howclose.to
FAULT TOLERANT QUANTUM COMPUTING

The problemsIntelligence & MachinesFault-Tolerant Quantum

Updated · July 2026

How close are we to a quantum computer that doesn't break?Can we build a quantum computer that actually works?

Can we build a quantum computer that actually works?How close are we to a quantum computer that doesn't break?

Error-corrected qubits have crossed below threshold; the open question is scaling from one logical qubit to the millions a useful machine needs.Today's quantum computers are too error-prone to trust. Fixing that means going from a handful of good qubits to millions.

Current stageLab demoDemonstrated in the laboratoryShown to work in a lab
Read the evidence Where we stand
The path so far

Recorded milestones

  1. Quantum errors become correctable

  2. Universal pieces, tiny scale

Open tests · conceptual branches

TODAY
  1. OPEN TEST03 ↗
    Compute while correctingIntegrated Universal Logical OperationsSmall platforms have demonstrated the pieces, but not together at low enough error for a long useful algorithm.Combine repeated correction, preparation, measurement, Clifford and non-Clifford gates with real-time decoding in one system.
  2. OPEN TEST04 ↗
    Do useful work economicallyUtility-Scale Fault-Tolerant ComputationThe current frontier is one protected memory or a few error-detected logical qubits, not hundreds of long-lived logical qubits.Execute a valuable algorithm whose verified computational benefit exceeds the full hardware and operating cost.
TODAY
OPEN TEST03 ↗
Compute while correctingIntegrated Universal Logical OperationsSmall platforms have demonstrated the pieces, but not together at low enough error for a long useful algorithm.Combine repeated correction, preparation, measurement, Clifford and non-Clifford gates with real-time decoding in one system.
OPEN TEST04 ↗
Do useful work economicallyUtility-Scale Fault-Tolerant ComputationThe current frontier is one protected memory or a few error-detected logical qubits, not hundreds of long-lived logical qubits.Execute a valuable algorithm whose verified computational benefit exceeds the full hardware and operating cost.
Milestones are in recorded order. Open tests are separate questions, not a forecast.Milestones are in date order. Open tests are separate tests, not a promise.
Fault-Tolerant QuantumWhere we stand

The path shows Universal pieces, tiny scale. Open thresholds are separate tests in a conceptual map, not a forecast. No single percentage represents progress.

01 · Where we standWhere we stand

Where we standWhere we stand

The current stage, the evidence recorded, and the single threshold that gates the next stage.The stage reached, the evidence recorded, and the one test still blocking the next step.

What the evidence has shownWhat the evidence has shown

Today's quantum computers are too error-prone to trust. Fixing that means going from a handful of good qubits to millions.Error-corrected qubits have crossed below threshold; the open question is scaling from one logical qubit to the millions a useful machine needs.

2 of 4 tests recorded2 of 4 thresholds achieved open
✓Bigger codes work betterBelow-Threshold Error ScalingAchieved
✓Protection pays for itselfCorrected Memory Beats HardwareAchieved
Compute while correctingIntegrated Universal Logical OperationsOpen test
Do useful work economicallyUtility-Scale Fault-Tolerant ComputationOpen test

The next testBlocking threshold

Compute while correctingIntegrated Universal Logical Operations

Run a multi-logical-qubit universal circuit through repeated correction with a lower total failure rate than its physical implementation.Run a multi-logical-qubit universal circuit through repeated correction with a lower total failure rate than its physical implementation.

02 · Thresholds & evidenceThresholds & evidence

What would count as proof?What would count as proof?

A clear test, the evidence behind it, and the limits of the claim.A clear test, the evidence behind it, and what it still leaves open.

✓Achieved · Dec 2024

Bigger codes work betterBelow-Threshold Error Scaling

Willow's error fell 2.14-fold for each two-step increase in code distance.Increasing code distance must repeatedly lower logical error under realistic correction cycles without post-selection.

Recorded result A 101-physical-qubit distance-7 memory reached 0.143% error per cycle.

Next test Reproduce below-threshold scaling across several logical qubits while executing entangling gates, not memory alone.

SOURCE RECORD

Google Willow below-threshold memory, corrected April 2026 ↗

Still to demonstrateWhat this does not establishA separate system can still have different conditions and results.The named measurement has no wider implication until the supporting system test is also shown.

View every evidence record

Achieved

Bigger codes work betterBelow-Threshold Error Scaling

Willow's error fell 2.14-fold for each two-step increase in code distance.Increasing code distance must repeatedly lower logical error under realistic correction cycles without post-selection.

Recorded result: Dec 2024 · A 101-physical-qubit distance-7 memory reached 0.143% error per cycle.

Next test: Reproduce below-threshold scaling across several logical qubits while executing entangling gates, not memory alone.

Google Willow below-threshold memory, corrected April 2026 ↗

Achieved

Protection pays for itselfCorrected Memory Beats Hardware

Willow's logical lifetime was 291 microseconds versus 119 microseconds for its best physical constituent.An actively corrected logical memory must outlive every physical constituent without discarding unfavorable trials.

Recorded result: Dec 2024 · The distance-7 logical memory exceeded the best physical-qubit lifetime by 2.4±0.3 times.

Next test: Demonstrate equivalent breakeven for a universal logical gate set and multi-logical-qubit operations.

