Chest Forge
STEAM PC / 2026.08.18

9 Best Fault-Tolerant Quantum Computing Providers for Enterprises

Enterprise technology leaders face an awkward planning problem with quantum computing. The machines expected to matter commercially, fault-tolerant systems that correct their own errors, are not yet generally available…

9 Best Fault-Tolerant Quantum Computing Providers for Enterprises

Enterprise technology leaders face an awkward planning problem with quantum computing. The machines expected to matter commercially, fault-tolerant systems that correct their own errors, are not yet generally available, yet the work of finding suitable problems, building skills and setting budgets takes years. Waiting means starting late; committing too early means paying for experiments that may not transfer.

Choosing a provider is complicated by vocabulary. “Fault tolerant” now appears in nearly every roadmap, but it covers everything from error detection on a handful of qubits to fully error-corrected machines running millions of operations. The evidence behind the dates ranges from peer-reviewed experiments to slideware.

This guide mentions nine fault-tolerant quantum computing providers, evaluating each on the evidence behind its fault-tolerance claims, the specificity of its roadmap, and the enterprise engagement available before fault-tolerant hardware arrives.

What Enterprise Leaders Should Know

  • Most providers now publish dated fault-tolerance targets between 2028 and 2030. All remain targets.
  • Error-detected and error-corrected logical qubits are different milestones, and announcements do not always make the distinction obvious.
  • Among the vendors in this list, QuEra pairs a 2028 target of 256 logical qubits with peer-reviewed 2025 results in logical algorithms and magic-state distillation.
  • Engagement options before fault tolerance range from pay-per-use cloud access to multi-year co-design programs.
  • Hyperscaler channels are becoming the default enterprise route to quantum hardware.

Selection Criteria

We assessed providers from the viewpoint of a corporate buyer with a multi-year horizon. Each company needed a public roadmap that names fault tolerance as a goal and attaches at least one quantitative milestone to it, such as a logical qubit count, logical error rate or operation count. 

We gave more weight to milestones supported by published experiments than to projections, and we checked whether reported logical qubits were error-corrected or only error-detected. We also looked at how enterprises can engage today, whether through cloud access, on-premises systems, software tools or structured co-design. 

We weighed each architecture’s approach to reducing error-correction overhead.

Providers in Brief

The 9 providers use five hardware approaches. Every milestone below is a target.

The Top Fault Tolerant Quantum Computing Providers List

1. QuEra

QuEra computes with individual neutral atoms that lasers arrange into programmable arrays. Moving atoms during a circuit gives flexible connectivity and lets a single operation act on many qubits at once, which suits error-correcting codes built on transversal logical gates. The machines run at room temperature, which simplifies placement in a data center.

Its fault-tolerance plan comes with numbers attached. Libra, aimed at Amazon Braket in 2028, targets 256 logical qubits, a logical error rate of one in a million and about a million reliable operations. The generation after it targets more than 1,000 logical qubits at one error in a billion. The supporting evidence is peer reviewed: 2025 Nature papers with Harvard and MIT collaborators reported fault-tolerant algorithms on up to 96 logical qubits and logical magic state distillation, the step that unlocks universal fault-tolerant algorithms.

Enterprises do not have to wait for Libra to begin. The analog Aquila system is on Amazon Braket today, and the FTQC Founders Circle offers a multi-year collaboration in which organizations bring high-value problems for evaluation and algorithm co-design, with a path toward priority access where technical and business fit are established. Co-design work covers use-case triage, classical baselines and resource estimates, so a team can decide whether a problem justifies fault-tolerant hardware at all.

Key features:

  • Neutral-atom arrays with reconfigurable connectivity
  • Parallel, transversal logical operations
  • Fault-tolerant algorithms on up to 96 logical qubits
  • Logical magic state distillation, published in Nature
  • Libra planned for Amazon Braket in 2028 with 256 logical qubits
  • Next-generation target of more than 1,000 logical qubits
  • Founders Circle co-design program for enterprises
  • Room-temperature systems with no cryogenic plant

2. Quantinuum

Quantinuum enters the enterprise market with a hyperscaler deal in hand. In August 2026 it announced a multi-year partnership with Oracle to install a Helios system in a US Oracle Cloud Infrastructure data center, for access through a managed OCI service. Helios has 98 qubits with 99.921% average two-qubit gate fidelity, and the roadmap runs to Sol in 2027 and the fault-tolerant Apollo in 2029.

The OCI service is still planned, with preview timing and pricing left to Oracle, so it is a future channel rather than a current one.

Key features:

  • 98-qubit Helios with 99.921% two-qubit fidelity
  • Planned managed access through Oracle Cloud Infrastructure
  • Sol in 2027 and Apollo in 2029

3. IBM

For companies already standardized on Qiskit, IBM offers a continuous path from today’s experiments to its fault-tolerant plans. Starling, to be built at a new IBM Quantum Data Center in Poughkeepsie, New York, targets 200 logical qubits running 100 million quantum operations by 2029. A successor, Blue Jay, targets 2,000 logical qubits and a billion operations.

Until Starling arrives, enterprise work runs on IBM’s pre-fault-tolerant processors, which means results from current projects come from hardware without full error correction.

