- OLIX, the two-year-old London AI chip startup founded by James Dacombe, announced a $312m Series B financing round at a $3.3 billion valuation.
- The financing round included Fundomo, Arm, and Hudson River Trading, along with angel investors such as Reed Hastings, co-founder of Netflix.
- OLIX aims to produce AI chips that are faster and cheaper than those of Nvidia, the current market leader.
- This funding round has tripled OLIX's valuation from just over $1 billion in February when it raised $220m.
- The financing will support the delivery of OLIX's DX-1 chip to its first customers by H2 2027.
OLIX, a London-based AI chip startup founded by James Dacombe, has successfully raised $312 million in a Series B financing round, elevating its valuation to $3.3 billion. This funding round saw participation from notable investors including Fundomo, Arm, and Reed Hastings, co-founder of Netflix.123
The startup aims to address the limitations of current AI inference hardware by developing specialized chips for each stage of token production. OLIX's innovative X-1 platform utilizes a unique 'slow and wide' optical interconnect, allowing data transfer between chips using light, which significantly reduces latency and energy costs.
The first chip, DX-1, is designed to optimize reasoning and output generation, achieving over 10,000 tokens per second per user for models with 100 billion parameters. This architecture is scalable to models with 10 trillion parameters and above, positioning OLIX to compete with industry giants like Nvidia.
The financing will support the delivery of DX-1 to customers by H2 2027 and the expansion of OLIX's custom silicon platform. The company is also actively hiring across various engineering disciplines in multiple locations, including London and San Francisco.5
Despite the enthusiasm in private markets, the broader chip industry has faced volatility, with concerns over debt funding impacting publicly listed chip stocks. OLIX's innovative approach and strategic funding position it well within the booming AI infrastructure market.
“OLIX's X-1 platform unrolls models across chips over a 'slow and wide' optical interconnect; its DX-1 accelerator targets 100B-parameter models at over 10,000 tokens per second per user. The round funds DX-1 delivery by H2 2027 and the wider platform build-out, while the design avoids advanced packaging and high-bandwidth memory amid supply shortages.”