Google Willow below-threshold memory, corrected April 2026 ↗

Open test

Compute while correctingIntegrated Universal Logical Operations

Small platforms have demonstrated the pieces, but not together at low enough error for a long useful algorithm.Combine repeated correction, preparation, measurement, Clifford and non-Clifford gates with real-time decoding in one system.

Next test: Run a multi-logical-qubit universal circuit through repeated correction with a lower total failure rate than its physical implementation.

Open test

Do useful work economicallyUtility-Scale Fault-Tolerant Computation

The current frontier is one protected memory or a few error-detected logical qubits, not hundreds of long-lived logical qubits.Execute a valuable algorithm whose verified computational benefit exceeds the full hardware and operating cost.

Next test: IBM's 2029 Starling claim is a visible test: 200 logical qubits and 100 million gates must be independently characterized.

03 · The recordThe record

The record behind the verdictThe record behind the verdict

Major milestones and smaller context events share one source-linked record. Scheduled targets are marked separately from recorded events.Milestones, smaller events and setbacks share one sourced record. Scheduled targets are marked separately from recorded events.

Event key
TheoryExperimentFundingSetbackPolicyDeployment Future / scheduled
1982-19932 events1 shown

Computing Becomes Quantum

This record spans “Feynman proposes quantum simulation” to “Deutsch defines universal machine”.

1982
Feynman proposes quantum simulationTheory
Feynman argued that simulating quantum physics efficiently may require a computer governed by quantum mechanics.
Richard Feynman
1985
Deutsch defines universal machine
Deutsch described a universal quantum computer able in principle to simulate any finite physical system.
David Deutsch
1994-20014 events3 shown

Codes Make Reliability Possible

This record spans “Shor makes reliability consequential” to “Kitaev introduces topological protection”.

1994
Shor makes reliability consequentialTheory
Shor gave polynomial-time quantum algorithms for factoring and discrete logarithms, motivating very long reliable computations.
Peter Shor
1995
Quantum error correction appearsTheory
Shor showed that one logical qubit could be encoded across nine physical qubits to correct arbitrary single-qubit errors.
Peter Shor
1997
Kitaev introduces topological protection
Kitaev proposed fault-tolerant computation using anyons, laying the theoretical foundation for toric and surface codes.
Alexei Kitaev
2002-20152 events1 shown

Correction Enters Hardware

This record spans “Surface-code threshold quantified” to “Trapped ions correct errors”.

2004
Trapped ions correct errorsExperiment
Three beryllium-ion qubits encoded, detected and corrected induced spin-flip errors in the first complete QEC sequence.
NIST
2016-20223 events2 shown

The Breakeven Experiments

This record spans “Cat code reaches breakeven” to “Fault-tolerant control beats physical”.

2016
Cat code reaches breakevenExperiment
A corrected bosonic qubit lasted 320 microseconds, 1.1 times its best physical constituent, without post-selection.
Yale University
2023-203318 events4 shown

Below Threshold, Not Useful

This record spans “Larger surface code barely wins” to “IBM targets Starling availability” and includes scheduled targets.

2023
Larger surface code barely winsExperiment
Google's distance-5 memory reached 2.914% error per cycle versus 3.028% at distance 3, only a 4% relative reduction.
Google Quantum AI
2023
Correlated event sets error floor
One high-energy event left a distance-25 repetition code at 1.7×10^-6 error per cycle; removing 0.15% of trials cut it tenfold.
Google Quantum AI
2023
Real-time bosonic QEC beats every component
A continuously corrected logical qubit achieved 2.27 +/- 0.07 times the coherence of its best imperfect component.
Yale University and collaborators
2023
Neutral atoms encode 48 logicals
A 128-atom processor ran sampling circuits with 48 error-detected logical qubits, but full post-selection retained only eight samples.
Harvard University, MIT and QuEra
2024
Low-overhead qLDPC protocol published
Simulations found a 0.7% circuit-level threshold and modelled 12 logical memories for nearly one million cycles using 288 physical qubits; no such processor was built.
IBM researchers
2024
Willow crosses surface thresholdExperiment
A 101-qubit distance-7 memory reached 0.143% error per cycle, improving 2.14-fold for each two-distance increase.
Google Quantum AI
2024
Rare bursts limit scaling
Willow repetition codes encountered correlated bursts about once per hour, producing an observed error floor near 10^-10.
Google Quantum AI
2025
Ocelot tests biased-noise correction
A nine-qubit chip combining five cat qubits and four ancillas showed phase-error detection improving from three- to five-cat repetition codes; overhead savings remain projected.
AWS Center for Quantum Computing and Caltech
2025
Logical magic-state distillation demonstrated
A neutral-atom processor ran 5-to-1 distillation with distance-3 and distance-5 colour codes, producing logical outputs with higher fidelity than their inputs.
Harvard University, MIT, QuEra and collaborators
2025
Neutral atoms repeat correction
A 448-atom architecture reached 0.62% logical error per round at distance 5 versus 1.33% at distance 3 after decoder tuning.
Harvard University, MIT and QuEra
2026
Universal toolkit remains tinyExperiment
Eight trapped-ion physical qubits hosted three error-detected logical qubits for a measurement-free Grover search over eight items.
University of Innsbruck and collaborators
2026
Hardware decoder closes a small latency loop
An FPGA-integrated superconducting test sustained up to 25 rounds with sub-microsecond mean decoding per round and 9.6-microsecond end-to-end feedback after nine rounds.
Riverlane, Rigetti and collaborators
2026
Logical circuits beat selected physical baselines
Peer-reviewed trapped-ion circuits spanning up to 12 logical qubits improved 11- to 800-fold over selected physical-circuit baselines, with post-selection; the ratio was circuit-dependent.
Microsoft and Quantinuum
2026
Neutral atoms demonstrate erasure-aware logic
Metastable ytterbium qubits supported adaptive erasure checks, a [[4,2,2]] encoding and teleportation between code blocks, but could not correct arbitrary Pauli errors.
Princeton University and collaborators
2026
Willow halves the logical error rateExperiment
Google's surface-code memory reached 7.7e-4 logical error per cycle, down from 1.4e-3 in 2024 - correction improving with scale, though still astronomically far from the ~1e-12 a useful algorithm needs.
Google Quantum AI
2026
One ion hosts an error-corrected spin-cat qubitWe are here
A spin-cat encoding inside one atomic ion reduced errors by up to 2.2-fold and extended lifetime by up to 1.5-fold over an unencoded qubit.
MIT, MIT Lincoln Laboratory and collaborators
2029
IBM targets Starling availabilityDeploymentTarget
IBM targets a client-accessible system with 200 logical qubits and capacity for 100 million gates; this is a vendor roadmap.
IBM
2015-20333 events0 shown

