Redefining connectivity: Transforming networks for the AI era

The explosive growth of Artificial Intelligence (AI) has sparked a data revolution, driving unprecedented demand for faster, more efficient communication systems. Yet, this revolution has brought its own set of challenges.

- SHUBHENDU PARTH

Consider this: training a single large language model involves the seamless transfer of vast data streams across thousands of processors in parallel computing environments. The bottleneck? The existing silicon-based optical transceivers that convert electrical signals to optical ones—critical for transmitting data in modern networks—are struggling to keep pace with the speed and scale required by these AI models. This bottleneck not only slows AI innovation but also leaves behind a carbon footprint that is too substantial to ignore.

To tackle this, researchers at Karlsruhe Institute of Technology (KIT) are taking a bold step. Supported by the European Research Council, KIT’s ATHENS project aims to combine silicon, which is relatively cheaper and widely available, with advanced materials like organic compounds and crystals to develop hybrid optical transceivers.