What Is a Gigaquop?
A gigaquop is an informal name for a quantum computer capable of executing on the order of one billion (\(10^9\)) quantum operations; naively, that corresponds to a logical fault probability per operation of order \(10^{-9}\) for an order-one total failure probability. It sits three orders of magnitude above the megaquop, the informal term physicist John Preskill proposed for machines beyond NISQ on the road toward fault-tolerant, application-scale quantum computers.
The unit counts operations, not qubits. A "quop" is one quantum operation, such as a gate or a measurement. If each operation fails with probability \(p\), the probability that no operation fails is approximately
\[ (1-p)^N \approx e^{-Np}. \]
Thus \(Np\sim1\) marks an order-one chance of at least one fault, while a high probability of a clean run requires \(Np\ll1\).
So a gigaquop machine corresponds, at the order-of-magnitude level, to a logical fault probability per operation of \(10^{-9}\) or lower, depending on the target success probability and error model. That is the logical error rate that error correction is meant to deliver.
"Gigaquop" is shorthand for a reliable logical-operation budget, or circuit volume. It is not a certified benchmark, and no standards body defines it.
Gigaquop, Megaquop and Teraquop: A Ladder of Reliable Depth
The SI prefixes mark rungs on one ladder. Preskill's 2018 paper introduced NISQ (noisy intermediate-scale quantum) devices, which run without error correction. His 2025 paper on the megaquop machine proposed a target near \(10^6\) operations on the way to full fault tolerance. The gigaquop and teraquop follow the same prefix pattern, and are used here in that sense.
| Regime | Reliable operations | Per-operation error for \(Np\sim1\), not a high-success guarantee |
|---|---|---|
| Illustrative uncorrected circuit | e.g. \(10^3\) physical operation locations | \(p\ll10^{-3}\) for high probability of no faults; \(p\sim10^{-3}\) gives order-one fault probability |
| Megaquop | \(10^6\) | \(10^{-6}\) |
| Gigaquop | \(10^9\) | \(10^{-9}\) |
| Teraquop | \(10^{12}\) | \(10^{-12}\) |
Each rung is a thousandfold increase in how long a computation can run, which is why the ladder says more about usefulness than any qubit total does.
What a Gigaquop Demands From Error Correction
Physical qubits are far too noisy for a billion operations, so a gigaquop machine relies on quantum error correction to build logical qubits. For a code such as the surface code, once the physical error rate \(p\) is below a threshold \(p_{\mathrm{th}}\), the logical error rate falls roughly exponentially with code distance \(d\):
\[ p_L \approx A\left(\frac{p}{p_{\mathrm{th}}}\right)^{(d+1)/2} \]
Here \(A\) is a constant of order one. This is a heuristic, but it shows the trade: reaching \(10^{-9}\) generally takes a larger code distance, better physical error rates, or a code and architecture that achieve the same logical error rate with lower overhead.
Three costs tend to dominate:
- Non-Clifford operations. Operations such as the T gate cannot be done simply on most codes, and usually rely on magic state distillation or related methods, which consume space and time.
- Real-time decoding. Classical decoders must keep pace with every round of syndrome measurements.
- Calibration and control. Errors must stay low and stable across thousands of physical qubits for the length of the run.
Which Algorithms Sit Near This Scale?
Resource estimates give a feel for the numbers, with the caveat that they change as methods improve. Gidney and Ekerå (2021) estimated that Shor's algorithm could factor a 2048-bit RSA integer in about eight hours using about 20 million noisy physical qubits, with a Toffoli count on the order of a billion.
A Toffoli count is not a quop count. A Toffoli count is not, by itself, a full logical-operation count: the circuit also contains Clifford gates and measurements, and a chosen fault-tolerant implementation decomposes each Toffoli or CCZ into logical operations and physical error-correction cycles.
Estimates for quantum chemistry on industrially relevant molecules also reach into the billions of operations or beyond, depending heavily on the algorithm chosen. Depth is only half the story: these workloads need many logical qubits at once, which the quop count does not capture.
Where QuEra's Roadmap Meets the Gigaquop Scale
QuEra's blog post on the software stack fault tolerance needs states that Libra is designed for 256 error-corrected logical qubits built from more than 10,000 physical qubits at a \(10^{-6}\) logical error rate, arriving on Amazon Braket in 2028. The same post says a next-generation system aimed at the gigaquop scale is planned for 2028/2029.
By the \(Np\sim1\) heuristic above, a rate of \(10^{-6}\) corresponds to a million-operation budget only at the order-of-magnitude, order-one-failure-probability level, so a billion is a separate target and not implied by those design figures. The post also makes the broader point that qubit counts are only the entry ticket: compilers, schedulers, calibration specifications and trustworthy simulators decide whether a machine returns a useful answer.
FAQ
Is a gigaquop the same as a billion-qubit computer?
No. A gigaquop counts reliable operations, about \(10^9\), not qubits. A machine with far fewer qubits could in principle run a billion reliable operations, and a machine with many noisy qubits could not.
How is a gigaquop different from a megaquop?
A megaquop and a gigaquop are informal labels for machines capable of executing on the order of \(10^6\) and \(10^9\) quantum operations, respectively. Under the \(N \lesssim 1/p\) heuristic, that means cutting the error rate per operation from about \(10^{-6}\) to about \(10^{-9}\).
Has any machine reached a gigaquop?
To our knowledge, no machine has publicly demonstrated a billion reliable logical operations. Error-correction experiments so far have shown logical error suppression on small numbers of logical qubits, at depths far below this scale.
Would a gigaquop machine be able to break RSA?
Not by that fact alone. The peer-reviewed Gidney-Ekerå 2021 estimate for 2048-bit RSA uses about \(2.6\times10^9\) Toffoli gates and about \(2\times10^7\) noisy physical qubits under its surface-code assumptions; later estimates change the qubit-runtime tradeoff. Depth and width both matter, and the estimates keep changing.
Key Takeaways
- A gigaquop is an informal label for a machine that can execute on the order of \(10^9\) quantum operations; the naive order-of-magnitude target is a logical fault probability per operation of \(10^{-9}\) or lower, depending on the desired success probability.
- The unit measures logical-operation budget, not qubit count, and it is informal shorthand rather than a certified benchmark.
- Reaching it relies on fault-tolerant quantum computing: larger code distances, magic state distillation and real-time decoding.
- QuEra's roadmap post places a next-generation system aimed at this scale in 2028/2029, after Libra on Amazon Braket in 2028.
.webp)
