# How quantum computing could transform IT: what changes, and when

> Quantum computers should help with chemistry and materials, not everyday IT. What the evidence shows, vendor roadmaps and why post-quantum crypto starts now.

- URL: https://computese.com/how-quantum-computing-could-transform-it/
- Author: Duong Quan Nguyen, CEO, Computese
- Published: 2024-07-24
- Updated: 2026-09-25
- Topics: Cloud, Security

## In short
- Quantum computers will not replace servers, databases or AI clusters. The strongest case is simulating molecules and materials; speedups for optimization and machine learning are unproven, and data-heavy work stays classical.
- The major quantum advantage claims so far have been on scientific benchmarks, and several were later matched by classical simulations. The July 2026 results from IBM and partners are also physics problems, not business workloads.
- IBM targets its first fault-tolerant machine for 2029 and Quantinuum by 2030; Google expects commercially relevant machines by then too. These are goals: DARPA is still testing whether any approach can deliver more value than it costs by 2033.
- The action for every IT team is post-quantum cryptography. NIST published ML-KEM, ML-DSA and SLH-DSA in August 2024, and its draft transition plan deprecates RSA and elliptic-curve keys after 2030 and disallows them after 2035.
- You can experiment for little or nothing through Amazon Braket, Azure Quantum or IBM Quantum Platform, but for most organizations a cryptography inventory is worth more than a quantum pilot.

Quantum computing will not replace the servers, databases or AI clusters your business runs. It is expected to help with a narrow class of problems, mainly simulating molecules and materials, once error-corrected machines arrive; vendors target the first around 2029 to 2030. The change that reaches every IT team first is cryptographic: replacing RSA and elliptic-curve keys.

This guide is written for IT leaders, not physicists. After one short section on what a quantum computer does differently (the full explainer is [what a quantum computer is, and whether it is really faster](https://computese.com/quantum-computer-a-new-faster-type-of-computing/)), it covers which problems quantum computers are expected to help with and how strong the evidence is, what they will not replace, where each vendor's roadmap stands as of September 2026, how to experiment through cloud services, and the post-quantum cryptography work to start now. For how circuits, gates and error correction work, see our explainer on [quantum circuit design](https://computese.com/advancements-in-quantum-circuit-design/).

## What a quantum computer does differently

A classical bit is 0 or 1. A [qubit can be in a superposition](https://learn.microsoft.com/en-us/azure/quantum/overview-understanding-quantum-computing) of both, and several qubits can be entangled, so that their state can only be described together. A quantum algorithm uses entanglement and interference to steer that combined state toward the answer. Measuring it returns one ordinary result, so a program runs many times, each run called a shot, and the output is a count of results rather than a single number.

Describing a quantum system on a classical computer gets exponentially harder as the system grows. Microsoft's documentation gives the example of particles that can sit in 40 locations: storing their quantum state takes 2⁴⁰ numbers, more than 130 GB, and every added location doubles that. A machine built from quantum parts represents such states natively. That is why simulating nature, meaning molecules, materials and physics, is the application with the firmest theory behind it.

Two limits shape everything else in this guide:

- **Speedups exist only for specific algorithms.** A quantum computer is not a faster general-purpose processor. Each useful speedup comes from an algorithm built for one kind of problem.
- **Today's qubits are noisy.** In 2018 John Preskill named the current stage [NISQ, noisy intermediate-scale quantum](https://arxiv.org/abs/1801.00862): machines whose gate errors limit how large a circuit they can run reliably. The fix is error correction, which combines many physical qubits into one logical qubit; Google describes each logical qubit as [made of hundreds of physical qubits](https://research.google/blog/safeguarding-cryptocurrency-by-disclosing-quantum-vulnerabilities-responsibly/). A machine that runs long algorithms on logical qubits is called a fault-tolerant quantum computer. In December 2024 Google showed its Willow chip [halving the logical error rate each time the code grew](https://blog.google/innovation-and-ai/technology/research/google-willow-quantum-chip/), which it called the most convincing prototype of a scalable logical qubit to date.

