JPCUB Competitive Landscape v2.0: System-Level Joules-per-Solution Estimates for 17 Quantum Computing Platforms from Published Specifications

In the paper's words

The JPCUB P0 protocol (https://papers.qnfo.org/papers/joules-per-solution-metric/) defines the joules-per-solution metric — total system energy per correct answer — as a universal, physics-grounded benchmark for computational platforms. The qwav.tech competitive landscape displays six platforms with one published measurement (IBM Eagle at $0.89$ J/solution) and five entries listed as "Not yet measured" or "design target." This paper v2.0 expands the landscape to 17 platforms — 13 gate-model quantum processors spanning three architectures (superconducting, trapped-ion, neutral-atom) from 7 vendors, plus 4 non-gate-model or pre-commercial entries. JPCUB estimates are computed from published specifications using a consistent system-level power model.

The abstract, verbatim, and its opening sentence is the claim. Every line below is quoted from the text.

Numbers it reports

1. A 15 kW superconducting platform with 300 ns gates outperforms a 3.5 kW trapped-ion platform with 100 $\mu$s gates by approximately $100\times$ on JPCUB, because the factor-of-333 gate-time difference overwhelms the factor-of-4.3 power advantage.

   from 1.4 Core Claim

2. Correctness threshold $\varepsilon$: $0.95$ (95% success rate).

   from 2.1 Task Specification

3. Platform Qubits $P_{\mathrm{sys}}$ 1Q Gate 2Q Gate 2Q Fidelity Source Google Willow 105 25.0 kW 20 ns 30 ns 99.95% Google Sycamore 53 25.0 kW 25 ns 40 ns 99.8% IQM Garnet 20 12.0 kW 100 ns 200 ns 99.5% IBM Heron r2 133 15.0 kW 170 ns 300 ns 99.7% IBM Eagle r3 127 15.0 kW 288 ns 500 ns 99.0% $[$established — JPCUB P0 published at 0.89 J/sol$]$ Rigetti Ankaa-3 84 15.0 kW 200 ns 400 ns 98.0% Rigetti Aspen-M-3 80 15.0 kW 200 ns 400 ns 97.5% Architecture notes: All seven platforms share the same physical infrastructure — superconducting transmon qubits operating at approximately 15 mK in a dilution refrigerator.

   from 3.1 Superconducting Transmon (7 platforms)

4. Differences in JPCUB derive from gate speed (20–500 ns for two-qubit gates) and gate fidelity (97.5–99.95%).

   from 3.1 Superconducting Transmon (7 platforms)

5. Platform Qubits $P_{\mathrm{sys}}$ 1Q Gate 2Q Gate 2Q Fidelity Source Quantinuum H1-1 20 4.0 kW $10 \mu s$ $50 \mu s$ 99.8% Quantinuum H2 56 4.5 kW $10 \mu s$ $50 \mu s$ 99.8% IonQ Aria 25 3.0 kW $20 \mu s$ $100 \mu s$ 99.4% IonQ Forte 36 3.5 kW $20 \mu s$ $100 \mu s$ 99.5% Architecture notes: Trapped-ion systems operate at room temperature — no dilution refrigerator, no millikelvin cryogenics.

   from 3.2 Trapped Ions (4 platforms)

6. Platform Atoms $P_{\mathrm{sys}}$ 1Q Gate 2Q Gate 2Q Fidelity Source QuEra Aquila 256 4.0 kW 500 ns $1.5 \mu s$ 99.5% Pasqal Fresnel 100+ 4.0 kW 500 ns $2.0 \mu s$ 98.0% Architecture notes: Neutral-atom platforms use Rydberg blockade for two-qubit entanglement — gates in the 1–2 $\mu s$ range, significantly faster than trapped ions but approximately 5–50× slower than the best superconducting gates.

   from 3.3 Neutral Atoms (2 platforms)

How it is done

1. The JPCUB P0 protocol (https://papers.qnfo.org/papers/joules-per-solution-metric/) defines the joules-per-solution metric — total system energy per correct answer — as a universal, physics-grounded benchmark for computational platforms.

