Light can now be "Frozen." A research team in Italy recently took fast‑moving photons and put them into a supersolid state—light that flows yet also holds its shape. In practical terms, they proved you can pause light on a chip instead of letting it escape. That single trick opens the door to computers that process information at light‑speed with almost no heat.
This breakthrough represents a fundamental shift in how we think about computing. Instead of relying on electrons moving through silicon, we're now looking at photons that can be controlled and manipulated in ways that were previously impossible.
Why it feels like quantum magic
Picture a maze: a regular computer tests one corridor at a time. A photonic or quantum‑style chip lets a swarm of photons explore every corridor simultaneously, then collapse to the best route in one shot. That "maze‑all‑at‑once" effect is why these chips could crush optimisation and certain AI workloads.
Prototype hardware already exists
Canadian start‑up Xanadu and a Swiss‑led academic group have each shown palm‑size boards where tiny lasers, waveguides, and control electronics all live on the same slice of silicon. No liquid‑helium freezers—these run at room temperature. Key buzz term worth knowing: GKP qubits (a way to encode error‑tolerant quantum information in light).
Why CFOs should care
A top‑end Nvidia GPU gulps about 350 watts. Early photonic chips run on milliwatts, with analysts projecting up to 90% lower power bills for AI inference once the tech scales. Cheaper electricity plus fewer cooling systems means cloud providers could slash the hourly price you pay for heavy models.
Real‑world timeline
- 2025‑26: Limited cloud access (think "photonic instances" you can rent)
- 2027‑28: PCIe accelerator cards that slot into normal servers
- 2029‑30: Dedicated light‑based AI appliances for edge data centres and research labs
Bottom line. If "frozen light" moves from lab demo to factory production, today's GPU arms race could give way to an era where photons—not silicon transistors—do the heavy lifting. That shift could cut AI costs by an order of magnitude and wipe out talk of GPU shortages in the process.
Sources
- Supersolid "frozen light" breakthrough — Italian National Research Council, reported in Interesting Engineering
- ScienceAlert coverage of the same supersolid‑light experiment
- Xanadu press release — first on‑chip error‑resistant photonic GKP qubit
- Quantum Computing Report summary of Xanadu's GKP chip demo
- Harvard / ScienceDaily — ultra‑thin chip entangles photons at room temperature
- Nature paper — integrated photonic source of Gottesman–Kitaev–Preskill qubits
- SPIE review — thin‑film lithium‑niobate quantum photonics
- On‑chip lithium‑niobate entanglement demo (APS)