## Which problems quantum computers are expected to help with

A useful filter comes from a 2023 analysis by [Hoefler, Haener and Troyer in _Communications of the ACM_](https://arxiv.org/abs/2307.00523). A future error-corrected quantum computer will run each operation far more slowly than a classical chip, so they argue a problem needs two properties before quantum hardware can win. It needs a speedup better than quadratic, because a Grover-style square-root saving is swallowed by the overhead, and it needs small inputs, because loading classical data into qubits is slow. Their summary: quantum computers will be practical for big compute on small data, not for big data. Their shortlist puts chemistry and materials science first, along with breaking public-key cryptography with Shor's algorithm.

| Problem area                                   | Expected benefit                                                         | Evidence as of September 2026                                                     |
| ---------------------------------------------- | ------------------------------------------------------------------------ | --------------------------------------------------------------------------------- |
| Chemistry and materials simulation             | Exponential speedups for some quantum systems                            | The strongest theoretical case; no demonstrated business result yet               |
| Optimization (routing, scheduling, portfolios) | Unclear: speedups shown only in theory, for specific structured problems | An open research question; not part of the 2026 advantage demonstrations          |
| Sampling and physics experiments               | Beyond-classical benchmarks                                              | Demonstrated repeatedly since 2019, with several claims later matched classically |
| Machine learning on large datasets             | Mostly quadratic speedups                                                | Unlikely: data loading and error-correction overhead cancel the gain              |
| Breaking RSA and elliptic-curve cryptography   | Exponential (Shor's algorithm)                                           | Needs a large fault-tolerant machine; resource estimates keep falling             |

**Chemistry and materials.** Molecules are quantum systems, so this is the natural fit. The evidence, though, is more cautious than the marketing. A [2023 study in _Nature Communications_](https://www.nature.com/articles/s41467-023-37587-6) examined ground-state energy estimation, the most common task in quantum chemistry, and concluded that evidence for an exponential quantum advantage across chemical space has yet to be found, while polynomial speedups may still make quantum computers useful there. In October 2025 Google ran a [proof-of-principle experiment](https://blog.google/innovation-and-ai/technology/research/quantum-echoes-willow-verifiable-quantum-advantage/) on two molecules of 15 and 28 atoms whose results matched traditional NMR measurements: a direction for drug and materials research, separate from the benchmark on which Google claimed its speed advantage.

**Optimization.** This is the area where the evidence is least settled. Google's quantum team wrote in November 2025 that whether quantum machines can [succeed on optimization problems where classical ones fall short](https://research.google/blog/a-new-quantum-toolkit-for-optimization/) remains largely open. Its new algorithm, Decoded Quantum Interferometry, published in _Nature_, offers a speedup for specific structured problems on a sufficiently large error-corrected machine. IBM noted in August 2026 that optimization [was not a focus of its July advantage demonstrations](https://www.ibm.com/quantum/blog/qoblib), and it tracks candidates in an open benchmarking library, QOBLIB.

**Machine learning and simulation of fluids.** Hoefler and colleagues list many current machine learning training approaches, Grover-based drug design, turbulent fluid dynamics, and [weather and climate simulation](https://computese.com/ai-for-climate-research/) among the areas unlikely to see a quantum advantage with current algorithms in the foreseeable future.

## Quantum advantage claims, and how they held up

Quantum advantage means a quantum computer completing a task beyond what classical computing can do alone, with a result that can be validated, which is how [IBM defined it in July 2026](https://www.ibm.com/quantum/blog/quantum-advantage). Every major claim so far has been on a scientific benchmark, and classical algorithms have kept improving in response:

| Date          | Claim                                                                                                                                                                                                                                                                               | What happened next                                                                                                                                                                             |
| ------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| October 2019  | Google's 53-qubit Sycamore [sampled random circuits in about 200 seconds](https://www.nature.com/articles/s41586-019-1666-5), a task estimated at 10,000 years on a supercomputer (_Nature_)                                                                                        | A classical tensor-network method [produced the samples in about 15 hours on 512 GPUs](https://arxiv.org/abs/2111.03011) (_Physical Review Letters_, 2022)                                     |
| June 2023     | IBM's 127-qubit processor measured values [beyond brute-force classical simulation](https://www.nature.com/articles/s41586-023-06096-3) (_Nature_)                                                                                                                                  | Within two weeks a [classical simulation more accurate than the quantum results](https://arxiv.org/abs/2306.14887) was posted, later published in _PRX Quantum_                                |
| December 2024 | Google's 105-qubit Willow ran the same kind of sampling benchmark in under five minutes, against an estimated 10²⁵ years on a supercomputer                                                                                                                                         | Google itself says random circuit sampling has [no known real-world application](https://blog.google/innovation-and-ai/technology/research/google-willow-quantum-chip/)                        |
| 2025          | D-Wave's quantum annealer simulated spin-glass dynamics [beyond classical methods](https://arxiv.org/abs/2403.00910) (_Science_)                                                                                                                                                    | Two groups posted classical simulations of comparable accuracy as preprints ([tensor networks](https://arxiv.org/abs/2503.05693), [variational Monte Carlo](https://arxiv.org/abs/2503.08247)) |
| October 2025  | Google's Quantum Echoes measured a verifiable quantum effect [13,000 times faster](https://blog.google/innovation-and-ai/technology/research/quantum-echoes-willow-verifiable-quantum-advantage/) than the best classical algorithm on one of the fastest supercomputers (_Nature_) | The [paper itself](https://www.nature.com/articles/s41586-025-09526-6) calls its learning demonstration a toy model of practically relevant problems                                           |
| July 2026     | IBM and partners reported [three advantage demonstrations](https://www.ibm.com/quantum/blog/quantum-advantage): a 70-logical-qubit sampling task and two physics simulations of 56 and up to 74 qubits                                                                              | Submitted to the open Quantum Advantage Tracker, where classical groups try to match them                                                                                                      |