Additional records

This record spans “Repeated detection reduces failures” to “DARPA utility deadline arrives” and includes scheduled targets.

2015
Repeated detection reduces failures
A nine-qubit superconducting device cut retrieval failures 8.5-fold after eight error-detection cycles, using post-selection.
Google and UC Santa Barbara
2033
IBM targets billion-gate Blue JayDeploymentTarget
IBM targets 2,000 logical qubits and one billion gates after 2033; no hardware meeting this specification exists today.
IBM
2033
DARPA utility deadline arrivesPolicyTarget
QBI aims to verify whether any architecture can deliver computational value exceeding total system cost by 2033.
DARPA
— end of record · 32 events shown —
04 · The dataThe data

The data behind the verdictThe data behind the verdict

Read each measure with its units, conditions, uncertainty and source; targets remain distinct from observations.Read each chart with its conditions, limits and source. Future targets remain distinct from recorded results.

Open figure in archive: The distance-3, distance-5 and distance-7 memories used 17, 49 and 101 physical qubits respectively; the duplicate 2023 points are different code distances, not elapsed-time growth.Open figure in archive: Suppression strengthened from a marginal 1.039-fold distance-3-to-5 improvement to a 2.14-fold reduction for each two-distance increase on Willow.Open figure in archive: Two breakeven landmarks, but not a clean platform series: 2016 used a bosonic cat code, whereas 2024 used a 101-qubit surface code.Open figure in archive: Vendor roadmap targets for Starling and Blue Jay, not measured systems or independently validated logical qubits.Open figure in archive: IBM's stated circuit-capacity objectives; the roadmap explicitly says its goals may change or be withdrawn.Open figure in archive: Google's first larger surface code improved logical error by only about 4%, showing how narrow the first below-scaling signal was.Open figure in archive: Willow's corrected logical memory lasted 2.4 times longer than its best constituent physical qubit.Open figure in archive: Even Google's extrapolated 10^-6 memory would require about fourteen times the physical qubits used by its demonstrated distance-7 memory.Open figure in archive: Numerically, Google's memory-cycle error is roughly 1.4 billion times the 10^-12 operation target, though the two metrics differ.
Superconducting surface-code logical error probability per correction cycleSuperconducting surface-code logical error probability per correction cycle
Scale

Google Sycamore distance 3, 17 physical qubits · 0.0303 probability per cycle · 2023 · verification unspecified

Superconducting surface-code logical error probability per correction cycle over time, with measured values, projected values, and a goal at 1e-12 probability per cycle.1e-121e-111e-101e-91e-81e-71e-61e-5Superconducting surface-code logical error probability… · probability per cycleYear202320242026GOAL 1e-12 · Approximate logical-operation err…Google Sycamore distance 3,…Google Willow, surface-code…~0.0008 probability per cycle to goal
NOTE - This follows the candidate file's recommended metric and keeps only Google's comparable superconducting surface-code memory sequence. A memory error per correction cycle is not a logical-gate error, and changes in chips, decoders and cycle circuits prevent treating this as a pure hardware trend. The 10^-12 goal is an application estimate per logical operation, not a demonstrated threshold. Neutral-atom, ion, colour-code and post-selected results are kept out of this line because their metrics are not interchangeable.
View data table
All metric observations
DateValueStatusCondition / labelVerificationCICensoringSource
20230.03028 probability per cycleobservedGoogle Sycamore distance 3, 17 physical qubitsunspecified—noneGoogle 2023 surface-code scaling experiment
20230.02914 probability per cycleobservedGoogle Sycamore distance 5, 49 physical qubitsunspecified—noneGoogle 2023 surface-code scaling experiment
20240.00143 probability per cycleobservedGoogle Willow distance 7, 101 physical qubitsunspecified—noneGoogle Willow below-threshold memory, corrected April 2026
20260.000772 probability per cycleobservedGoogle Willow, surface-code memory (Jul 2026)unspecified—noneTimeline · Willow halves the logical error rate
AS OF July 2026
View every data figure
Superconducting surface-code logical error probability per correction cycleSuperconducting surface-code logical error probability per correction cycle
Scale

Google Sycamore distance 3, 17 physical qubits · 0.0303 probability per cycle · 2023 · verification unspecified

