OLIX Secures $312 Million to Scale Photonic AI Compute Infrastructure

The post-money valuation lands at $3.3 billion — a number that converts a UK lab project into a line item on a limited partner's quarterly report.

Capital stack and who's exposed

Fundomo anchors the round. Arm — itself a UK-domiciled chip designer — adds strategic optics without aggressive dilution; the alignment reads well in Westminster policy memos and shores up OLIX's pitch as the answer to a politically convenient question: where is Britain's domestic compute stack? Hudson River Trading rounds out the cap table with the quant angle. Photonic compute has been pitched for years as a latency-killer for inference workloads, and HRT's signature trade is on microsecond edges where conventional silicon leaves money on the table.

The structure matters more than the headline figure. Series B by label, but $312 million against a $3.3 billion valuation puts OLIX in the territory where institutional crossovers begin to circle. The undisclosed primary-versus-secondary split will determine whether this is genuine runway extension or a partial liquidity event for early backers. Either way, at this cap table the company is no longer priced as a science project.

The sober read

Photonic chips remain unproven at production scale. The valuation imports Nvidia-adjacent multiples onto a startup whose commercial silicon story, on the disclosed record, still has to be built. The path from $3.3 billion to a credible exit runs through one of two doors: hyperscaler M&A — historically a buyer's market for chip pretenders — or a public listing into a market that has shown limited patience for hardware narratives in recent quarters.

Watch the lead investor on the next round. A crossover fund stepping in signals maturity and clean optics for an eventual IPO. Another specialist VC at the next milestone means the cap table is full, the round was tight, and patience inside the syndicate is running thin. The dollar amount raised today is the easy number to remember. The multiples implied are the harder one to defend.