About

OLIX develops optical tensor processing units (OTPUs) that use light-based circuits to address the memory bandwidth bottlenecks inherent in GPU-based AI infrastructure. Founded in 2024, the company operates at the intersection of photonics, optical circuits, and AI accelerator architecture, with a contrarian approach that favors scaling SRAM integrated with photonics over conventional HBM-based designs.

The company's first product, the OLIX Decode Accelerator 1 (DX-1), is purpose-built for the decode phase of AI inference. Its architecture involves rack-scale co-design spanning logic, data movement, packaging, optics, and interconnect - reflecting a system-level engineering philosophy rather than a chip-only focus.

OLIX works across several technical domains including high-performance computing and system-level co-design, positioning itself within the AI infrastructure sector. As an early-stage company, it is pursuing an alternative compute paradigm at a time when the dominant architectures face increasing pressure from the demands of large-scale AI workloads.

Similar companies

Lumai logoLU

Lumai

Lumai, a University of Oxford spin-out, builds optical compute servers that use light for AI inference matrix multiplication, targeting data centre performance and power efficiency.

Lightmatter logoLI

Lightmatter

Lightmatter manufactures photonic chips and optical interconnects that enable high-bandwidth communication for AI supercomputers and data centres.

MatX logoMA

MatX

MatX builds high-throughput chips specifically designed for large language model training and inference, targeting frontier AI labs with purpose-built hardware.

Fractile logoFR

Fractile

Fractile is a UK-based semiconductor company building AI acceleration hardware to radically improve frontier model inference performance.

Eridu AI logoEA

Eridu AI

Silicon Valley hardware startup building infrastructure solutions to accelerate AI model training and inference in data centres.

Astera Labs logoAL

Astera Labs

Astera Labs provides purpose-built connectivity solutions for rack-scale AI infrastructure, specializing in semiconductor-based technologies like PCIe, CXL, and Ethernet for data centers.