Olix's AI Chip Breakthrough
UK Startup Olix Secures $312M Series B, Valued at $3.3B for AI Hardware
Anonymous
Key Points
1
Founder James Dacombe is a 25-year-old Thiel Fellow who started Olix in March 2024; the company was formerly Flux Computing before rebranding to Olix in January 2026.
2
DX-1 decode accelerator is designed for the output stage of large AI models, with a focus on models up to 100 billion parameters, and relies on fast on‑chip SRAM rather than HBM to avoid current supply-chain bottlenecks, aiming for high tokens-per-second.
3
The X-1 platform uses a 'slow and wide' optical interconnect to link multiple chips, enabling an end-to-end system that includes lasers and a dedicated photonic interconnect rather than copper-based links.
4
Olix's Series B was led by Fundomo with participation from Arm and Hudson River Trading, and Netflix co‑founder Reed Hastings; the round also includes backing from the UK Sovereign AI fund, with plans to tape out the DX-1 and begin customer deliveries around 2027 while expanding teams across London, Bristol, Toronto, Austin, and San Francisco.
5
Olix faces competition from other players challenging Nvidia in AI compute, including Cerebras and Groq, as it pushes a photonic/ specialized-chip approach for AI inference.
UK startup Olix Computing has raised $312 million in a Series B round, valuing the company at $3.3 billion — a more than threefold jump from its roughly $1 billion valuation earlier this year, according to Sifted. The round was led by New York-based investment firm Fundomo, with backing from chipmaker Arm, trading firm Hudson River Trading, Netflix co-founder Reed Hastings, and the UK Sovereign AI fund.
Olix is building photonic AI chips — processors that use light instead of electrical signals — designed to run AI models faster and at lower energy cost than today's Nvidia hardware, Electronics Weekly reported. The company was founded in March 2024 and was previously called Flux Computing before rebranding to Olix in January 2026.
25-Year-Old Thiel Fellow Behind the $3.3B Bet
Founder James Dacombe is 25 years old and a Thiel Fellow — a program that pays students to drop out and build companies. He started Olix just over a year ago, according to Sifted. The company has grown from a valuation of roughly $220 million to $3.3 billion in a matter of months, per Tech.eu.
Olix plans to use the new funds to expand teams across London, Bristol, Toronto, Austin, and San Francisco. It expects to tape out — finalize the chip design for manufacturing — later this year, with first customer deliveries targeting around 2027, Yahoo Finance reported.
Light-Based Chips Skip Nvidia's Biggest Bottleneck
Olix's first product is the DX-1 decode accelerator. It targets the output stage of large AI models — the part that generates each word or token in a response. The chip is built around fast on-chip memory called SRAM, avoiding high-bandwidth memory, or HBM, which faces major supply-chain shortages, according to Electronics Weekly.
The DX-1 is designed for models up to 100 billion parameters. Its X-1 platform links multiple chips using a 'slow and wide' optical — light-based — interconnect instead of copper cables. This end-to-end design, from chips to lasers to interconnects, is central to Olix's pitch against conventional AI hardware, Crypto Briefing reported.
Big-Name Backers Signal Confidence in Photonic AI
The investor list reads like a who's who of tech money. Arm — the chip designer inside nearly every smartphone — joined alongside Hudson River Trading, one of the world's largest quantitative trading firms. Reed Hastings, who co-founded Netflix, also participated, per Sifted. The UK Sovereign AI fund's involvement signals government-level interest in keeping Britain competitive in AI chips.
The deal values Olix at more than $3.3 billion, making it one of the most valuable AI chip startups in Europe. Yahoo Finance noted the round highlights growing investor appetite for alternatives to Nvidia in AI inference — the process of running, not training, AI models.
Olix Enters Crowded Race Against Nvidia Challengers
Olix is not alone in targeting Nvidia's dominance. Cerebras and Groq have both built specialized chips for AI inference and attracted significant funding. But Olix's photonic approach — using light rather than electricity to move data between chips — is a distinct technical bet that few others are making at this scale, according to Crypto Briefing.
The core argument is simple: moving data with light is faster and uses less energy than copper. If Olix can deliver that at scale, it could undercut Nvidia on both speed and power costs — two things AI data centers care about deeply. The company's first real test comes when the DX-1 reaches customers around 2027, Electronics Weekly noted.
Sentiment
Mostly PositiveNEGATIVENEUTRALPOSITIVE
Publishers
21
Articles
7
Reach
28
A
AnonymousAuthor