According to reporting in the Financial Times, the funding will support the design and rollout of next‑generation optical‑based AI chips optimised for inference workloads – the computationally intensive process of running large AI models in real‑world environments. OLIX expects its first commercial products to reach customers in 2027 and plans to expand its engineering teams in the UK and North America as it scales development and prepares for manufacturing.
Founded in 2024 by James Dacombe, OLIX has now been valued at $1 billion following this latest round, led by Hummingbird Ventures. The company has also attracted backing from Plural, Vertex Ventures, LocalGlobe, and Entrepreneurs First.
Rather than compete directly with Nvidia’s current GPU architecture, OLIX is pursuing a different route by developing an optical digital processor featuring what it calls a “novel memory and interconnect architecture.” The design is intended to support existing AI models while delivering high throughput and responsiveness for demanding inference tasks.
The company argues that conventional semiconductor designs are increasingly unable to keep pace with user requirements. On its website, OLIX says it is “building a new class of accelerator designed to achieve high throughput and high interactivity on the most demanding inference workloads, free from the architectural and supply chain constraints of the current regime.”
A key aspect of its approach is the integration of photonics with an SRAM‑based memory architecture, which OLIX claims can outperform high‑bandwidth memory (HBM) systems in efficiency, throughput per watt, and overall cost. By avoiding HBM and advanced packaging, the company says it can sidestep supply‑chain pressures that are affecting the wider industry and are expected to continue into 2027. “From a supply chain and fabrication perspective, a new architecture must forego HBM, advanced packaging, or any other technology that is supply chain‑constrained by current incumbents. When even the biggest hyperscalers are struggling to secure capacity, a startup simply cannot compete,” OLIX wrote in its Compute Manifesto.
Investors see the company’s strategy as an attempt to resolve the core computational bottleneck limiting widespread AI deployment. OLIX’s processor – described as an Optical Tensor Processing Unit (OTPU) – performs bit‑perfect digital computation using optical techniques, differentiating it from earlier analogue optical systems. The company says it has already demonstrated integration across optical, electronic, and software components and is now developing custom ASICs to further support its architecture.
With interest in alternatives to Nvidia’s GPUs rising across the AI hardware landscape, OLIX’s move into optical compute places it among a growing cohort of startups hoping to reshape the future of inference‑focused silicon.
Leave a comment