This London AI Chip Startup Just Raised $312M Two Years After Launch

Jan de Vries · 2026-08-03

Some startups spend years grinding before they catch a big break. OLIX, a London-based AI chip company, just pulled off something remarkable: a $312 million Series B round at a $3.3 billion valuation—only two years after being founded. That kind of trajectory turns heads. And when you look at who's backing them, it's clear the smart money sees something special here. ## Who's Investing in OLIX and Why It Matters The round brought together some heavyweight names. Fundomo, Arm, and Hudson River Trading all participated, alongside angel investors like Reed Hastings—yes, the Netflix co-founder. Existing investors also doubled down, which usually signals confidence from people who've seen the numbers up close. But the funding is only part of the story. OLIX also made two key hires that show they're thinking long-term. Nick McKeown, a Stanford professor emeritus and winner of the 2025 Marconi Prize, is joining the board of directors. And Matt Briers, a former executive at the fintech Wise, is coming on board as Chief Financial Officer. These aren't just decorative appointments—they bring serious operational and technical firepower. ## The Problem with Today's AI Inference Hardware Here's the thing about how AI chips work right now: they're generalists. OLIX thinks that's the problem. "A datacenter is a factory whose product is the token," the company explains. "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." That analogy really lands. Imagine a car factory where every station—welding, painting, assembly—uses the exact same tool. It would be inefficient, right? That's essentially what the AI industry is doing. OLIX believes each new chip generation has just pushed single-chip specs higher. Better generalists, sure, but never true specialists. Their bet is that specialized chips for each stage of token production will unlock a huge leap in both performance and cost. ## The X-1 Platform: A Different Approach OLIX's X-1 platform takes a production-line approach. Models get fully unrolled across many chips, with each chip handling one specific part of the model. The key insight is flexibility—as AI architectures evolve, each chip keeps a flexible compute fabric rather than hard-coding any particular model design. The system also uses what they call a "slow and wide" optical interconnect. Instead of copper, data moves between chips using light. That means ultra-low latency and energy costs, made possible by rack-scale codesign across every link. A fully deterministic compiler schedules workloads across racks. If this works as described, it could make frontier AI dramatically more affordable and abundant. More importantly, it might unlock far more powerful models down the road. ## DX-1: The First Chip in the Pipeline The first chip in the X-1 platform is DX-1, a decode accelerator designed for the stage where a model reasons and generates output. For 100-billion-parameter models, OLIX claims it hits Pareto-optimal inference performance—delivering over 10,000 tokens per second per user. That's faster output at better energy efficiency than general-purpose chips running large batch sizes. The architecture scales to models of 10 trillion parameters and beyond, thanks to a multi-rack scale-up domain. DX-1 holds models in fast on-chip SRAM memory, avoiding advanced packaging and high-bandwidth memory—the exact components facing supply chain shortages across the industry. ## What's Next for OLIX The company plans to use this funding to deliver DX-1 to first customers by the second half of 2027. They're also building out the wider custom silicon platform, plus the manufacturing and supply chain commitments that scaling frontier inference hardware requires. OLIX is hiring across silicon, photonics, compiler, and systems engineering in London, Bristol, Austin, Toronto, and San Francisco. For a company that's only two years old, this is an aggressive timeline. But with $312 million in the bank and some serious talent on board, they've got the resources to try. Whether the specialized-chip approach wins out remains to be seen, but OLIX is certainly making a compelling case.