Published Jul 6, 2023

Celestial AI Raises $100M to Massively Boost AI Model Performance with "Photonic Fabric"

Explore Celestial AI's groundbreaking innovation 'Photonic Fabric' as they secure $100M in funding to revolutionize AI model performance, with insights from Jaeden Schafer on the technology's market impact and transformative potential.
Episode Highlights
AI Chat: ChatGPT & AI News, Artificial Intelligence, OpenAI, Machine Learning logo

Popular Clips

Episode Highlights

  • Technical Details

    Celestial AI's Photonic Fabric technology addresses the limitations of traditional electrical interconnects in data transfer. explains that as AI models grow in size, the need for faster data transfer becomes critical. Photonic Fabric offers low latency and high bandwidth, enabling more efficient use of memory and computing resources 1. This innovation allows data to be delivered directly to any point on the silicon die, significantly reducing latency issues 2.

    By enabling each processor to store and process larger chunks of data, the photonic fabric reduces the number of processors needed and provides the fastest link between processors, allowing them to process much faster and really just increases the speed overall.

    ---

    This advancement is particularly beneficial for large AI models, potentially reaching trillions of parameters, as seen with models like GPT-4 and those developed by Meta 1.

       

    Potential Benefits

    The Photonic Fabric technology not only enhances data transfer speeds but also improves overall system efficiency. highlights that this technology can help reduce waste and improve memory utilization, which is crucial for handling large-scale AI models 1. By partnering with top-tier tech companies, Celestial AI aims to overcome current limitations in data management and memory capacity 2.

    This new technology has reportedly overcome one of the really, I mean, you could call it like a quote, unquote, beachfront problems, but essentially, it's offering much higher bandwidth and lower delay times.

    ---

    The recent funding and collaboration with companies like Broadcom indicate a promising future for this technology, with prototypes expected to be ready within the next 18 months 2.

Related Episodes