   from Abstract

2. The joules-per-solution metric $J_S(T, \varepsilon)$ $[$established$]$ defines the total system-level energy (joules) consumed by system $S$ to produce a solution to computational task $T$ at correctness threshold $\varepsilon$: $$J_S(T, \varepsilon) = E_S(T, \varepsilon) = E_{\text{comp}} + E_{\text{mem}} + E_{\text{io}} + E_{\text{cool}} + E_{\text{conv}} + E_{\text{mfg}}$$ The protocol is published, open, and falsifiable.

   from 1.1 The JPCUB Metric

3. Applies consistent methodology — the same system-level power model to all 13 gate-model platforms, using the same task ($N = 15$, Shor's algorithm, $\varepsilon = 0.95$) and the same gate-count model.

   from 1.2 The Gap in the qwav.tech Landscape

4. It would be disconfirmed if independent JPCUB measurement following the P0 protocol on at least one trapped-ion platform and at least one superconducting platform showed the trapped-ion platform achieving lower joules-per-solution than the superconducting platform for the same task. $[$speculative$]$

   from 1.4 Core Claim

What would falsify it

1. The joules-per-solution metric $J_S(T, \varepsilon)$ $[$established$]$ defines the total system-level energy (joules) consumed by system $S$ to produce a solution to computational task $T$ at correctness threshold $\varepsilon$: $$J_S(T, \varepsilon) = E_S(T, \varepsilon) = E_{\text{comp}} + E_{\text{mem}} + E_{\text{io}} + E_{\text{cool}} + E_{\text{conv}} + E_{\text{mfg}}$$ The protocol is published, open, and falsifiable.

   from 1.1 The JPCUB Metric

2. It would be disconfirmed if independent JPCUB measurement following the P0 protocol on at least one trapped-ion platform and at least one superconducting platform showed the trapped-ion platform achieving lower joules-per-solution than the superconducting platform for the same task. $[$speculative$]$

   from 1.4 Core Claim

3. This has implications for the JPCUB calibration register [, §6]: the CAL-01 prediction ("No gate-model quantum computer will solve a commercially relevant problem at lower joules-per-solution than the best classical alternative by 2030") must account for both gate-speed and fidelity trajectories.

   from 6.2 Fidelity as the Second Factor

Where the claim stops

1. This finding has direct implications for the JPCUB research program: the P1 quantum energy audit must account for both gate-speed and cooling-power trajectories across modalities; the P9 Comparative Atlas must include an explicit time–power decomposition for each paradigm; and the CAL-01 prediction ("no quantum computer will beat classical on JPCUB by 2030") must specify gate-speed assumptions as well as fidelity assumptions.

   from 10. Conclusion

2. The qwav.tech competitive landscape displays six platforms with one published measurement (IBM Eagle at $0.89$ J/solution) and five entries listed as "Not yet measured" or "design target." This paper v2.0 expands the landscape to 17 platforms — 13 gate-model quantum processors spanning three architectures (superconducting, trapped-ion, neutral-atom) from 7 vendors, plus 4 non-gate-model or pre-commercial entries.

   from Abstract

3. Trapped-ion platforms, despite room-temperature operation eliminating the dilution refrigerator, rank lowest because microsecond-scale gate times (approximately $125\times$ slower than superconducting) produce execution energies that overwhelm the cooling-power advantage.

   from Abstract

Where to read

1. Abstract209 words · 1 min
2. 1. Introduction366 words · 2 min
3. 2. Methodology310 words · 1 min
4. 3. Platform Specifications402 words · 2 min
5. 4. Results: JPCUB Estimates for 13 Gate-Model Platforms296 words · 1 min
6. 5. Architecture Group Analysis511 words · 2 min
7. 6. Discussion816 words · 4 min
8. 7. Limitations182 words · 1 min
9. 8. JPCUB Baselines: Existing Computing Architectures530 words · 2 min
10. 9. Total Joules for Classically-Infeasible Problems1,165 words · 5 min
11. 10. Conclusion263 words · 1 min
12. Declarations193 words · 1 min