Quantum Resource Estimator

RSA-2048 with Gidney's 2025 algorithm

In 2025 Craig Gidney cut the published cost of factoring RSA-2048 from 20 million noisy qubits to under one million, at the price of a longer run (under a week). This page runs his logical-level figures through the standard surface-code model so you can see how much of the saving comes from the algorithm and how much from newer error-correction tricks.

How many qubits does Gidney's 2025 RSA-2048 algorithm need?

Using 1,399 logical qubits and 6.50 billion Toffoli gates, a surface-code machine at 0.1% physical error and 1 microsecond cycles needs code distance 27, about 3.56 million - 4.49 million physical qubits, and runs for about 2.03 days (serial depth). Order-of-magnitude planning estimate.

Precomputed with model version 2: 33% error budget, serial T-depth. Source: Gidney 2025 (arXiv:2505.15917), Table 5: 1,399 logical qubits and 6.5e9 Toffoli gates for n = 2048 using approximate residue arithmetic. The paper reaches <1M physical qubits in <1 week with yoked surface codes and magic state cultivation; this tool models plain surface codes and 15-to-1 factories, so expect a higher qubit count and a shorter runtime here (the paper deliberately uses few factories, trading time for space).
Physical error rateHardwareCode distancePhysical qubitsRuntime
0.1%superconducting, 1 µs273.56 million - 4.49 million2.03 days
0.05%superconducting, 1 µs212.16 million - 2.71 million38 hours
0.01%superconducting, 1 µs13826,072 - 1.04 million23 hours
0.001%superconducting, 1 µs9395,928 - 498,60416 hours
0.1%trapped ion, 200 µs273.56 million - 4.49 million406 days
0.05%trapped ion, 200 µs212.16 million - 2.71 million316 days
0.01%trapped ion, 200 µs13826,072 - 1.04 million196 days
0.001%trapped ion, 200 µs9395,928 - 498,604135 days

Try your own assumptions

Physical qubits (total)
Estimated runtime
Code distance
Data-block physical qubits
T-gate count
Physical per logical

Highlighted share: magic-state factories (conservative bound shown).

How the model works and where every constant comes from: methodology and sources.