The pattern matters more than any single row. Benchmarks are chosen to be hard for classical computers, classical researchers then look for shortcuts, and the frontier moves, as Google itself [expected when it announced Willow](https://blog.google/innovation-and-ai/technology/research/google-willow-quantum-chip/). IBM's own post makes the same point: announcing an advantage opens a result to scrutiny rather than closing the case. None of these results is a business workload. For an IT leader, an advantage headline is a milestone for the field, not a reason to buy anything.

## What quantum computers will not replace

Most of an IT estate moves and stores data: web and API traffic, transactions, databases, analytics, file storage, and AI training and inference. That is exactly what quantum computers are worst at. [Hoefler and colleagues](https://arxiv.org/abs/2307.00523) compared one classical chip of today with an optimistic future quantum computer of 10,000 error-corrected logical qubits. The quantum machine's input and output bandwidth came out at about 1 Gbit/s against 10,000 Gbit/s for the classical chip, and its 16-bit floating-point throughput at 10.5 thousand operations per second against 195 trillion. Their conclusion is that any problem limited by access to classical data, such as searching a database, will be solved faster by classical computers.

![A tall stack of databases feeds a wide pipe into a rack of classical servers and only a thin orange line into a small quantum processor, which works on a small molecule.](https://computese.com/images/blog/how-quantum-computing-could-transform-it/data-bottleneck.da8390440e-1536.webp)

*A quantum computer can only take in data slowly, so it pays off on small inputs that need enormous computation, not on big data.*

Three consequences for planning:

- **It is an accelerator, not a server.** In hybrid quantum computing, [a classical computer and a quantum computer work together](https://learn.microsoft.com/en-us/azure/quantum/overview-azure-quantum) on one problem: the classical side prepares jobs, sends them to the quantum processor and processes the results. Some qubit technologies must be kept [near absolute zero](https://learn.microsoft.com/en-us/azure/quantum/overview-understanding-quantum-computing), so most organizations reach the hardware through the cloud; IBM also offers a dedicated on-premises system, priced by quote.
- **Nothing in a typical estate moves to it.** The workloads that fit are research computations in chemistry, materials and physics, not the systems IT runs today, so there is no capacity, licence or data center decision to make for quantum today.
- **Classical methods keep improving.** The tensor-network and Monte Carlo techniques that answered the advantage claims above run on ordinary clusters and GPUs, and they set the bar any quantum pilot has to clear.

## Where the hardware is: vendor roadmaps as of September 2026

The dates below are the vendors' own targets, taken from their pages. IBM, for example, states that its roadmap represents only goals and objectives and is subject to change. Treat them as intentions, and weigh each milestone by whether the vendor met its previous ones.

| Vendor            | Where it is                                                                                                                                                                                                                                                                                            | What it targets                                                                                                                                                                                                                                                                                                                                                                                          |
| ----------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| IBM               | Nighthawk r2: [120 qubits, accurate results on circuits of 7,500+ gates](https://www.ibm.com/quantum/blog/nighthawk-r2) (August 2026)                                                                                                                                                                  | [Roadmap](https://www.ibm.com/roadmaps/quantum/): Starling, IBM's first large-scale fault-tolerant system, in 2029, running [100 million operations on 200 logical qubits](https://newsroom.ibm.com/2025-06-10-IBM-Sets-the-Course-to-Build-Worlds-First-Large-Scale,-Fault-Tolerant-Quantum-Computer-at-New-IBM-Quantum-Data-Center); Blue Jay, 1 billion operations on 2,000 logical qubits, from 2033 |
| Google Quantum AI | Willow: 105 qubits and error correction below threshold (December 2024); Quantum Echoes (October 2025)                                                                                                                                                                                                 | [Increasingly confident](https://blog.google/innovation-and-ai/technology/research/neutral-atom-quantum-computers/) that commercially relevant superconducting quantum computers arrive by the end of this decade (March 2026); now also building neutral-atom machines                                                                                                                                  |
| Quantinuum        | Helios: [98 physical qubits, and 48 error-corrected logical qubits](https://www.quantinuum.com/press-releases/quantinuum-announces-commercial-launch-of-new-helios-quantum-computer-that-offers-unprecedented-accuracy-to-enable-generative-quantum-ai-genqai) that beat physical ones (November 2025) | [Apollo](https://www.quantinuum.com/blog/technical-perspective-by-the-end-of-the-decade-we-will-deliver-universal-fault-tolerant-quantum-computing), a universal fault-tolerant machine with hundreds of logical qubits, by the end of the decade (roadmap of September 2024)                                                                                                                            |
| Microsoft         | Majorana 1: [eight topological qubits](https://azure.microsoft.com/en-us/blog/quantum/2025/02/19/microsoft-unveils-majorana-1-the-worlds-first-quantum-processor-powered-by-topological-qubits/) on a chip designed to hold a million (February 2025)                                                  | A fault-tolerant prototype "in years, not decades", in the final phase of DARPA's US2QC program                                                                                                                                                                                                                                                                                                          |
| IonQ              | Trapped-ion processors offered through Amazon Braket and Azure Quantum                                                                                                                                                                                                                                 | [Roadmap](https://www.ionq.com/roadmap): 10,000 physical and 800 logical qubits in 2027; 2 million physical and 80,000 logical qubits by 2030                                                                                                                                                                                                                                                            |

