Gscl

Pioneering Fault Tolerance

Google Surface-code logical prototype Research prototype

Google's research prototype demonstrates the foundational principles of fault-tolerant quantum computing by successfully encoding and operating a logical qubit with error suppression.

Google Quantum AI Superconducting transmon qubits Research prototype Logical qubits confidence: high

The quest for a fault-tolerant quantum computer represents one of the most significant challenges in modern physics and computer science. Unlike classical bits, quantum bits (qubits) are inherently fragile, susceptible to noise and decoherence from their environment. This fragility leads to errors that can quickly corrupt quantum computations, making large-scale, reliable quantum algorithms impossible without robust error correction.

Google Quantum AI's Surface-code logical prototype, unveiled in 2023, marks a pivotal step towards overcoming this fundamental hurdle. This system is not a general-purpose quantum computer, but rather a dedicated research platform designed to demonstrate the viability of quantum error correction (QEC) in a real-world hardware setting. Specifically, it focuses on the implementation and characterization of a 'logical qubit' – an error-corrected qubit encoded across multiple physical qubits – using the highly promising surface code architecture.

The core idea behind a logical qubit is to distribute quantum information across an ensemble of physical qubits in such a way that errors affecting individual physical qubits can be detected and corrected without disturbing the encoded quantum state. This redundancy is analogous to how classical error correction codes protect data in conventional computing, but with the added complexity of preserving delicate quantum superpositions and entanglement. The surface code, a type of topological quantum error correction code, is particularly attractive due to its relatively high error threshold and its compatibility with 2D qubit architectures, making it a leading candidate for scalable fault-tolerant quantum computing.

This prototype leverages Google's advanced superconducting transmon qubit technology, specifically utilizing a subset of a Sycamore-class processor. The demonstration of a functional logical qubit, capable of suppressing errors as the code distance increases, provides crucial empirical evidence that the theoretical promises of quantum error correction can be realized in practice. For data analysts, understanding this system means appreciating the intricate interplay between hardware capabilities, error characteristics, and the sophisticated algorithms required to maintain quantum coherence over extended computational periods. It represents a transition point from merely demonstrating quantum phenomena to actively engineering resilience against quantum noise, paving the way for future generations of truly fault-tolerant quantum processors.

The significance of this work cannot be overstated. While the prototype itself is limited in scale and functionality, its success validates the fundamental approach to building reliable quantum computers. It provides concrete data points on the performance of a logical qubit, offering insights into the challenges and opportunities for scaling up these complex systems. The ability to measure and compare error rates for different code distances (d3 vs. d5) directly addresses a core tenet of QEC theory: that increasing redundancy should lead to lower logical error rates. This empirical validation is critical for guiding future hardware design and error correction strategies, moving the field closer to the realization of quantum computers that can tackle problems currently intractable for even the most powerful classical supercomputers.

Key metrics

Logical qubits
1 (2023)
Error-corrected logical qubit using surface code
Benchmark headline
4
Logical error suppression 4% relative d5 vs d3 (2023) | Distance-25 repetition code error 1.7e-6
Error-correction readiness
50/100
Heuristic score from topology + mode + error hints
Historical importance
0/100
Heuristic score from milestones + roadmap language
Native gates
SQ gates | CZ | Measurement | Reset
Gate alphabet you compile to
Connectivity
Distance-5 surface code on 49 physical qubits
Mapping overhead + routing depth sensitivity

Technical specifications

Spec Details
System ID GSCL
Vendor Google Quantum AI
Technology Superconducting transmon qubits
Status Research prototype
Primary metric Logical qubits
Metric meaning Error-corrected logical qubit using surface code
Qubit mode Surface code encodes logical qubit in multiple physical qubits for error detection/correction
Connectivity Distance-5 surface code on 49 physical qubits
Native gates SQ gates | CZ | Measurement | Reset
Error rates & fidelities Logical error per cycle 2.914% d5 vs 3.028% d3 (2023) | Component errors SQ 0.1-0.5% TQ 1-2% Meas 1-2%
Benchmarks Logical error suppression 4% relative d5 vs d3 (2023) | Distance-25 repetition code error 1.7e-6
How to access Internal research only
Platforms Google Quantum AI internal
SDKs Cirq
Regions Santa Barbara CA
Account requirements Google collaborator
Pricing model Not applicable
Example prices Not applicable
Free tier / credits Not applicable
First announced 2023-02-22
First available 2023
Major revisions Distance 3/5 surface | Repetition code
Retired / roadmap Superseded by Willow QEC (2024)
Notes Used Sycamore device; first scaling suppression demo