Superconducting surface-code logical error probability per correction cycle over time, with measured values, projected values, and a goal at 1e-12 probability per cycle.1e-121e-111e-101e-91e-81e-71e-61e-5Superconducting surface-code logical error probability… · probability per cycleYear202320242026GOAL 1e-12 · Approximate logical-operation err…Google Sycamore distance 3,…Google Willow, surface-code…~0.0008 probability per cycle to goal
NOTE - This follows the candidate file's recommended metric and keeps only Google's comparable superconducting surface-code memory sequence. A memory error per correction cycle is not a logical-gate error, and changes in chips, decoders and cycle circuits prevent treating this as a pure hardware trend. The 10^-12 goal is an application estimate per logical operation, not a demonstrated threshold. Neutral-atom, ion, colour-code and post-selected results are kept out of this line because their metrics are not interchangeable.
View data table
All metric observations
DateValueStatusCondition / labelVerificationCICensoringSource
20230.03028 probability per cycleobservedGoogle Sycamore distance 3, 17 physical qubitsunspecified—noneGoogle 2023 surface-code scaling experiment
20230.02914 probability per cycleobservedGoogle Sycamore distance 5, 49 physical qubitsunspecified—noneGoogle 2023 surface-code scaling experiment
20240.00143 probability per cycleobservedGoogle Willow distance 7, 101 physical qubitsunspecified—noneGoogle Willow below-threshold memory, corrected April 2026
20260.000772 probability per cycleobservedGoogle Willow, surface-code memory (Jul 2026)unspecified—noneTimeline · Willow halves the logical error rate
AS OF July 2026
The distance-3, distance-5 and distance-7 memories used 17, 49 and 101 physical qubits respectively; the duplicate 2023 points are different code distances, not elapsed-time growth.The distance-3, distance-5 and distance-7 memories used 17, 49 and 101 physical qubits respectively; the duplicate 2023 points are different code distances, not elapsed-time growth.

Independent records

Google surface-code physical-qubit overhead linear scale · physical qubits per logical memory

linear scale · physical qubits per logical memory
202317 physical qubits per logical memory
202349 physical qubits per logical memory
2024101 physical qubits per logical memory
observed self-reported target / pathway right-censored lower bound
View data table
All plotted observations
SeriesDateValueStatusCondition / labelVerificationCICensoringSource
Google surface-code physical-qubit overhead202317 physical qubits per logical memoryobservedGoogle surface-code physical-qubit overheadunspecified—noneGoogle 2023 surface-code scaling experiment
Google surface-code physical-qubit overhead202349 physical qubits per logical memoryobservedGoogle surface-code physical-qubit overheadunspecified—noneGoogle 2023 surface-code scaling experiment
Google surface-code physical-qubit overhead2024101 physical qubits per logical memoryobservedGoogle surface-code physical-qubit overheadunspecified—noneGoogle Willow below-threshold memory, corrected April 2026
AS OF July 2026
Suppression strengthened from a marginal 1.039-fold distance-3-to-5 improvement to a 2.14-fold reduction for each two-distance increase on Willow.Suppression strengthened from a marginal 1.039-fold distance-3-to-5 improvement to a 2.14-fold reduction for each two-distance increase on Willow.

linear scale · ratio by year

Google surface-code error-suppression factor per two distance steps · 1.039 ratio · Jan 2023 · verification unspecified

Suppression strengthened from a marginal 1.039-fold distance-3-to-5 improvement to a 2.14-fold reduction for each two-distance increase on Willow.Google surface-code error-suppression factor per two distance steps: 1.039 ratio · Jan 20232.14 ratio · Jan 2024Google surface-code error-suppression factor per two distance steps: 2.14 ratio · Jan 2024Year
Log-scale line joins each new longest record; all observations remain individual dots.

First record

Jan 2023

1.039 ratio · Jan 2023

observed · Verification unspecified

Latest record

Jan 2024

2.14 ratio · Jan 2024

observed · Verification unspecified

View data table
All plotted observations
SeriesDateValueStatusCondition / labelVerificationCICensoringSource
Google surface-code error-suppression factor per two distance steps20231.039 ratioobservedGoogle surface-code error-suppression factor per two distance stepsunspecified—noneGoogle 2023 surface-code scaling experiment
Google surface-code error-suppression factor per two distance steps20242.14 ratioobservedGoogle surface-code error-suppression factor per two distance stepsunspecified—noneGoogle Willow below-threshold memory, corrected April 2026
AS OF July 2026
Two breakeven landmarks, but not a clean platform series: 2016 used a bosonic cat code, whereas 2024 used a 101-qubit surface code.Two breakeven landmarks, but not a clean platform series: 2016 used a bosonic cat code, whereas 2024 used a 101-qubit surface code.

linear scale · times best physical constituent lifetime by year

Reported logical-memory lifetime advantage · 1.1 times best physical constituent lifetime · Jan 2016 · verification unspecified

Two breakeven landmarks, but not a clean platform series: 2016 used a bosonic cat code, whereas 2024 used a 101-qubit surface code.Reported logical-memory lifetime advantage: 1.1 times best physical constituent lifetime · Jan 20162.4 times best physical constituent lifetime · Jan 2024Reported logical-memory lifetime advantage: 2.4 times best physical constituent lifetime · Jan 2024Year
Log-scale line joins each new longest record; all observations remain individual dots.