Two caveats before you plan around any of these:

- **Microsoft's topological approach carries a caveat from its own paper.** The [_Nature_ paper](https://www.nature.com/articles/s41586-024-08445-2) published alongside Majorana 1 states that its measurement alone does not distinguish the topological states Microsoft needs from ordinary states that can mimic them.
- **There is an independent check.** DARPA's [Quantum Benchmarking Initiative](https://www.darpa.mil/research/programs/quantum-benchmarking-initiative) is testing whether any approach can reach utility scale by 2033, meaning a machine whose computational value exceeds its cost. It verifies vendors' designs rather than forecasting, which makes it a useful neutral yardstick.

## How to experiment with quantum computing today

You do not need hardware of your own. A typical experiment is a loop: code on a laptop or in a notebook builds a small circuit, submits it as a task of many shots, receives counts back, and uses a classical optimizer to adjust the next circuit. Amazon Braket runs a managed version of this loop as [Hybrid Jobs](https://aws.amazon.com/braket/pricing/), with PennyLane, NVIDIA CUDA-Q or its own SDK.

![A laptop sends a small circuit through the cloud to a queue and an orange quantum processor; counts come back as a bar histogram, and a gear on the laptop adjusts the next circuit.](https://computese.com/images/blog/how-quantum-computing-could-transform-it/hybrid-loop.3d4ed02aea-1536.webp)

*Most of the work in a quantum experiment is classical: the processor runs short circuits and returns counts.*

Three cloud services give access to real quantum processors and simulators, billed per use:

| Service                                                      | Hardware (September 2026)                                                 | How you pay                                                                                                                                                         | Tools                                                                    |
| ------------------------------------------------------------ | ------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------ |
| Amazon Braket                                                | AQT, IonQ, IQM, QuEra and Rigetti processors; SV1, DM1 and TN1 simulators | $0.30 per task plus a per-shot fee from $0.000425 (Rigetti) to $0.08 (IonQ Forte); hourly reservations; one free hour of simulation a month for the first 12 months | Braket SDK, PennyLane, CUDA-Q                                            |
| Azure Quantum                                                | IonQ, Pasqal, Quantinuum and Rigetti                                      | An Azure subscription and workspace; [each provider sets its own prices](https://learn.microsoft.com/en-us/azure/quantum/pricing)                                   | Quantum Development Kit (Q#, Qiskit, Cirq, OpenQASM), resource estimator |
| [IBM Quantum Platform](https://www.ibm.com/quantum/products) | IBM processors of 100+ qubits                                             | Open Plan: up to 10 minutes a month free; Pay-As-You-Go from $96 a minute                                                                                           | Qiskit, Qiskit Functions                                                 |

Costs climb quickly once error mitigation is on. On Braket, a single IonQ Forte task with error mitigation needs at least 2,500 shots and costs $200.30. On Azure, the minimum charge for one IonQ Forte program with error mitigation on, the default, is about $168.

A sensible first project:

1. **Pick a candidate that fits:** a small-data, compute-heavy kernel such as a molecule, a material or a small combinatorial core, not a data-heavy workload.
2. **Establish the best classical baseline first,** with the same care you give the quantum side.
3. **Prototype on a simulator.** The Braket SDK includes a free local simulator, and Microsoft's Quantum Development Kit is free, open source and needs no Azure account.
4. **Run a few small jobs on hardware** with a hard budget and a fixed number of shots.
5. **Compare honestly** with the baseline and record the cost per useful result.
6. **Estimate the full-scale version.** The Azure Quantum resource estimator calculates the physical resources an algorithm would need on future fault-tolerant hardware, which tells you how far away your use case really is.