System profile

Core Technology and Architecture: The Google Surface-code logical prototype is built upon Google Quantum AI's established superconducting transmon qubit technology. These qubits are fabricated on a silicon chip and operate at millikelvin temperatures to minimize thermal noise. The choice of transmon qubits is driven by their relatively long coherence times, high fidelity gate operations, and scalability potential in a 2D lattice. The system specifically employs a subset of a larger Sycamore-class device, demonstrating the integration of error correction within an existing high-performance quantum processor. The physical qubits are arranged in a grid, facilitating the nearest-neighbor interactions essential for surface code implementation.

Logical Qubit Implementation: The primary metric for this system is the demonstration of a single logical qubit. This logical qubit is encoded using a distance-5 surface code, which requires 49 physical qubits. The surface code is a topological error correction scheme where quantum information is stored non-locally across an array of physical qubits. The 'distance' of the code (d) refers to the minimum number of physical qubit errors required to cause a logical error. A higher code distance implies greater error protection. The prototype also explored a distance-3 surface code for comparative analysis, which would typically require fewer physical qubits but offer less protection. The encoding process involves preparing specific entangled states among the physical qubits, and subsequent operations on the logical qubit are performed by applying sequences of gates to these physical qubits, followed by syndrome measurements to detect and correct errors.

Native Gate Set and Connectivity: The system supports a standard set of native gates crucial for universal quantum computation and error correction. These include single-qubit (SQ) gates, two-qubit Controlled-Z (CZ) gates, qubit measurement, and qubit reset operations. The connectivity topology is specifically designed to support the distance-5 surface code, implying a grid-like arrangement where each physical qubit can interact with its nearest neighbors. This local connectivity is a key advantage of the surface code, simplifying the physical wiring and control requirements compared to all-to-all connected architectures.

Performance Metrics and Error Rates: The most critical performance indicator is the logical error rate per cycle. For the distance-5 (d5) surface code, the observed logical error rate was 2.914% per cycle. This was directly compared against a distance-3 (d3) surface code implementation, which exhibited a logical error rate of 3.028% per cycle. The key finding here is the demonstration of logical error suppression: the d5 code showed a 4% relative reduction in logical errors compared to the d3 code. This is a crucial empirical validation that increasing the code distance, and thus the redundancy, indeed leads to improved error protection. While the logical error rate is still relatively high for practical applications, the *trend* of suppression with increasing distance is the fundamental proof of concept.

The underlying physical qubit error rates contribute to these logical error rates. Typical component errors observed were in the range of 0.1-0.5% for single-qubit gates, 1-2% for two-qubit (CZ) gates, and 1-2% for measurement operations. These physical error rates are near the theoretical threshold for the surface code, meaning that with further improvements, significant gains in logical error suppression can be expected. Another benchmark demonstrated was a distance-25 repetition code, achieving an error rate of 1.7e-6. While a repetition code only protects against bit-flip errors (not phase errors), this result showcases the system's ability to maintain very low error rates for specific types of error correction over extended periods.

Operational Limits and Benchmarks: The experiments conducted on this prototype typically involved circuit depths of up to 25 cycles. A 'cycle' in this context refers to a full round of syndrome measurements and error correction operations. This depth is sufficient to demonstrate the error suppression effect over time. Information regarding limits on shots, queueing, or other operational parameters is not publicly confirmed, as this is an internal research prototype. The primary focus was on validating the error correction mechanism rather than offering a general-purpose computational service.

