What is quantum computing used for today — and what is it projected to be used for?

As of October 6, 2026, quantum computing is used in production for research, not for business advantage: peer-reviewed physics on analog simulators, algorithm benchmarking on pre-registered suites, training on free SDKs, and one applied certified-randomness demonstration (Nature, March 2025). The industry’s more than $1 billion of 2025 revenue (McKinsey Quantum Technology Monitor 2026) comes from selling machines, cloud access and services — not from any workload with independently verified quantum advantage: that count, in Rosetta Quantum’s ledger and in the public record, is 0. The projections — IBM’s fault-tolerant Starling for 2029, up to $2.7 trillion of economic value by 2035 (McKinsey) — are declared targets, and this post will keep them in the future tense.

Status as of: October 6, 2026.

What is quantum computing actually used for today?

For research, for measurement and for training — and that is a measured description, not a dismissal. Four uses are live and documented. First, physics: analog quantum simulators are producing peer-reviewed condensed-matter results today, including quench dynamics on a 256-atom machine, the one lane where the hardware already does science that gets published and re-run (the map of which machine does what is in our analog-vs-universal post, with the Wiersema re-run caveats filed). Second, benchmarking and algorithm development: on QOBLIB, the pre-registered optimization suite of 10 problem classes and 1,264 instances (arXiv:2504.03832), the quantum side holds 2 best-known solutions against 533 classical optima and makes 0 advantage claims — numbers from our QOBLIB post, where the scoreboard lives. Third, education and workforce: the standard SDKs are open-source and everything up to the QPU costs $0, which is why training is one of the largest real uses by headcount (our programming post maps the stack and its prices). Fourth, one applied demonstration with an advantage argument behind it — certified randomness — which gets its own section below because it is the only entry of its kind.

Two adjacent quantum branches do sell working product today, and the map is dishonest without them: quantum sensing invoiced about $470M in 2025 (our sensing post), and QKD networks are deployed — with 0 of 6 security agencies recommending QKD as a primary control (our QKD post). Computing is the branch where revenue is real but verified advantage is not.

What does the industry’s $1 billion actually buy?

Machines, cloud access, services and readiness — not solved business problems. The McKinsey Quantum Technology Monitor 2026 reports that quantum computing companies generated “more than $1 billion in revenue worldwide in 2025”, that “over 300 global companies are adopting” the technology, that a third of them allocated more than $10 million to it in 2025, and that start-up investment reached $12.6 billion in 2025 — 6.3 times the 2024 figure, with 90 percent going to computing. The same report projects revenue of as much as $4.4 billion by 2028; that number is a projection and will stay one until it is booked. What the corporate budgets buy in practice is pilots, talent and option value: how to run that evaluation without burning the budget is a post of its own, and the team it takes is another. None of this revenue is evidence of quantum advantage. It is evidence of demand for readiness — a different, measurable thing.

What is projected — and by whom?

Fault tolerance by 2029, according to IBM; up to $2.7 trillion of economic value by 2035, according to McKinsey’s model; chemistry and materials first, according to most of the analyst literature. The concrete vendor commitment: on June 10, 2025 IBM declared Quantum Starling — approximately 200 logical qubits running 100 million quantum operations — for 2029 in Poughkeepsie, with Blue Jay (on the order of 2,000 logical qubits) declared for 2033. Those will remain declarations until the machines run; the physical-to-logical arithmetic that converts such roadmaps into dates is in our logical-qubit post. On the analyst side, McKinsey’s 2026 monitor models that quantum computing “could create up to $2.7 trillion of economic value worldwide by 2035” and sizes the 2035 computing market at $43–71 billion — conditional modeling, not bookings. And one projected use carries a regulatory calendar rather than a promise: cryptanalysis. NIST’s draft timeline deprecates RSA-2048-class cryptography by 2030 and disallows it after 2035 (NIST IR 8547), which is why the migration is scheduled while the machine that would do the breaking does not exist — our RSA-2048 post keeps that arithmetic.

The full map, with one source per row:

Which uses have independently verified advantage behind them today?

On commercial workloads, none; on synthetic benchmarks, a short and contested list. The first peer-reviewed history of the field (LaRose, Quantum 10, 2198, September 2026) classifies Google’s 2019 supremacy claim as refuted — the seven-year erosion is charted in our retrospective — while later random-circuit claims stand unrefuted, which means not yet beaten, not verified useful. The one applied entry is certified randomness: 71,313 bits of entropy certified on Quantinuum’s 56-qubit H2, published in Nature on March 26, 2025, with the certification itself consuming 1.1 ExaFLOPS of classical supercomputing — and Scott Aaronson, who invented the protocol, calling it “a first step toward using quantum computers to generate certified random bits for actual cryptographic applications”. IBM’s July 2026 trio of advantage claims (arXiv:2607.25998, arXiv:2607.24937, arXiv:2607.25941) is the newest entry and already under classical counterattack — the first of them is tracked in our doped-Clifford post, and Rosetta will re-measure the trio as the classical responses land, in the margin unit adopted on October 5. Rosetta Quantum’s own counter is unchanged: advantaged solves on business workloads in the public record, 0 — and when that number moves, the ledger will print the ratio, not just the medal.

What we don’t know

We don’t know whether any of the 300-plus corporate programs McKinsey counts has an internal workload where a quantum machine already beats that company’s best classical baseline — companies do not publish their baselines, so the public zero could undercount private wins; it could equally be counting readiness spend as progress. We don’t know whether IBM’s 2026 advantage trio will survive: on the Algorithmiq case, about half of the classical methods tried could not match the quantum results — and disagreed with each other (The Next Platform, 31-Jul-2026), which is evidence of a hard instance, not yet of a settled verdict. We don’t know whether the 2029 Starling date will hold; Rosetta keeps no private forecast and will measure the machine when it runs. And we don’t know how the $1B+ of 2025 revenue splits between government or research customers and commercial ones — the public figure does not break it out, and the split is exactly what a decision-maker would want.

Sources

• McKinsey Quantum Technology Monitor 2026 — A commercial tipping point
• Certified randomness on Quantinuum H2 (Nature, 26-Mar-2025) — The Quantum Insider
• JPMorganChase — Certified randomness
• IBM fault-tolerant roadmap, announced 10-Jun-2025 (summary)
• The Next Platform — IBM: three demonstrations prove quantum advantage (31-Jul-2026)
• LaRose — A brief history of quantum vs classical computational advantage (Quantum 10, 2198, 2026)
• QOBLIB — Quantum Optimization Benchmarking Library (arXiv:2504.03832)
• Sanders et al. — Compilation of fault-tolerant quantum heuristics (PRX Quantum 1, 020312, 2020)
• Wiersema et al. — arXiv:2609.01719 (2026)
• NIST IR 8547 — Transition to Post-Quantum Cryptography Standards
• Rosetta Quantum — What is an “advantaged solve” and why does it matter?
• Rosetta Quantum — By how much does the winner win?
• Rosetta Quantum — Are quantum sensors the quantum industry that already sells today?
• Rosetta Quantum — What is QKD and does your company need it?
• Rosetta Quantum — How do you program a quantum computer today?
• Rosetta Quantum — Analog quantum simulator vs universal quantum computer
• Rosetta Quantum — What is QOBLIB?
• Rosetta Quantum — Google’s 2019 supremacy claim, seven years on

Rosetta Quantum publishes verdicts with raw, reproducible data. This is educational content, not a product claim.