Olix raises $220M at $1B valuation to challenge Nvidia in AI inference

What’s the deal? London-based AI chip startup Olix, previously known as Flux Computing, has raised $220 million in a round that values it at just over $1 billion. Founded in 2024 by 25-year-old James DacombeDealroom has a profile for this one. Try Dealroom →, the company has secured $250 million in total funding.

Olix is building what it calls a “new class of accelerator” for AI inference — the computing needed to run models after training. It says its optical digital processor, built on a novel memory and interconnect architecture, aims to be faster and cheaper than Nvidia’s chips while remaining compatible with existing AI models.

Why now? Inference demand is rising as AI agents grow more complex. Tools such as AnthropicDealroom has a profile for this one. Try Dealroom →’s Claude Code require significantly more computing power than earlier chatbots. Most challengers to Nvidia have adapted graphics processing unit designs originally built for gaming. Olix is betting inference requires a ground-up rethink.

Investor interest in AI chips has also revived. Nvidia’s recent partnership with rival Groq has renewed attention on a sector long seen as too risky and capital intensive.

What could go wrong? Nvidia still dominates AI infrastructure, and no startup has meaningfully dented its position. Chip design is expensive, slow, and dependent on tight global supply chains.

US competitors are raising far larger sums, increasing pressure on smaller players. Delivering a fundamentally new chip architecture is technically demanding and execution risk remains high.

The signal: Capital is shifting from AI applications to the infrastructure layer. As inference workloads surge, investors are backing new hardware architectures rather than incremental GPU improvements.

The deal also reflects the UK’s push to strengthen its position in AI infrastructure, a market largely led by US companies.

Sources:
Financial Times
NiftyGPT

A.M.