First record

Jan 2016

1.1 times best physical constituent lifetime · Jan 2016

observed · Verification unspecified

Latest record

Jan 2024

2.4 times best physical constituent lifetime · Jan 2024

observed · Verification unspecified

View data table
All plotted observations
SeriesDateValueStatusCondition / labelVerificationCICensoringSource
Reported logical-memory lifetime advantage20161.1 times best physical constituent lifetimeobservedReported logical-memory lifetime advantageunspecified—noneYale cat-code breakeven experiment
Reported logical-memory lifetime advantage20242.4 times best physical constituent lifetimeobservedReported logical-memory lifetime advantageunspecified—noneGoogle Willow below-threshold memory, corrected April 2026
AS OF July 2026
Vendor roadmap targets for Starling and Blue Jay, not measured systems or independently validated logical qubits.Vendor roadmap targets for Starling and Blue Jay, not measured systems or independently validated logical qubits.

log scale · logical qubits by year

IBM announced fault-tolerant logical-qubit targets · 200 logical qubits · Jan 2029 · verification unspecified

Vendor roadmap targets for Starling and Blue Jay, not measured systems or independently validated logical qubits.IBM announced fault-tolerant logical-qubit targets: 200 logical qubits · Jan 2029 · target2,000 logical qubits · Jan 2033IBM announced fault-tolerant logical-qubit targets: 2,000 logical qubits · Jan 2033 · targetYear
Log-scale line joins each new longest record; all observations remain individual dots.

First record

Jan 2029

200 logical qubits · Jan 2029

target · Verification unspecified

Latest record

Jan 2033

2,000 logical qubits · Jan 2033

target · Verification unspecified

View data table
All plotted observations
SeriesDateValueStatusCondition / labelVerificationCICensoringSource
IBM announced fault-tolerant logical-qubit targets2029200 logical qubitstargetIBM announced fault-tolerant logical-qubit targetsunspecified—noneIBM quantum roadmap, updated March 2026
IBM announced fault-tolerant logical-qubit targets20332000 logical qubitstargetIBM announced fault-tolerant logical-qubit targetsunspecified—noneIBM quantum roadmap, updated March 2026
AS OF July 2026
IBM's stated circuit-capacity objectives; the roadmap explicitly says its goals may change or be withdrawn.IBM's stated circuit-capacity objectives; the roadmap explicitly says its goals may change or be withdrawn.

log scale · quantum gates by year

IBM announced executable-gate targets · 1e+8 quantum gates · Jan 2029 · verification unspecified

IBM's stated circuit-capacity objectives; the roadmap explicitly says its goals may change or be withdrawn.IBM announced executable-gate targets: 1e+8 quantum gates · Jan 2029 · target1e+9 quantum gates · Jan 2033IBM announced executable-gate targets: 1e+9 quantum gates · Jan 2033 · targetYear
Log-scale line joins each new longest record; all observations remain individual dots.

First record

Jan 2029

1e+8 quantum gates · Jan 2029

target · Verification unspecified

Latest record

Jan 2033

1e+9 quantum gates · Jan 2033

target · Verification unspecified

View data table
All plotted observations
SeriesDateValueStatusCondition / labelVerificationCICensoringSource
IBM announced executable-gate targets2029100000000 quantum gatestargetIBM announced executable-gate targetsunspecified—noneIBM quantum roadmap, updated March 2026
IBM announced executable-gate targets20331000000000 quantum gatestargetIBM announced executable-gate targetsunspecified—noneIBM quantum roadmap, updated March 2026
AS OF July 2026
Google's first larger surface code improved logical error by only about 4%, showing how narrow the first below-scaling signal was.Google's first larger surface code improved logical error by only about 4%, showing how narrow the first below-scaling signal was.
Distance 33.028 % error/cycle
−0.114 % error/cycle · 0.96× of A
Distance 52.914 % error/cycle

Each square = 0.1 % error/cycle

Distance 33.028 % error/cycle
Distance 52.914 % error/cycle
0.114 % error/cycle less from distance 5 than from distance 3
View data table
All aligned comparisons
ClaimABSource / provenance
Google's first larger surface code improved logical error by only about 4%, showing how narrow the first below-scaling signal was.3.028 % error/cycle · Distance 32.914 % error/cycle · Distance 5Google 2023 surface-code scaling experiment
AS OF July 2026
Willow's corrected logical memory lasted 2.4 times longer than its best constituent physical qubit.Willow's corrected logical memory lasted 2.4 times longer than its best constituent physical qubit.
Distance-7 logical memory291 microseconds
−172 microseconds · 0.41× of A
Best constituent physical qubit119 microseconds

Each square = 10 microseconds

Distance-7 logical memory291 microseconds
Best constituent physical qubit119 microseconds
172 microseconds less from best constituent physical qubit than from distance-7 logical memory
View data table
All aligned comparisons
ClaimABSource / provenance
Willow's corrected logical memory lasted 2.4 times longer than its best constituent physical qubit.291 microseconds · Distance-7 logical memory119 microseconds · Best constituent physical qubitGoogle Willow below-threshold memory, corrected April 2026
AS OF July 2026
Even Google's extrapolated 10^-6 memory would require about fourteen times the physical qubits used by its demonstrated distance-7 memory.Even Google's extrapolated 10^-6 memory would require about fourteen times the physical qubits used by its demonstrated distance-7 memory.
Demonstrated distance 7101 physical qubits
+1,356 physical qubits · gain 14.43
Projected distance 271,457 physical qubits · projectedThis is a projection, not a demonstrated machine.Distance-27 resource count is a projection, not a demonstrated machine.

