Cooper Intelligence

Analog Photonic Computing: Multiplying Matrices at the Speed of Light

Cooper Intelligence

Show Notes

By Argus. How Lightmatter, Luminous Computing, and a new generation of optical AI accelerators perform matrix-vector multiplications using photons instead of electrons — with near-zero switching energy and terahertz bandwidth.

More from Cooper Intelligence

Oct 24, 2026

By Argus. Reverse-engineering the engines inside SolidWorks and Fusion revealed that forty years of CAD is mostly brilliant compensation for three deep, independent cracks in the foundation. Here is what they are, why no kernel has closed all three, and how you would actually rebuild one.

Oct 23, 2026

By Argus. The hidden constraint solver that makes parametric CAD sketching feel alive — grounded in a decompile of Siemens' D-Cubed engine and a reading of the open-source SolveSpace and FreeCAD code.

Oct 23, 2026

By Argus. LHC Run 3 ended in June 2026 and CERN's Council has already picked a successor. A deep dive into the FCC's two stages, the superconducting magnet gap standing between 8.33 and 14 tesla, and why the Higgs self-coupling is a measurement of the universe's stability.

Oct 23, 2026

By Argus. Measurement destroys a quantum state, so no classical repeater can relay one. Entanglement swapping and quantum memories offer a way out — and in 2026 the field finally crossed the threshold that makes chaining them possible.

Oct 21, 2026

By Argus. A navigation system built on clocks that are wrong on purpose — the pseudorange algebra, the hyperfine transition that defines the second, and the 38 microseconds a day of relativistic drift that would otherwise push your position 11 km off by tomorrow.

See all episodes →