Q/C Technologies (Nasdaq: QCLS) said on September 30, 2026 it has begun a research collaboration with Sandia National Laboratories' Center for Integrated Nanotechnologies (CINT) to identify and verify component technologies for an optical processing unit (OPU) targeted at AI inference. The work covers nanophotonic devices and the physical primitives required to compute optically at scale. The two parties framed the effort as a hardware roadmap exercise rather than a finished architecture.

What is the scope of the Q/C–CINT collaboration?

The joint program focuses on nanophotonic components — devices that manipulate light at sub-wavelength scales — and on the building blocks needed to run AI workloads optically. According to the announcement, the teams will catalogue candidate optical operations, prioritise device technologies and document failure modes in five areas: nonlinear operations, memory, numerical precision, optical loss and electrical integration.

Why Sandia's Center for Integrated Nanotechnologies?

Sandia, managed for the Department of Energy's National Nuclear Security Administration, employs roughly 17,000 staff and runs an annual budget of approximately $5 billion. The lab has established programmes in microelectronics, photonics, advanced materials and high-performance computing.

CINT, a nanoscience user facility within Sandia, offers fabrication and characterisation tools that most commercial start-ups cannot afford to operate in-house. Financial terms of the collaboration were not disclosed.

What does management say?

Q/C Executive Chairman Joshua Silverman framed the work as an architectural evaluation. "Access to CINT's scientific expertise and advanced research capabilities in photonics and nanoscale technologies represents an important step as we develop our proprietary optical computing platform," he said. Optical computing, he added, can address "increasing constraints in power consumption, bandwidth and scalability" in AI infrastructure.

Chief Technology Officer Yossef Ehrlichman, Ph.D., described the collaboration as path-finding. "This collaboration provides an opportunity to accelerate our work toward a practical, scalable optical computing architecture," he said. "We believe this work with CINT will help us define the pathway from fundamental optical capabilities to an integrated OPU architecture."

What technical challenges remain?

The announcement lists five open problems that determine whether an OPU can compete with current GPUs and accelerators:

• Nonlinear operations, where optics typically requires inefficient electro-optic conversion
• Memory access, because no optical equivalent of DRAM exists
• Numerical precision, since analogue optical compute accumulates noise differently than digital electronics
• Optical loss, which compounds across cascaded stages
• Integration with electronic control planes, I/O and thermal management

Neither party published benchmark data, simulation results or projected throughput figures in the announcement.

What is Q/C Technologies' standing?

The San Francisco-based company recently opened a 4,800-square-foot integrated photonics laboratory to support OPU development. On September 30, QCLS traded at $0.92 in delayed pricing, up 2.52% on relative volume of 114.9× its trailing average, with a market capitalisation near $7.5 million and an intraday range of $0.80 to $1.12.

That capital base underlines why a national-laboratory user facility matters. Photonic process development — e-beam lithography, cleanroom time, advanced metrology — typically exceeds the capex of a small public company. CINT access converts fixed R&D overhead into variable cost.

What should R&D managers watch?

The deliverable is a verified component and integration roadmap, not a working chip. Researchers tracking the programme should watch for:

• Peer-reviewed publications from CINT staff describing measured device performance
• Q/C patent filings on optical-memory or nonlinear schemes
• Foundry or packaging partner announcements indicating a path beyond CINT

If the roadmap delivers device-level specifications that survive independent characterisation, Q/C will have a credible basis to raise development capital for a first-generation OPU. If the bottlenecks — especially memory and nonlinearity — resist photonic solutions, the company may need to recast its pitch around hybrid optoelectronic architectures rather than a fully optical pipeline.