New superconducting chip aims to help tame fusion plasma (www.scientificamerican.com)

Brookhaven National Laboratory has unveiled a groundbreaking superconducting chip designed to enable real-time monitoring of fusion plasma, developed in collaboration with other U.S. Department of Energy facilities and AI hardware startup Great Sky. This innovative chip aims to address the challenge of managing erratic plasma behavior within tokamaks—fusion reactors that use magnetic fields to contain high-temperature plasma. Disruptions in plasma can lead to catastrophic energy releases, making rapid response critical to achieving stable fusion energy production. What sets this chip apart is its use of neuromorphic AI systems, which mimic human brain structure to process vast amounts of data simultaneously, potentially allowing it to respond much faster than traditional control systems. While the chip has not yet been implemented in a working tokamak, it successfully processed real-time data from the DIII-D tokamak at a staggering rate of a million samples per second, identifying anomalies and predicting disruptions. Looking ahead, the team plans to develop a more advanced chip featuring a larger neural network, which they anticipate will be tested with fusion plasma to further explore its potential applications, extending beyond fusion to fields like edge computing where traditional data centers may be impractical.