Run experiments in a separate sandbox account or subscription with budgets, alerts and least-privilege access, so a loop of hardware tasks cannot run up a bill unnoticed. If your cloud accounts lack that structure, our [cloud transformation](https://computese.com/services/cloud-transformation/) service builds a governed landing zone, with SSO, least-privilege roles, tagging, budgets and guardrails from the first day.

## The one action for every IT team now: post-quantum cryptography

Quantum computing's first effect on most IT teams is defensive. This section is the summary; the full picture, with every deadline, what already runs post-quantum and a step-by-step migration plan, is in our guide to [quantum computing in cybersecurity](https://computese.com/quantum-computing-in-cybersecurity-breakthrough/). [Shor's algorithm](https://learn.microsoft.com/en-us/azure/quantum/overview-understanding-quantum-computing) factors integers exponentially faster than the best known classical method, which would let a large fault-tolerant quantum computer break RSA and elliptic-curve cryptography: the public-key algorithms behind TLS, VPNs, SSH, code signing and email encryption.

The estimates of what that takes keep falling. In 2019 Craig Gidney and Martin Ekerå estimated 20 million noisy qubits running for eight hours to factor a 2048-bit RSA key; in a [2025 preprint](https://arxiv.org/abs/2505.15917) Gidney cut that to under a million noisy qubits running for under a week. In March 2026 Google estimated that [256-bit elliptic-curve keys could be broken](https://research.google/blog/safeguarding-cryptocurrency-by-disclosing-quantum-vulnerabilities-responsibly/) with fewer than 500,000 physical qubits in a few minutes, about 20 times fewer physical qubits than before. IBM's, Google's and Quantinuum's flagship processors have 98 to 120 physical qubits, and neutral-atom arrays of about 10,000 qubits have yet to [run deep circuits](https://blog.google/innovation-and-ai/technology/research/neutral-atom-quantum-computers/). NIST describes a machine that can break today's public-key cryptography as possibly [years or decades away](https://csrc.nist.gov/projects/post-quantum-cryptography).

The problem is timing. In a harvest now, decrypt later attack, adversaries [collect encrypted data now](https://csrc.nist.gov/pubs/ir/8547/ipd) to decrypt it once the hardware exists, so anything that must stay confidential for years is already exposed. NIST's draft transition plan frames it as simple arithmetic: if data must stay secret for X years and migration takes Y years, you must start before X plus Y exceeds the years until such a machine exists. Canada's Cyber Centre tells federal departments to make systems that carry confidential data over public networks a [high priority for migration](https://www.cyber.gc.ca/en/guidance/roadmap-migration-post-quantum-cryptography-government-canada-itsm40001).

![A laptop and a server exchange padlocked envelopes while a tap copies them into an orange storage cabinet; after an hourglass, a quantum computer opens a stored padlock.](https://computese.com/images/blog/how-quantum-computing-could-transform-it/harvest.bd2ba2fedb-1536.webp)

*Traffic recorded today can be decrypted when the hardware exists, so the deadline for long-lived secrets comes before the machine does.*

The standards are ready. In August 2024 NIST published [FIPS 203, 204 and 205](https://www.nist.gov/news-events/news/2024/08/nist-releases-first-3-finalized-post-quantum-encryption-standards): ML-KEM for key establishment, ML-DSA for digital signatures, and SLH-DSA as a hash-based signature backup, and it urged administrators to start integrating them immediately. In March 2025 NIST [selected HQC](https://www.nist.gov/news-events/news/2025/03/nist-selects-hqc-fifth-algorithm-post-quantum-encryption) as a backup to ML-KEM, with a final standard expected in 2027. As of its [August 2026 project update](https://csrc.nist.gov/projects/post-quantum-cryptography), the HQC and Falcon-based signature standards were still in progress.

Quantum key distribution (QKD), sometimes presented as an alternative, does not replace this work. It needs specialized hardware and does not authenticate the parties, and the UK NCSC [will not support it for government or military use](https://www.ncsc.gov.uk/paper/quantum-networking-technologies), naming post-quantum cryptography as the best mitigation.

