Quantum Powerhouse

Research · applications

What quantum computing is actually good for

Not “everything, faster.” Quantum computers change the cost of specific structured problems— and leave most computation exactly where it is. This map sorts the claims by evidence, each with a primary source you can open. Same rule as the rest of the site: nothing here is asserted that can't be traced.

Quantum or not? — call it

1/8 · score 0/0

“Quantum computers do ordinary arithmetic faster than classical ones.”

Real — evidence exists

Chemistry & materials simulation

Nature is quantum, so simulating strongly-correlated electrons — catalysts, batteries, magnets, drug targets — is the application with the deepest physics behind it. In May 2026 Q-CTRL reported a ~3,000× speedup over performance-optimized classical software on a commercially relevant materials problem on IBM hardware, the first credible 'practical advantage' claim; IBM hardware has also simulated a 300-atom pharmaceutical system. This site's own VQE solves H₂ in your browser — the same idea at toy scale.

source: Q-CTRL, May 2026 ↗

Breaking public-key cryptography — and replacing it

Shor's algorithm factors and takes discrete logs efficiently; a large fault-tolerant machine would break RSA and elliptic-curve cryptography. The practical consequence is already here: NIST finalized post-quantum standards in 2024 (ML-KEM, ML-DSA, SLH-DSA) and migrations are underway. Quantum key distribution (BB84 — play it in the arcade) is deployed today on real fiber links.

source: Assessing the benefits and risks (arXiv 2401.16317) ↗

Promising — unproven at scale

Finance: Monte Carlo via amplitude estimation

A genuine quadratic (square-root) speedup for derivative pricing and risk — but fragile: estimates put ~4,700 logical qubits and ~10⁹ T-gates at 45 MHz just to match classical Monte Carlo, roughly 1,000× today's clock rates, while GPUs keep eroding the edge. Banks are piloting; nobody is in production.

source: Quantum computing by 2033 — industry map ↗

Optimization: logistics, scheduling, engineering design

QAOA and annealing are real, and hybrid pilots exist for routing and scheduling. But careful evaluations on prototypical industrial problems — robot path planning, vehicle options — find no clear advantage yet over strong classical heuristics. Worth watching; not worth promising.

source: Practicality of quantum optimization (arXiv 2311.11621) ↗

Quantum machine learning

The hype runs ahead of the evidence: loading classical data into qubits is inefficient, most results are on small or simulated datasets, and even Microsoft places AI-workload advantage 5–10 years out. The opposite direction is already paying: classical ML accelerates quantum simulation.

source: QML survey (arXiv 2310.10315) ↗

Not this — and why

General-purpose speed

Quantum computers are not faster computers. For a broad class of problems they offer at most a polynomial speedup, and for most everyday computation none. They cannot solve anything a classical machine cannot; they change the cost of specific structured problems. Most researchers believe NP-complete problems stay hard.

source: The Quantum Frontier (arXiv 1206.0785) ↗

Video and image generation

Quantum generative models exist as research on MNIST-scale, low-resolution images; hybrid models need the classical half to reach anything usable. Quantum video processing is at the 'basic research is just beginning' stage. GPUs own this for the foreseeable future.

source: Quantum GANs survey (arXiv 2506.18002) ↗

Robots, construction, 'automating everything'

Robots run on classical control and classical AI. Quantum's plausible role is indirect — an optimization sub-routine for path planning or design search. Claims of quantum-driven seismic optimization on construction sites are preprints, not peer-reviewed results. The intelligence people feel from automation comes from AI; quantum doesn't add 'more intellect', it adds specific math.

source: Quantum computing & AI automation perspectives (arXiv 2505.10012) ↗

The honest one-liner

Quantum computing is a simulation and number-theory machine with a cryptography deadline attached. Chemistry, materials, and the post-quantum migration are the bets with evidence; finance, optimization, and ML are bets with promise; general speed, media generation, and robotics are not quantum problems. If you want to feel the difference rather than read it, the arcade runs the real algorithms, and the research wing shows how claims get checked. For the cryptography deadline in practice, see pqc-scan, a tool that inventories quantum-vulnerable cryptography in a codebase.