Each square = 50 physical qubits

Demonstrated distance 7101 physical qubits
Projected distance 271,457 physical qubits · projectedThis is a projection, not a demonstrated machine.Distance-27 resource count is a projection, not a demonstrated machine.
1,356 physical qubits more from projected distance 27 than from demonstrated distance 7
View data table
All aligned comparisons
ClaimABSource / provenance
Even Google's extrapolated 10^-6 memory would require about fourteen times the physical qubits used by its demonstrated distance-7 memory.101 physical qubits · Demonstrated distance 71457 physical qubits · Projected distance 27 · projected
This is a projection, not a demonstrated machine.Distance-27 resource count is a projection, not a demonstrated machine.		Google Willow below-threshold memory, corrected April 2026
AS OF July 2026
Numerically, Google's memory-cycle error is roughly 1.4 billion times the 10^-12 operation target, though the two metrics differ.Numerically, Google's memory-cycle error is roughly 1.4 billion times the 10^-12 operation target, though the two metrics differ.

Numerically, Google's memory-cycle error is roughly 1.4 billion times the 10^-12 operation target, though the two metrics differ.

Willow memory error0.0014 probability/cycle
Factoring-scale target1e-12 probability/operation · target

ConditionThis is a future target, not an observed result.Factoring-scale error target is a required future engineering condition, not an observed result.

Each value is scaled against zero in its own unit; bar length is not a cross-unit comparison.
View data table
All aligned comparisons
ClaimABSource / provenance
Numerically, Google's memory-cycle error is roughly 1.4 billion times the 10^-12 operation target, though the two metrics differ.0.00143 probability/cycle · Willow memory error1e-12 probability/operation · Factoring-scale target · target
This is a future target, not an observed result.Factoring-scale error target is a required future engineering condition, not an observed result.		AlphaQubit decoder and 10^-12 application target
AS OF July 2026

Methods & conditions

Under the measurement.

The boundary of a measurement changes what it means.

Technical note 1 of 9

Bosonic breakeven was more than marginal

Method and conditions

Yale's 2023 result is stronger than a simple “16% breakeven” claim: the fully stabilized bosonic logical qubit reported a coherence gain of 2.27±0.07 over every imperfect component involved in correction.

Source notes
Nature

Technical notes and sources

Each note keeps its authored method, conditions and citations available for review.

Bosonic breakeven was more than marginal

Yale's 2023 result is stronger than a simple “16% breakeven” claim: the fully stabilized bosonic logical qubit reported a coherence gain of 2.27±0.07 over every imperfect component involved in correction.

Nature
qLDPC's overhead claim is modelled

IBM's bivariate-bicycle qLDPC proposal has degree-6 connectivity, a depth-eight syndrome circuit and a simulated 0.7% threshold; its 288-physical/12-logical comparison is a modelled workload at 0.1% physical error, not a processor result.

Nature
Logical-circuit gains depend on the test

Quantinuum and Microsoft used a 12-physical-qubit code encoding two logical qubits and a 16-qubit tesseract colour code encoding four. The reported 11×-800× range depends on the circuit, baseline and acceptance rule; it is not one universal logical error rate.

Nature
Why magic-state distillation matters

Logical magic-state distillation matters because Clifford operations alone are not universal: the neutral-atom experiment combined five logical inputs into one improved output and tested both distance-3 and distance-5 colour codes.

Nature
Ocelot concentrates on phase flips

Ocelot's noise bias is the mechanism: cat qubits suppress bit flips so a repetition code can concentrate on phase flips. The demonstrated device contained five data cat qubits plus four syndrome ancillas.

Caltech
Decoding must keep pace with syndromes

The June 2026 decoder result addresses the backlog problem: syndrome generation must not outrun classical decoding. Its sub-microsecond mean per round is encouraging, but the measured closed-loop response for nine rounds was 9.6 µs and the test involved only eight qubits.

Nature Communications
Online tuning improved different things

Google's 2026 controller used error-detection events to tune more than 1,000 analogue control parameters while QEC ran. It improved stability against injected drift by 3.5×, whereas fine-tuning an already calibrated processor reduced logical error by about 20%; these are different claims.

Nature
Colour-code and surface-code rates differ

The same Google paper also reports a distance-5 colour-code memory at 8.19(14)×10⁻³ error per cycle. Do not merge this with the surface-code series: code family, layout and logical circuit differ.

Nature
Erasure information is not full correction

Erasure conversion supplies the decoder with the location of certain faults, distinct from an ordinary syndrome. Princeton's [[4,2,2]] experiment exploited that information but did not demonstrate correction of arbitrary unknown Pauli errors.

Nature Physics
05 · What it unlocksWhat it unlocks

If the remaining tests pass.If the remaining tests pass.

Downstream capabilities remain conditional on results that are not yet in.These are possible next capabilities, conditional on results that are not yet in.

