Avicena is a privately held company headquartered in Sunnyvale, California, with a subsidiary in Edinburgh, Scotland. Founded in 2019 by leading technologists from the optical networking industry, Avicena specializes in developing ultra-low energy optical interconnects based on microLED technology. These interconnects offer class-leading bandwidth density and energy efficiency for medium-reach applications, including chip-to-chip connections in high-performance computing (HPC), artificial intelligence/machine learning (AI/ML), and memory disaggregation. Avicena's technology is a key building block in the evolution of new networking and computing architectures that aim to reduce the energy impact on our planet.
Key Features and Functionality:
- Ultra-Low Power Consumption: Avicena's microLED-based interconnects consume up to five times less power than traditional alternatives, achieving energy efficiencies of less than 1 picojoule per bit.
- High Bandwidth Density: The company's LightBundle™ technology supports shoreline densities exceeding 1 terabit per second per millimeter, enabling scalable, high-performance interconnects.
- Extended Reach: Avicena's interconnects can extend up to 10 meters, facilitating the scaling of GPU clusters across multiple racks and connecting thousands of GPUs.
- Innovative MicroLED Integration: As the first company to utilize microLEDs for high-speed communication, Avicena leverages high-volume, high-yield, low-cost components to ensure reliability and cost-effectiveness.
- Wide Operating Temperature Range: The interconnects operate efficiently across a broad temperature spectrum, from -55°C to 125°C, ensuring performance in diverse environments.
Primary Value and User Solutions:
Avicena's microLED-based interconnects address critical challenges in modern computing infrastructures by providing energy-efficient, high-bandwidth, and scalable solutions for chip-to-chip communication. By enabling GPU clusters to scale across multiple racks and connecting thousands of GPUs, Avicena's technology overcomes the limitations of traditional copper interconnects, which are constrained by reach and power consumption. This advancement is particularly beneficial for applications in AI/ML, HPC, and memory disaggregation, where high-speed, low-latency, and energy-efficient data transfer is essential. By reducing power consumption and increasing interconnect reach, Avicena's solutions contribute to more sustainable and scalable computing architectures.