The deadlines are set too:

| Source                                                                                                                                                                         | Milestones                                                                                                                                      |
| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------- |
| [NIST IR 8547](https://csrc.nist.gov/pubs/ir/8547/ipd), initial public draft (November 2024)                                                                                   | Quantum-vulnerable algorithms such as RSA and elliptic-curve cryptography: deprecated after 2030 at 112-bit strength, all disallowed after 2035 |
| [UK NCSC](https://www.ncsc.gov.uk/guidance/pqc-migration-timelines) (March 2025)                                                                                               | Discovery and an initial plan by 2028; highest-priority migrations by 2031; all systems by 2035                                                 |
| [Canadian Centre for Cyber Security](https://www.cyber.gc.ca/en/guidance/roadmap-migration-post-quantum-cryptography-government-canada-itsm40001) (June 2025, federal systems) | Departmental plans by April 2026; high-priority systems by the end of 2031; the rest by the end of 2035                                         |
| [Google](https://blog.google/innovation-and-ai/technology/safety-security/cryptography-migration-timeline/) (March 2026, its own systems)                                      | Migration by 2029, prioritizing authentication and digital signatures                                                                           |

What to do now, in order (each step is worked through in the [migration plan](https://computese.com/quantum-computing-in-cybersecurity-breakthrough/)):

1. **Inventory your cryptography:** where RSA and elliptic-curve keys protect TLS endpoints, VPNs, SSH, code signing, certificates and PKI, databases, hardware security modules and vendor products.
2. **Rank systems by how long their data must stay secret** and by exposure to recording, starting with confidential traffic over public networks.
3. **Ask vendors for their PQC roadmaps** and write post-quantum support into procurement.
4. **Turn on hybrid key exchange where your stack supports it.** OpenSSL 3.5, released in April 2025, [supports ML-KEM, ML-DSA and SLH-DSA](https://openssl-library.org/news/openssl-3.5-notes/) and offers the hybrid X25519MLKEM768 key share by default.
5. **Build crypto agility:** automate how keys and certificates are replaced, the same discipline our guide to [renewing an SSL certificate](https://computese.com/ssl-certificate/) covers, so that changing an algorithm becomes a configuration change rather than a project.

The internet-facing part of the inventory is the easiest place to begin. Our [security scanning](https://computese.com/services/security-testing/) service discovers your domains, subdomains, IP ranges and certificates and checks exposed services and their TLS configuration on a schedule.

> [!IMPORTANT]
> Start with the inventory. [NIST notes](https://csrc.nist.gov/pubs/ir/8547/ipd) that moving from a new algorithm standard to full integration has historically taken 10 to 20 years, and the 2035 deadline is less than ten years away.

## A realistic quantum timeline for IT teams

Standards bodies and governments set the firm dates; hardware dates are vendor goals. This table uses only the sourced milestones above.

| Horizon                         | What the sources say                                                                                                                                                                                                                                                  | What IT teams should do                                                                                                      |
| ------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------- |
| Now (2026)                      | Flagship processors of 98 to 120 physical qubits; advantage claims on physics problems; PQC standards final since August 2024; Canadian federal migration plans due April 2026                                                                                        | Inventory cryptography; enable hybrid TLS where supported; experiment in the cloud only if you have a candidate problem      |
| Next three years (2027 to 2029) | HQC standard expected in 2027; UK discovery deadline in 2028; IBM aims at circuits of up to 15,000 gates in 2028 and Starling in 2029; Google's own PQC migration by 2029                                                                                             | Finish discovery; migrate systems with long-lived secrets first; require PQC in procurement; review quantum use cases yearly |
| Later (2030 to 2035)            | Vendors expect fault-tolerant, commercially relevant machines by the end of the decade; NIST deprecation after 2030; UK and Canadian high-priority deadline 2031; DARPA's utility test by 2033; IBM Blue Jay from 2033; RSA and elliptic curves disallowed after 2035 | Complete the migration; evaluate fault-tolerant cloud services if your business has chemistry or materials problems          |

Plan against the first set of dates and track the second. A vendor milestone that slips moves your quantum pilot; it does not move your cryptography deadline.

## Key terms
- **Qubit**: The unit of quantum information. It can hold a superposition of 0 and 1; measuring it returns 0 or 1, with probabilities set by its state.
- **Entanglement**: A correlation between qubits so strong that their states can only be described together. Quantum algorithms combine it with interference to steer a computation toward the answer.
- **Quantum advantage**: A quantum computer completing a task beyond what classical computers can do alone, with a result that can be validated. The major claims so far have been on scientific benchmarks.
- **NISQ**: Noisy intermediate-scale quantum: John Preskill's 2018 name for quantum computers whose gate noise limits how large a circuit they can run reliably. He expected the first to have 50 to 100 qubits.
- **Logical qubit**: A qubit encoded across many physical qubits by error correction, so that it fails far less often than any one of them. Google describes each one as made of hundreds of physical qubits.
- **Fault-tolerant quantum computer**: A machine that runs long algorithms on logical qubits, correcting errors faster than they build up. IBM targets its first, Starling, for 2029.
- **Shor's algorithm**: Peter Shor's 1994 quantum algorithm that factors integers exponentially faster than the best known classical method. On a large enough machine it breaks RSA and elliptic-curve cryptography.
- **Post-quantum cryptography (PQC)**: Public-key algorithms that run on ordinary computers but are designed to resist quantum attacks, such as ML-KEM (FIPS 203) and ML-DSA (FIPS 204).
- **Harvest now, decrypt later**: Recording encrypted traffic or data today in order to decrypt it once a capable quantum computer exists. It makes data that must stay secret for years urgent to protect.