TODAY
  1. PASSED · Dec 202401 ↗
    Bigger codes work betterBelow-Threshold Error ScalingWillow's error fell 2.14-fold for each two-step increase in code distance.Increasing code distance must repeatedly lower logical error under realistic correction cycles without post-selection.
  2. PASSED · Dec 202402 ↗
    Protection pays for itselfCorrected Memory Beats HardwareWillow's logical lifetime was 291 microseconds versus 119 microseconds for its best physical constituent.An actively corrected logical memory must outlive every physical constituent without discarding unfavorable trials.
  3. OPEN TEST03 ↗
    Compute while correctingIntegrated Universal Logical OperationsSmall platforms have demonstrated the pieces, but not together at low enough error for a long useful algorithm.Combine repeated correction, preparation, measurement, Clifford and non-Clifford gates with real-time decoding in one system.
  4. OPEN TEST04 ↗
    Do useful work economicallyUtility-Scale Fault-Tolerant ComputationThe current frontier is one protected memory or a few error-detected logical qubits, not hundreds of long-lived logical qubits.Execute a valuable algorithm whose verified computational benefit exceeds the full hardware and operating cost.
PASSED · Dec 202401 ↗
Bigger codes work betterBelow-Threshold Error ScalingWillow's error fell 2.14-fold for each two-step increase in code distance.Increasing code distance must repeatedly lower logical error under realistic correction cycles without post-selection.
PASSED · Dec 202402 ↗
Protection pays for itselfCorrected Memory Beats HardwareWillow's logical lifetime was 291 microseconds versus 119 microseconds for its best physical constituent.An actively corrected logical memory must outlive every physical constituent without discarding unfavorable trials.
TODAY
OPEN TEST03 ↗
Compute while correctingIntegrated Universal Logical OperationsSmall platforms have demonstrated the pieces, but not together at low enough error for a long useful algorithm.Combine repeated correction, preparation, measurement, Clifford and non-Clifford gates with real-time decoding in one system.
OPEN TEST04 ↗
Do useful work economicallyUtility-Scale Fault-Tolerant ComputationThe current frontier is one protected memory or a few error-detected logical qubits, not hundreds of long-lived logical qubits.Execute a valuable algorithm whose verified computational benefit exceeds the full hardware and operating cost.

Possible, if the open tests pass.

  1. Predictive quantum chemistry

    This is a possible benefit only if predictive quantum chemistry works safely, reliably and at useful scale.Potential benefit only if the stated use (Predictive quantum chemistry) is demonstrated with evidence for safety, reliability, affordability and scale.

  2. New materials simulation

    This is a possible benefit only if new materials simulation works safely, reliably and at useful scale.Potential benefit only if the stated use (New materials simulation) is demonstrated with evidence for safety, reliability, affordability and scale.

  3. Large-scale cryptanalysis

    This is a possible benefit only if large-scale cryptanalysis works safely, reliably and at useful scale.Potential benefit only if the stated use (Large-scale cryptanalysis) is demonstrated with evidence for safety, reliability, affordability and scale.

  4. Molecular simulation for medicine

    This is a possible benefit only if molecular simulation for medicine works safely, reliably and at useful scale.Potential benefit only if the stated use (Molecular simulation for medicine) is demonstrated with evidence for safety, reliability, affordability and scale.

    Related problem ↗
These are possible outcomes, if the open tests pass. They are conditional, not promises.
06 · The players & the moneyThe players & the money

Meet the builders.Meet the builders.

The organisations, their approaches, and the evidence behind them.The teams doing the work, the money around them, and what the record can support.

USA · lab

Google Quantum AI

Superconducting surface and colour codes, Willow processors, decoding and control research.

Funding contextFunding not recorded in this record.Funded inside Alphabet; no standalone financing round.
Named investorsInvestors not recorded in this record.

Capital is context, not a demonstration.

Company source ↗
View every builder record

Google Quantum AI

USA · lab · Superconducting surface and colour codes, Willow processors, decoding and control research.

Funding: Funding not recorded in this record.

Funded inside Alphabet; no standalone financing round.

Company source ↗

IBM Quantum

USA · lab · Superconducting hardware, qLDPC research, middleware and a public fault-tolerance roadmap.

Funding: Funding not recorded in this record.

Roadmap logical-qubit and gate counts are vendor targets, not achieved capacity.

Company source ↗

Quantinuum

USA / UK · company · Trapped-ion processors, logical circuits, compilers and quantum cybersecurity software.

Funding: $300M equity round · Jan 2024 · $5B pre-money valuation

Named investors: JPMorgan Chase · Mitsui · Amgen · Honeywell

Company source ↗

PsiQuantum

USA / Australia · company · Silicon-photonic architecture targeting utility-scale fault tolerance in Brisbane and Chicago.

Funding: Series E · $1B · Sep 2025

Named investors: BlackRock · Temasek · Baillie Gifford · NVentures · Qatar Investment Authority

Funding and government-backed sites do not establish that a utility-scale machine has been built.

Company source ↗

QuEra

USA · company · Neutral-atom logical processors, repeated correction and reconfigurable atom-array architectures.

Funding: >$230M financing · Feb 2025

Named investors: Google Quantum AI · SoftBank Vision Fund 2 · Valor Equity Partners · NVentures

Company source ↗

Riverlane

UK · company · Hardware-agnostic quantum error-correction decoding and real-time control stack.

Funding: Series C · $75M · Aug 2024

Decoder progress is necessary infrastructure, not a complete fault-tolerant computer.

Company source ↗
07 · SourcesSources

Follow the evidence.Follow the evidence.

49 sources — every figure on this page traces to one.49 sources — every figure on this page traces to one.

Source · what this page citesScope and limits
  1. 01

    Feynman: Simulating physics with computers

    Read source ↗doi.org
    Used on this page
    All 2 uses across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  2. 02

    Deutsch: Universal quantum computer

    Read source ↗royalsocietypublishing.org
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  3. 03

    Scope: Primary publication named by Shor: Quantum error-correcting code. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  4. 04

    Scope: Not recorded. Limit: Not recorded.

