Olix’s $312M Series B Pushes Valuation to $3.3B as It Rebuilds AI Datacenters

James Dacombe’s Olix has secured $312 million in Series B funding at a $3.3 billion valuation, a deal that signals the company’s ambition to rebuild the datacenter from the ground up for the age of large AI models. The round, announced on August 3, 2026, comes without a disclosed lead investor, but the capital will accelerate Olix’s work on what it calls the “token factory” — a reimagined hardware architecture that turns AI inference into a production line rather than a single-chip bottleneck.

Olix’s core insight is that the way tokens are produced today is fundamentally broken. Every AI model runs on general-purpose chips, even though generating a token involves many distinct operations, each with different demands on compute, memory, and bandwidth. Moving work between specialized chips has historically been too slow and energy-intensive to be practical, so the industry has settled for monolithic processors that try to do everything at once. The result, Olix argues, is that no matter how large a single chip grows, the system cannot deliver high throughput, low latency, and low cost simultaneously. Tokens remain scarce and expensive, and the rising demand for extended reasoning in frontier models only widens the gap.

Olix’s answer is to break the process apart. The company is building dedicated hardware for different stages of token generation, with fast, cheap interconnects that make it viable to shuttle data between them. The approach parallels the factory production line, where specialized machines handle separate tasks, and the product flows efficiently from one station to the next. Olix’s immediate focus is the memory layer for inference — a critical bottleneck in serving large models that the company sees as a key unlock for scaling AI without compromising interactivity or cost.

The funding will be used to advance that production line, scaling the team and building out the infrastructure necessary to bring the token factory to market. Olix is hiring across the US, Canada, and the UK, and the website explicitly ties the Series B to an accelerated buildout of its memory-layer technology. The company’s messaging frames the raise as a direct response to an industry-wide constraint: AI progress is compounding, but the infrastructure beneath it is not keeping pace. By shifting from general-purpose chips to a purpose-built production system, Olix aims to make frontier AI more accessible and economically viable.

In a market where AI infrastructure spending is soaring and the cost of inference is a growing concern for enterprises and developers alike, Olix’s valuation reflects investor belief that the next wave of efficiency gains will come from architectural overhauls rather than incremental chip improvements. The company’s approach pits it against the prevailing model of scaling ever-larger GPUs, but it also places it in a small cadre of startups betting that the datacenter itself must become a specialized machine. With a $3.3 billion valuation and a founder who has kept the company’s profile low until now, Olix steps into the spotlight as a potentially disruptive force in AI infrastructure.