Key features:

  • Starling target of 200 logical qubits by 2029
  • Blue Jay target of 2,000 logical qubits
  • Qiskit software and IBM cloud access

4. IonQ

IonQ makes trapped-ion hardware available through the cloud accounts many enterprises already hold, including Amazon Braket, Microsoft Azure and Google Cloud. Its acquisition of Oxford Ionics added 2D ion-trap technology and chip-based qubit control, the basis for a roadmap that projects 40,000 to 80,000 logical qubits by 2030.

Those logical-qubit figures are projections resting on a platform still moving from R&D prototype to product, so near-term enterprise work runs on IonQ’s current systems.

Key features:

  • Access through three major public clouds
  • 2D trap and electronic qubit control technology
  • Roadmap to tens of thousands of logical qubits by 2030

5. Microsoft

Microsoft is pursuing fault tolerance on two tracks at once. Its own research centers on topological qubits, and in June 2026 it announced Majorana 2, reporting a 1,000-fold reliability improvement over Majorana 1 and a path to a commercially useful, scalable machine by 2029. In parallel, it is building Magne, a Level 2 system, with Atom Computing in Denmark, and Azure Quantum gives enterprises a familiar entry point.

Independent physicists have questioned whether Microsoft’s Majorana devices demonstrate topological qubits, and Microsoft has not published a qubit count for Majorana 2.

Key features:

  • Majorana 2 topological processor
  • Magne logical-qubit system with Atom Computing
  • Azure Quantum platform

6. Infleqtion

Infleqtion runs two businesses: neutral-atom quantum computing and a quantum sensing and timing portfolio that generates much of its current revenue. In September 2026 it announced that it had entangled 30 logical qubits encoded in 80 atoms on its Sqale system. It targets 100 logical qubits in 2028 and more than 1,000 by 2030.

The 30-qubit result used an error-detecting code with post-selection, not error correction, and Infleqtion has not published a logical error rate for it.

Key features:

  • Sqale neutral-atom quantum computer
  • 30 error-detected logical qubits demonstrated
  • NVIDIA NVQLink integration

7. Xanadu

Xanadu’s most widely used product is software. Its open-source PennyLane library averaged about 160,000 monthly downloads in 2025, a low-cost way for enterprise developers to build skills. On the hardware side, Xanadu introduced Aurora, a modular, networked photonic system. It targets fault-tolerant operations by 2028 and up to 500 logical qubits in 2029 to 2030.

Aurora is a 12-qubit system, so the distance between current hardware and those targets is large.

Key features:

  • PennyLane open-source software library
  • Aurora modular photonic architecture
  • Target of up to 500 logical qubits by 2030

8. Alice & Bob

Paris-based Alice & Bob attacks error-correction overhead at the qubit level. Its superconducting cat qubits suppress bit-flip errors by design, reducing error correction from a two-dimensional problem to a one-dimensional one. In June 2026 it unveiled Helium, an on-premises system with Slurm compatibility, engineered to encode the company’s first logical qubit from as few as 18 cat qubits. Its roadmap ends with Graphene, a 100-logical-qubit machine planned for 2030.

That first logical qubit remains a design target, so enterprise engagement today is research-oriented.

Key features:

  • Cat qubits with built-in bit-flip suppression
  • Helium on-premises research system
  • Graphene target of 100 logical qubits by 2030

9. Rigetti

Rigetti builds superconducting processors from small chiplets tiled together. Its Cepheus-1-108Q system, generally available since April 2026, combines twelve 9-qubit chiplets with gate times of about 60 nanoseconds and is offered through Rigetti’s cloud and Amazon Braket.

The system’s 99.1% median two-qubit fidelity sits below the 99.5% Rigetti targets later in 2026, and the company has not yet published a dated fault-tolerant system.

Key features:

  • 108-qubit chiplet-based processor
  • Gate times of about 60 nanoseconds
  • Access through Amazon Braket

FAQs

What is fault-tolerant quantum computing?

Fault-tolerant quantum computing uses error correction to keep a computation reliable even though individual qubits and operations make mistakes. Information is spread across many physical qubits to form logical qubits, and errors are found and fixed in real time. This is what allows long, complex algorithms with commercial value to run to completion.

When will fault-tolerant quantum computers be available?

Most published targets fall between 2028 and 2030. QuEra plans Libra for Amazon Braket in 2028, while IBM and Quantinuum each target 2029 for their fault-tolerant systems. These are roadmap targets, and early machines will suit selected workloads rather than every problem an enterprise has.

What is the difference between quantum error detection and error correction?

Error detection identifies that an error occurred and discards the affected run. Error correction identifies the error and fixes it during the computation, so the run continues. Detection is a useful research step, but only correction scales to the long computations that fault-tolerant applications require.

How should an enterprise prepare for fault-tolerant quantum computing?

Start by identifying a consequential problem, establishing a strong classical baseline and estimating the logical resources a quantum approach would need. Programs like QuEra’s Founders Circle structure this work as co-design with the hardware team. Building internal skills through cloud access and open-source tools runs in parallel.

Is it worth investing in quantum computing before fault tolerance arrives?

For organizations with problems in chemistry, materials, optimization or simulation, a modest, staged investment usually makes sense. Early work builds skills, clarifies which problems fit and positions teams to use fault-tolerant hardware once it arrives. A sound outcome can also be a decision to wait.

The post 9 Best Fault-Tolerant Quantum Computing Providers for Enterprises appeared first on Hardware Times.