Great Sky

Great Sky is a Boulder, Colorado-based AI hardware company developing a computing architecture it calls Superconducting Optoelectronic Networks (SOENs). Rather than running neural networks on conventional silicon processors, the company builds superconducting circuits that operate at cryogenic temperatures and behave like biological neurons, links them with optical communication at the single-photon level, and uses semiconductor circuitry to bridge electronic and photonic domains through amplification, light emission and control logic. Memory and processing are co-located, which the company says reduces data movement, enables high-speed programmability, and supports on-device learning so systems can adapt to new data streams without constant retraining cycles.

The technology stems from 12 years of research at the U.S. National Institute of Standards and Technology, where the founding team produced 23 peer-reviewed publications and eight patents. The company publicly debuted in March 2026 alongside a seed financing, reporting multiple successful chip tape-outs and first-generation chips running inference. Great Sky positions the platform for high-throughput, low-latency workloads including spoken language processing, continuous video analysis and large multimodal models, citing video analysis at more than 60 million frames per second versus roughly 30 frames per second on typical GPU-based systems. Longer term, the company plans multi-modular cognitive systems formed by tiling wafer-scale networks with fiber-optic interconnects.

Founding story

Great Sky was founded by four former NIST researchers specializing in photonics, superconductors and semiconductor systems: CEO Jeff Shainline, CTO Jeff Chiles, VP of Fabrication Saeed Khan and VP of Architecture Bryce Primavera. The company spun out of a 12-year research program at NIST in which the team proved individual superconducting and single-photon components and subsequently integrated them into complete neural networks, aided by advances in circuit concepts, manufacturing capability and modeling.

Business model

Great Sky designs and manufactures purpose-built AI compute hardware, positioning its superconducting optoelectronic systems as an alternative compute platform to GPU-based infrastructure for AI workloads.

Traction

As of its March 2026 debut, Great Sky reported multiple successful chip tape-outs, first-generation chips running inference, and benchmark claims of processing more than 60 million video frames per second with substantially lower energy per video than conventional GPU-based models.

Latest developments

In March 2026 the company announced its public debut, a $14 million seed round, chip tape-outs and commercial milestones, and said funds would go toward hardware development, expanding engineering and scientific teams, and commercialization in defense, energy and large-scale AI infrastructure.