## Common questions

### When will quantum computers be useful for business?

For most businesses, not before fault-tolerant machines arrive. IBM targets its first for 2029, and Quantinuum and Google say by the end of the decade; DARPA is testing whether an industrially useful machine is possible by 2033. The first value is expected in chemistry and materials research, not in general IT.

### Will quantum computers replace classical computers?

No. They work as accelerators: a classical computer prepares each job, sends it to the quantum processor and processes the results. Loading data into qubits is slow, so databases, analytics, web workloads and AI training stay on classical hardware.

### Can a quantum computer break RSA encryption today?

No. The flagship processors of 2025 and 2026 have 98 to 120 physical qubits, while a 2025 estimate for breaking 2048-bit RSA assumes a machine of just under a million noisy qubits running for up to a week. The risk today is data recorded now and decrypted later, which is why the migration to post-quantum cryptography should start now.

### How can my company start experimenting with quantum computing?

Use a cloud service such as Amazon Braket, Azure Quantum or IBM Quantum Platform. Start on free simulators or IBM's Open Plan (up to 10 minutes of hardware time a month), set a budget, and compare every result with the best classical method.

### Is quantum key distribution (QKD) an alternative to post-quantum cryptography?

Not for most organizations. QKD needs specialized hardware and does not authenticate the parties, so it still depends on other cryptography. The UK NCSC will not support QKD for government or military use and names post-quantum cryptography as the best mitigation.

