Quantum computing is a path to energy-efficient AI (www.fastcompany.com)

As AI workloads drive a steep rise in data‑center electricity demand, the AI industry is exploring quantum computing—especially annealing quantum processors—as a path to far more energy‑efficient computation. Proponents argue quantum systems can search complex optimization landscapes much faster than classical hardware, making them well suited to problems that underlie AI infrastructure: grid planning, optimal data‑center siting, hard combinatorial optimization, and certain scientific simulations. Early real‑world uses include GE Vernova and E.ON applying quantum annealing to grid stability and vulnerability analysis, and Forschungszentrum Jülich integrating an annealer with its JUNIQ infrastructure to pair with the JUPITER exascale system. Technical results suggest substantial energy and time savings on niche but costly problems: D‑Wave’s annealer reportedly solved a magnetic‑materials simulation in minutes using ~12 kW—an equivalent classical GPU supercomputer would need nearly a million years and enormous power. Hybrid quantum‑classical workflows have improved drug‑discovery molecule generation (Japan Tobacco) and sped complex particle‑collision simulations (TRIUMF), while blockchain and hashing scenarios claim potential electricity reductions up to 1,000×. The takeaway for AI/ML practitioners: annealing quantum devices are emerging as complementary accelerators for optimization‑heavy tasks, and tools (e.g., PyTorch plug‑ins) are lowering integration friction—offering a pragmatic route to reduce AI’s growing energy footprint before full‑scale fault‑tolerant quantum computers arrive.