Great Sky, in collaboration with Brookhaven National Lab, Princeton Plasma Physics Laboratory, and the DIII-D National Fusion Facility, has developed a superconducting analog neural-network AI chip designed for real-time sensing and control of fusion plasmas, and demonstrated its capabilities for real-time fusion plasma control.

The team showcased a high-speed AI chip built with superconducting optoelectronic networks to analyze fusion plasmas, able to process data at up to one million samples per second from the DIII-D tokamak.

The release provides contact information along with a link to a full multimedia release and the company website.

Richard Buttery of DIII-D praised the technology’s potential impact on diagnostics and control for fusion technology.

Genesis Mission tests demonstrated the chip processing data from the DIII-D fusion reactor tokamak at unprecedented speeds, illustrating its potential for fusion diagnostics and control.

The work aligns with DOE priorities in advanced AI, quantum technologies, and transformative energy, underscoring potential benefits for fusion control and other fast, complex systems.

It also highlights potential applications beyond fusion, including quantum system control, cybersecurity, and continual-learning AI.

Great Sky asserts significant cost advantages for AI token generation, estimating up to one million output tokens for about a penny, highlighting energy and cost efficiencies versus conventional AI hardware.

Great Sky has completed four chip tapeouts for under $4 million and plans to produce a chip 10x larger, aiming to finish by year’s end and test at Brookhaven in early 2027, followed by deployment at DIII-D the same year.

The project traces Great Sky’s origins to the U.S. National Institute of Standards and Technology, with backing from investors including Bison Ventures, Olive Tree Capital, Matchstick Ventures, Range Ventures, and Access Venture Partners.

Jeff Shainline, founder and CEO of Great Sky, calls this a major step forward and previews a next-generation chip with a neural network 10x larger to be demonstrated in December.

The chip successfully handled a tokamak-simulation test (TORAX) from Google DeepMind, validating responsiveness to rapidly changing plasma behavior.