A Rapid Rise in the AI Chip Race
Olix, the UK-based chip designer founded by 25-year-old James Dacombe, has raised a $312M Series B round at a $3.3BN valuation — a more than 3x increase from its $1BN valuation just six months ago, when it closed a $220M round in February 2025. The startup is only two years old.
The round was led by New York VC Fundomo, with notable participation from Arm, chip-industry heavyweight, and Hudson River Trading, the quantitative trading firm. Angel investors include Reed Hastings, co-founder of Netflix. Existing backers — including the UK government's Sovereign AI fund — all increased their commitments.
The Bet Against the Generalist Chip
Olix's core thesis is a direct challenge to Nvidia's dominance in AI infrastructure. Rather than competing on raw performance within the generalist GPU paradigm, Olix is building an optical digital processor with a "novel memory and interconnect architecture" designed specifically for AI inference workloads.
The company's argument is essentially an industrial efficiency one:
"A datacenter is a factory whose product is the token. Producing a single token takes hundreds of operations, each placing different demands on hardware. Any other factory would give each stage a machine built for it. Instead, the token factory runs every stage on the same general purpose chip."
Olix contends that stacking ever-higher single-chip specs — Nvidia's iterative playbook — has hit its efficiency ceiling. Specialised chips for each stage of token production, it argues, will unlock step-change gains in both AI performance and cost.
Dacombe, who also serves as CEO of brain-monitoring startup CoMind, told the Financial Times that "we are moving into a world of specialists" — signalling the end of Nvidia handling all AI chip workloads.
Chips, Racks, and a 2026 Market Entry
According to the FT, Olix is in the final stages of designing and testing its chips and expects first customer deliveries next year. Notably, Olix plans to sell its chips not as standalone components but as part of a complete server rack bundling silicon, software, and networking equipment — a full-stack play similar in spirit to what companies like Groq and Cerebras have pursued.
Funding will go toward:
- Bringing chips to market
- Building out the broader silicon platform
- Funding manufacturing and supply chain commitments
Leadership Hires Signal Seriousness
Two high-profile appointments underscore the company's ambitions. Matt Briers, formerly CFO of Wise and the executive who took that company public, has joined as CFO. Nick McKeown — Stanford computer science professor and former Intel executive — has joined the board of directors. Both hires point to a company actively preparing for commercial scale and, plausibly, a future public offering.
What This Means for the Market
Olix is entering a crowded but high-stakes arena. Nvidia still commands an estimated 70–80% of the AI accelerator market, but challengers are multiplying — AMD, Intel Gaudi, Groq, Cerebras, and a wave of custom silicon efforts from hyperscalers like Google (TPUs) and Amazon (Trainium) are all chipping away at the monoculture.
What distinguishes Olix's pitch is its inference-first, specialisation-first framing at a moment when the industry's cost concerns are shifting from training (still Nvidia's stronghold) to inference at scale — where the economics of tokens-per-watt matter enormously to operators running production AI workloads.
For startup founders and technical buyers evaluating AI infrastructure, Olix's 2026 commercial launch is one to watch closely. If its architecture delivers on cost-per-token at production scale, it could meaningfully reshape procurement decisions across the inference stack.
