The round was led by Bullhound Capital, with the most striking detail being the simultaneous participation of Intel Capital and AMD Ventures two chip giants that compete fiercely in CPUs and GPUs, yet rarely agree on backing the same early-stage hard-tech company.
From Lab to Fab: A Decade-Long Journey for a Beam of Light
Salience Labs was founded in 2021, with its technology roots tracing back to over a decade of silicon photonics research at the University of Oxford and the University of Münster in Germany. CEO and co-founder Vaysh Kewada holds a physics degree from Imperial College London and an MBA, and previously worked at McKinsey before serving as an entrepreneur-in-residence at Oxford Science Enterprises to commercialise the technology. Co-founder Johannes Feldmann is a photonics physicist who helped develop and test the company’s early prototype.
The problem the company addresses is rapidly escalating: in AI data centres, thousands of GPUs need to interconnect across racks, and the power consumption and latency of traditional electrical switches have become a hard constraint on scale-up networks. Salience’s all-optical circuit switching (OCS) chips replace electrons with photons for signal routing. The company claims its technology can significantly reduce latency and power consumption compared with traditional electrical switching solutions.
64-Port Prototype Demonstrated, 128 and 256 Ports in Sight
A month before the funding announcement, Salience live-demonstrated a 64-port all-optical silicon photonic switch at ECOC 2026 in Málaga, Spain, which the company described as an industry first for silicon photonic OCS. The company had previously launched a 32-port version. The new capital will directly fund development of 128-port and 256-port products, while accelerating the build-out of high-volume manufacturing capabilities.
On the manufacturing side, Salience is working with Tower Semiconductor on wafer fabrication and Keysight Technologies on testing. The company also plans to establish a second West Coast office in Silicon Valley to be closer to AI data centre customers.
Why Now: A Near-Trillion-Dollar Market Gap
According to Dell’Oro Group, cumulative spending on data centre switches for AI back-end networks will approach $1 trillion from 2026 to 2030, with scale-up networks expected to account for more than half of the market by the end of that period. Intel Capital Managing Director Srini Ananth’s assessment in the investment statement is telling: “Photonics is quickly becoming essential to scaling AI infrastructure, and optical circuit switching can unlock significant gains in performance and efficiency.”
For this five-year-old Oxfordshire hard-tech company, $125 million is one of the most significant votes of confidence the sector has seen to date. But the real test is just beginning: from 64-port demonstrations to mass production delivery, from Oxfordshire laboratories to Silicon Valley customer racks, Salience must prove that light can not only run but run at scale.