Comparability and Tradeoffs: This prototype's significance lies in its direct empirical comparison of different code distances on the same hardware, a critical step for validating QEC theory. While other quantum computing platforms (e.g., ion traps, neutral atoms) also pursue fault tolerance, Google's demonstration with superconducting transmons highlights the progress in this specific modality. A key tradeoff identified is the impact of correlated errors, such as those potentially induced by high-energy events like cosmic rays, which can affect multiple physical qubits simultaneously and challenge the assumptions of independent errors that many QEC codes rely on. Although the system is operating below the fault-tolerance threshold (meaning errors are still accumulating faster than they are corrected in a fully fault-tolerant manner), it is notably 'near threshold.' This proximity is crucial, as it suggests that incremental improvements in physical qubit quality and control could push the system into a regime where true fault tolerance becomes achievable, allowing for exponential suppression of logical errors. The continuous battle against noise, whether from superconducting circuit imperfections or environmental factors, remains a central challenge, and this prototype provides invaluable data for navigating these complex tradeoffs.

Generation lineage (family-level)
Heuristic chain based on common naming. Verify by revision/date for strict claims.
Related systems (same vendor)
Cross-system comparison (same vendor)
System Status Primary metric
Google Willow Research prototype Physical qubits: 105 (2024)
Google Sycamore Research prototype Physical qubits: 53 (2019)

Access & pricing

How you access it
  • Public access to this prototype is currently unavailable.
  • Access is restricted to internal research teams and Google collaborators only.
  • The system operates exclusively within Google Quantum AI's internal platforms.
  • The hardware is located at Google's facilities in Santa Barbara, California.
  • Development and interaction with the system primarily utilize the Cirq SDK.
  • Account requirements necessitate being a Google collaborator or part of the internal research team.
  • This system functions as a prototype for demonstration and research purposes, not for external public use.
How costs sneak up
  • Public pricing for this research prototype is not applicable.
  • The system does not operate under a commercial pricing model.
  • There are no example prices or cost structures available for external users.
  • Cost drivers are internal research and development expenses, not user-based charges.
  • No free tier or credits are offered, as it is not a commercial service.
  • Pricing notes indicate that this system is exclusively for internal research and development, with no public commercial offering.

Status timeline

The journey towards fault-tolerant quantum computing has been a long and arduous one, rooted in theoretical breakthroughs from the 1990s, such as Peter Shor's algorithm and the development of quantum error correction codes. Google Quantum AI has been a significant player in this field, notably demonstrating 'quantum supremacy' in 2019 with its Sycamore processor, showcasing the ability of a quantum computer to perform a task intractable for classical supercomputers. This achievement, while groundbreaking, did not inherently address the issue of quantum errors, setting the stage for the next critical phase: fault tolerance.

February 22, 2023: Initial Announcement
Google Quantum AI formally announced the successful demonstration of a logical qubit prototype using a surface code on a Sycamore-class processor. This announcement was accompanied by a publication in Nature, detailing the experimental results and the significance of achieving error suppression with increasing code distance. This marked a crucial milestone, moving beyond theoretical proposals to empirical validation of quantum error correction principles on a physical hardware system.

2023: First Availability and Key Experiments
Throughout 2023, the prototype was actively used for internal research, culminating in the published results. The core experiments focused on:

  • Distance-3 vs. Distance-5 Surface Code Comparison: The most significant demonstration involved comparing the performance of a logical qubit encoded with a distance-3 surface code against one encoded with a distance-5 surface code. Both implementations utilized a subset of 49 physical superconducting transmon qubits. The experiment showed that the distance-5 code achieved a logical error rate of 2.914% per cycle, which was a 4% relative improvement over the 3.028% per cycle observed for the distance-3 code. This direct empirical evidence of error suppression with increasing code distance was a critical validation of the surface code's theoretical scaling properties.
  • Repetition Code Benchmarking: In parallel, the team also demonstrated a distance-25 repetition code, achieving an impressive error rate of 1.7e-6. While repetition codes only protect against bit-flip errors and not the full spectrum of quantum errors, this experiment showcased the high fidelity achievable for specific error correction tasks and provided valuable insights into the underlying noise characteristics of the hardware.
  • Validation of Error Correction Mechanisms: The experiments meticulously characterized the physical error rates (single-qubit gates, two-qubit gates, measurements) and their impact on logical qubit performance. This detailed analysis confirmed that the system was operating near the theoretical fault-tolerance threshold for the surface code, indicating that further incremental improvements in physical qubit quality could lead to exponential suppression of logical errors.

