We're building computer chips that lose zero energy to resistance — the post-CMOS foundation for faster climate models, drug discovery, and AI at a fraction of the power.
DISCoVER's superconducting chips are projected to deliver 100× the energy efficiency of CMOS at comparable performance, and 10× the processing speed at equal power draw — net of cryogenic cooling costs.1
The fields that hit the hardest limits of today's hardware first.
Large AI models consume more energy per training run than most cities use in a day. Zero-resistance hardware would let the same compute happen at a fraction of the energy — making large-scale AI sustainable long term.
Simulating how a drug molecule binds to a protein requires thousands of parallel calculations that overwhelm CMOS systems. Superconducting logic would make personalized medicine modeling fast enough to be practical.
Global climate simulations require compute that today's data centers can't sustain at scale. Cryogenic chips that waste almost no energy could run models at resolutions currently impossible — earlier warnings, better predictions.
Seminars, faculty spotlights, and the papers behind them.
This deep-dive episode explores the ideas from Jeff Shainline’s recent seminar on superconducting optoelectronic networks (SOENs) and radical rethink of AI hardware.
This paper presents compact, transformer-free interface circuits between SFQ and DQFP superconducting logic families
Superconductor electronics hasn't had its breakout moment yet — which means there's still room to define the field, not just join it. DISCoVER is a seven-university collaboration built around education and open participation, and welcomes contributors at every stage.
One of DISCoVER's five founding commitments is building a culture of collaboration and opportunity through education, open to all — not just those who already work in computing.