## Sources
1. [What is quantum computing?](https://learn.microsoft.com/en-us/azure/quantum/overview-understanding-quantum-computing), Microsoft Learn
2. [Quantum Computing in the NISQ era and beyond](https://arxiv.org/abs/1801.00862), arXiv (John Preskill, 2018)
3. [Safeguarding cryptocurrency by disclosing quantum vulnerabilities responsibly](https://research.google/blog/safeguarding-cryptocurrency-by-disclosing-quantum-vulnerabilities-responsibly/), Google Research
4. [Meet Willow, our state-of-the-art quantum chip](https://blog.google/innovation-and-ai/technology/research/google-willow-quantum-chip/), Google
5. [Disentangling Hype from Practicality: On Realistically Achieving Quantum Advantage](https://arxiv.org/abs/2307.00523), Communications of the ACM, 2023 (arXiv)
6. [Evaluating the evidence for exponential quantum advantage in ground-state quantum chemistry](https://www.nature.com/articles/s41467-023-37587-6), Nature Communications
7. [Our Quantum Echoes algorithm is a big step toward real-world applications for quantum computing](https://blog.google/innovation-and-ai/technology/research/quantum-echoes-willow-verifiable-quantum-advantage/), Google
8. [A new quantum toolkit for optimization](https://research.google/blog/a-new-quantum-toolkit-for-optimization/), Google Research
9. [QOBLIB: tracking progress in quantum optimization](https://www.ibm.com/quantum/blog/qoblib), IBM Quantum
10. [Researchers demonstrate quantum advantage through trusted quantum computation](https://www.ibm.com/quantum/blog/quantum-advantage), IBM Quantum
11. [Quantum supremacy using a programmable superconducting processor](https://www.nature.com/articles/s41586-019-1666-5), Nature (Arute et al., 2019)
12. [Solving the sampling problem of the Sycamore quantum circuits](https://arxiv.org/abs/2111.03011), Physical Review Letters, 2022 (arXiv)
13. [Evidence for the utility of quantum computing before fault tolerance](https://www.nature.com/articles/s41586-023-06096-3), Nature (Kim et al., 2023)
14. [Efficient tensor network simulation of IBM's Eagle kicked Ising experiment](https://arxiv.org/abs/2306.14887), PRX Quantum, 2024 (arXiv)
15. [Beyond-classical computation in quantum simulation](https://arxiv.org/abs/2403.00910), Science, 2025 (arXiv)
16. [Dynamics of disordered quantum systems with two- and three-dimensional tensor networks](https://arxiv.org/abs/2503.05693), arXiv preprint (2025)
17. [Challenging the Quantum Advantage Frontier with Large-Scale Classical Simulations of Annealing Dynamics](https://arxiv.org/abs/2503.08247), arXiv preprint (2025)
18. [Observation of constructive interference at the edge of quantum ergodicity](https://www.nature.com/articles/s41586-025-09526-6), Nature (Google Quantum AI and collaborators, 2025)
19. [What is Azure Quantum?](https://learn.microsoft.com/en-us/azure/quantum/overview-azure-quantum), Microsoft Learn
20. [IBM Quantum Nighthawk r2: more circuits, faster](https://www.ibm.com/quantum/blog/nighthawk-r2), IBM Quantum
21. [Quantum Roadmap](https://www.ibm.com/roadmaps/quantum/), IBM
22. [IBM Sets the Course to Build World's First Large-Scale, Fault-Tolerant Quantum Computer at New IBM Quantum Data Center](https://newsroom.ibm.com/2025-06-10-IBM-Sets-the-Course-to-Build-Worlds-First-Large-Scale,-Fault-Tolerant-Quantum-Computer-at-New-IBM-Quantum-Data-Center), IBM Newsroom
23. [Google Quantum AI to include neutral atom computing](https://blog.google/innovation-and-ai/technology/research/neutral-atom-quantum-computers/), Google
24. [Quantinuum announces commercial launch of new Helios quantum computer](https://www.quantinuum.com/press-releases/quantinuum-announces-commercial-launch-of-new-helios-quantum-computer-that-offers-unprecedented-accuracy-to-enable-generative-quantum-ai-genqai), Quantinuum
25. [Technical perspective: By the end of the decade, we will deliver universal, fully fault-tolerant quantum computing](https://www.quantinuum.com/blog/technical-perspective-by-the-end-of-the-decade-we-will-deliver-universal-fault-tolerant-quantum-computing), Quantinuum
26. [Microsoft unveils Majorana 1, the world's first quantum processor powered by topological qubits](https://azure.microsoft.com/en-us/blog/quantum/2025/02/19/microsoft-unveils-majorana-1-the-worlds-first-quantum-processor-powered-by-topological-qubits/), Microsoft Azure
27. [Interferometric single-shot parity measurement in InAs-Al hybrid devices](https://www.nature.com/articles/s41586-024-08445-2), Nature (Microsoft Azure Quantum, 2025)
28. [IonQ roadmap](https://www.ionq.com/roadmap), IonQ
29. [Quantum Benchmarking Initiative](https://www.darpa.mil/research/programs/quantum-benchmarking-initiative), DARPA
30. [Amazon Braket pricing](https://aws.amazon.com/braket/pricing/), Amazon Web Services
31. [Pricing plans for Azure Quantum providers](https://learn.microsoft.com/en-us/azure/quantum/pricing), Microsoft Learn
32. [IBM Quantum Computing: products and services](https://www.ibm.com/quantum/products), IBM Quantum
33. [How to factor 2048 bit RSA integers with less than a million noisy qubits](https://arxiv.org/abs/2505.15917), arXiv preprint (Craig Gidney, 2025)
34. [NIST IR 8547 (Initial Public Draft): Transition to Post-Quantum Cryptography Standards](https://csrc.nist.gov/pubs/ir/8547/ipd), NIST
35. [Roadmap for the migration to post-quantum cryptography for the Government of Canada (ITSM.40.001)](https://www.cyber.gc.ca/en/guidance/roadmap-migration-post-quantum-cryptography-government-canada-itsm40001), Canadian Centre for Cyber Security
36. [NIST Releases First 3 Finalized Post-Quantum Encryption Standards](https://www.nist.gov/news-events/news/2024/08/nist-releases-first-3-finalized-post-quantum-encryption-standards), NIST
37. [NIST Selects HQC as Fifth Algorithm for Post-Quantum Encryption](https://www.nist.gov/news-events/news/2025/03/nist-selects-hqc-fifth-algorithm-post-quantum-encryption), NIST
38. [Post-Quantum Cryptography project](https://csrc.nist.gov/projects/post-quantum-cryptography), NIST Computer Security Resource Center
39. [Timelines for migration to post-quantum cryptography](https://www.ncsc.gov.uk/guidance/pqc-migration-timelines), UK National Cyber Security Centre
40. [Quantum frontiers may be closer than they appear](https://blog.google/innovation-and-ai/technology/safety-security/cryptography-migration-timeline/), Google
41. [OpenSSL 3.5 release notes](https://openssl-library.org/news/openssl-3.5-notes/), OpenSSL Library
42. [Quantum networking technologies](https://www.ncsc.gov.uk/paper/quantum-networking-technologies), UK National Cyber Security Centre