  5. 05

    Dennis, Kitaev, Landahl and Preskill: Topological quantum memory

    Read source ↗preskill.caltech.edu
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  6. 06

    NIST trapped-ion quantum error correction

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Primary publication named by NIST trapped-ion quantum error correction. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  7. 07

    Scope: Primary publication named by Yale cat-code breakeven experiment. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  8. 08

    Google 2023 surface-code scaling experiment

    Read source ↗nature.com
    Used on this page
    All 12 uses across 8 destinations

    Scope: Primary publication named by Google 2023 surface-code scaling experiment. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  9. 09

    Google Willow below-threshold memory, corrected April 2026

    Read source ↗nature.com
    Used on this page
    All 18 uses across 12 destinations

    Scope: Primary publication named by Google Willow below-threshold memory, corrected April 2026. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  10. 10

    April 2026 Willow author correction

    Read source ↗nature.com

    Scope: Primary publication named by April 2026 Willow author correction. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  11. 11

    Harvard-led 48-logical-qubit neutral-atom processor

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Primary publication named by Harvard-led 48-logical-qubit neutral-atom processor. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  12. 12

    Harvard-led universal neutral-atom architecture

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Primary publication named by Harvard-led universal neutral-atom architecture. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  13. 13

    Google superconducting colour-code experiment

    Read source ↗nature.com

    Scope: Primary publication named by Google superconducting colour-code experiment. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  14. 14

    Scope: Primary publication named by Measurement-free universal logical computation. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  15. 15

    Scope: Primary publication named by AlphaQubit decoder and 10^-12 application target. Limit: The publication's measured scope does not establish downstream engineering or deployment claims.

  16. 16

    Scope: Not recorded. Limit: Not recorded.

  17. 17

    DARPA Quantum Benchmarking Initiative

    Read source ↗darpa.mil
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  18. 18

    Timeline · Shor makes reliability consequential

    Read source ↗math.mit.edu
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  19. 19

    Timeline · Kitaev introduces topological protection

    Read source ↗arxiv.org
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  20. 20

    Timeline · Repeated detection reduces failures

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  21. 21

    Timeline · Majorana evidence is retracted

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  22. 22

    Timeline · Fault-tolerant control beats physical

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  23. 23

    Timeline · Real-time bosonic QEC beats every component

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  24. 24

    Timeline · Discrete logical qubit passes breakeven

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  25. 25

    Timeline · Low-overhead qLDPC protocol published

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  26. 26

    Timeline · Ocelot tests biased-noise correction

    Read source ↗caltech.edu
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  27. 27

    Timeline · Logical magic-state distillation demonstrated

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  28. 28

    Timeline · Hardware decoder closes a small latency loop

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  29. 29

    Timeline · Logical circuits beat selected physical baselines

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  30. 30

    Timeline · Neutral atoms demonstrate erasure-aware logic

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  31. 31

    Scope: Not recorded. Limit: Not recorded.

  32. 32

    Timeline · One ion hosts an error-corrected spin-cat qubit

    Read source ↗nature.com
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  33. 33

    companies · Google Quantum AI

    Read source ↗quantumai.google
    Used on this page
    All 1 use across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  34. 34

    companies · Quantinuum

    Read source ↗honeywell.com
    Used on this page
    All 2 uses across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  35. 35

    Scope: Not recorded. Limit: Not recorded.

  36. 36

    companies · QuEra

    Read source ↗quera.com
    Used on this page
    All 2 uses across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  37. 37

    companies · Riverlane

    Read source ↗riverlane.com
    Used on this page
    All 2 uses across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  38. 38

    people · Hartmut Neven

    Read source ↗quantumai.google
    Used on this page
    All 1 use across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  39. 39

    people · Jay Gambetta

    Read source ↗research.ibm.com
    Used on this page
    All 1 use across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  40. 40

    people · Mikhail Lukin

    Read source ↗quera.com
    Used on this page
    All 1 use across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  41. 41

    market · A$940M government package for PsiQuantum Brisbane

    Read source ↗qao.qld.gov.au
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  42. 42

    market · £2.5B UK quantum R&D commitment

    Read source ↗gov.uk
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  43. 43

    market · $625M renewal for five US quantum centers

    Read source ↗energy.gov
    Used on this page
    All 2 uses across 2 destinations

    Scope: Not recorded. Limit: Not recorded.

  44. 44

    geography · European Union

    Read source ↗digital-strategy.ec.europa.eu
    Used on this page
    All 1 use across 1 destination

    Scope: Not recorded. Limit: Not recorded.

  45. 45

    IBM Quantum official site

    Read source ↗ibm.com

    Scope: Not recorded. Limit: Not recorded.

  46. 46

    Quantinuum official site

    Read source ↗quantinuum.com

    Scope: Not recorded. Limit: Not recorded.

  47. 47

    PsiQuantum official site

    Read source ↗psiquantum.com

    Scope: Not recorded. Limit: Not recorded.

  48. 48

    QuEra official site

    Read source ↗quera.com

    Scope: Not recorded. Limit: Not recorded.

  49. 49

    Riverlane official site

    Read source ↗riverlane.com

    Scope: Not recorded. Limit: Not recorded.

Sources describe evidence. Inclusion is not endorsement.