Major Revisions and Iterations: The development of this prototype involved continuous refinement of both the hardware and the control software. This included optimizing qubit connectivity, improving gate fidelities, and enhancing the classical control electronics necessary for rapid syndrome extraction and error correction. The comparison between distance-3 and distance-5 codes itself represents a major revision in experimental design, directly testing the scaling behavior of the surface code.

2024: Superseded by Willow QEC Roadmap
As of 2024, the specific roadmap for this prototype has been superseded by Google's broader 'Willow' Quantum Error Correction (QEC) initiative. This indicates a natural progression in research and development. The lessons learned and the empirical data gathered from the 2023 surface-code logical prototype are directly informing the design and implementation of the next generation of QEC hardware. The Willow roadmap likely aims to build upon these foundational demonstrations, potentially exploring larger code distances, more complex logical operations, and more robust error correction architectures, moving closer to a truly fault-tolerant quantum computer. This iterative approach is typical in cutting-edge research, where successful prototypes pave the way for more ambitious and capable successors.

What to verify next

  • Investigate and mitigate the impact of high-energy events, such as cosmic rays, which can cause correlated errors across multiple physical qubits, challenging current error correction models.
  • Explore the implementation and performance of larger code distances beyond d5 to demonstrate stronger error suppression and approach the true fault-tolerance threshold.
  • Characterize the performance of logical gates on the logical qubit, moving beyond simple state preparation and measurement to demonstrate universal logical operations.
  • Assess the overheads associated with scaling up the number of logical qubits and the complexity of the classical control systems required for real-time error correction.
  • Evaluate the long-term coherence and stability of logical qubits under continuous operation, including the effects of repeated error correction cycles.
  • Compare the performance of surface codes with alternative QEC codes or hybrid approaches to identify the most efficient path to fault tolerance.
  • Develop and test more sophisticated decoding algorithms that can handle complex error patterns and reduce the latency of error correction.

FAQ

What is a logical qubit and why is it important?

A logical qubit is an error-corrected quantum bit, encoded across multiple physical qubits. It's crucial because physical qubits are highly susceptible to noise and errors. By distributing quantum information redundantly, a logical qubit can detect and correct errors without corrupting the encoded quantum state, making large-scale, reliable quantum computation possible.

What is the surface code, and how does it work in this prototype?

The surface code is a type of topological quantum error correction code. It works by arranging physical qubits in a 2D grid and performing local measurements (syndrome measurements) to detect errors without directly measuring the logical qubit's state. This prototype used a distance-5 surface code, meaning it required 49 physical qubits to encode one logical qubit, providing greater error protection than a distance-3 code.

What were the key results and benchmarks demonstrated by this prototype?

The primary result was the demonstration of logical error suppression: the distance-5 surface code achieved a 4% relative reduction in logical errors per cycle compared to a distance-3 code (2.914% vs 3.028%). This empirically validated that increasing code distance improves error protection. Additionally, a distance-25 repetition code achieved an error rate of 1.7e-6, showcasing high fidelity for specific error types.

Is the Google Surface-code logical prototype publicly accessible?

No, this system is a research prototype and is not publicly accessible. Access is restricted to Google Quantum AI's internal research teams and collaborators. It is used for internal development and demonstration purposes rather than as a commercial service.

How many physical qubits were used to create the logical qubit?

The distance-5 surface code implementation on this prototype utilized 49 physical superconducting transmon qubits to encode a single logical qubit. The surface code's design requires a significant number of physical qubits to achieve redundancy and error protection for even one logical qubit.

What are the main challenges or tradeoffs observed with this system?

One significant challenge is the impact of correlated errors, such as those potentially caused by high-energy events like cosmic rays, which can affect multiple physical qubits simultaneously. While the system is 'near threshold,' it is still operating below the fault-tolerance threshold, meaning further improvements in physical qubit quality and control are needed for exponential error suppression. Balancing physical qubit performance with the overheads of error correction remains a key tradeoff.

What is the future outlook for this prototype, given the 'retired roadmap' status?

The specific roadmap for this prototype has been superseded by Google's broader 'Willow' Quantum Error Correction (QEC) initiative as of 2024. This indicates that the valuable insights and empirical data gained from this prototype are directly informing the design and development of Google's next generation of QEC hardware, aiming for larger code distances and more robust fault-tolerant systems.



Also in Quantum Computing

Xanadu X8

Xanadu X12

Xanadu Borealis

Subscribe