Quantum Advantage Economics and Kill Criteria — Operating Quantum Computers

Quantum computational advantage is a bounded claim that a quantum method outperforms the best relevant classical methods on a defined task and metric under stated constraints. Portfolio value is broader: faster learning, reduced risk, and strategic options can justify research without demonstrating computational advantage. This chapter evaluates both, while keeping those claims separate.

This chapter turns quantum advantage into a portfolio-management and kill-criteria discipline.

Diagram loads as you read

Quantum Advantage Economics and Kill Criteria · Figure 1

View diagram source

flowchart LR
    Idea[Use-case idea] --> Baseline[Classical baseline]
    Baseline --> Quantum[Quantum candidate]
    Quantum --> Evidence[Evidence package]
    Evidence --> Value[Decision value]
    Value --> Gate{Continue?}
    Gate -- yes --> Invest[Increase investment]
    Gate -- no --> Kill[Kill or park]

Advantage must be defined against a baseline

A quantum workflow must be compared with relevant alternatives. A computational-advantage claim requires strong classical algorithms for the same task, accuracy and resource accounting, including applicable heuristics and approximations. Comparisons with laboratory experiments or business-as-usual processes can establish practical value; they do not by themselves establish computational advantage.

Baseline	Quantum must improve
classical exact solver	scale, cost, or time-to-answer
classical heuristic	solution quality, robustness, or explanation
physical experiment	experiment planning, search efficiency, or risk
manual process	decision speed or consistency
no current capability	feasibility or strategic option value

Diagram loads as you read

Advantage must be defined against a baseline · Figure 2

View diagram source

flowchart TB
    Candidate[Quantum candidate] --> Compare[Compare against baseline]
    BaselineA[Exact classical] --> Compare
    BaselineB[Heuristic classical] --> Compare
    BaselineC[Physical experiment] --> Compare
    BaselineD[Business process] --> Compare
    Compare --> Advantage[Bounded advantage statement]

Unit economics

The economic unit should be “validated result” or “decision supported,” not “shot,” “job,” or “circuit.” Jobs and shots are inputs. Validated decisions are outputs.

Diagram loads as you read

Unit economics · Figure 3

View diagram source

flowchart LR
    Spend[Spend] --> QPU[QPU time]
    Spend --> Staff[Staff time]
    Spend --> Compute[Classical compute]
    Spend --> Evidence[Evidence and review]
    QPU --> Validated[Validated result]
    Staff --> Validated
    Compute --> Validated
    Evidence --> Validated
    Validated --> Decision[Decision value]

Cost element	Hidden multiplier
QPU access	queue delay, reservation waste
shot budget	mitigation overhead
staff time	experiment design and debugging
classical compute	simulation, optimization, post-processing
governance	review, audit, evidence retention
opportunity cost	delayed classical alternative

Kill criteria

Kill criteria make quantum programs stronger. They prevent sunk-cost behavior and free resources for better candidates.

Diagram loads as you read

Kill criteria · Figure 4

View diagram source

flowchart TD
    Gate[Portfolio gate] --> Evidence{Evidence improved?}
    Evidence -- no --> Baseline{Baseline beaten?}
    Evidence -- yes --> Continue[Continue]
    Baseline -- no --> Kill[Kill or park]
    Baseline -- yes --> Risk{Risk acceptable?}
    Risk -- yes --> Continue
    Risk -- no --> Redesign[Redesign]

Examples:

Gate	Kill or park if
feasibility	workload cannot fit target envelope within planned horizon
baseline	quantum path cannot beat agreed classical baseline on any useful dimension
evidence	result remains unreproducible after defined attempts
economics	cost per validated decision exceeds threshold
strategic	use case no longer maps to business or research priority

Real options framing

Some quantum investments are options, not immediate ROI projects. An option is valuable if it preserves future capability under uncertainty.

Diagram loads as you read

Real options framing · Figure 5

View diagram source

flowchart LR
    Investment[Small investment] --> Learning[Learning]
    Learning --> Option[Future option]
    Option --> Trigger{Trigger event?}
    Trigger -- hardware improves --> Exercise[Scale program]
    Trigger -- no progress --> Expire[Let option expire]

Option investments still need kill rules. A technology-watch project can be small and useful; it should not silently become a large program without evidence.

Portfolio tiers

Diagram loads as you read

Portfolio tiers · Figure 6

View diagram source

quadrantChart
    title Quantum portfolio map
    x-axis Low strategic value --> High strategic value
    y-axis Low evidence --> High evidence
    quadrant-1 Scale selectively
    quadrant-2 Watch and learn
    quadrant-3 Kill or archive
    quadrant-4 Harden evidence
    Chemistry surrogate: [0.75, 0.55]
    Optimization demo: [0.40, 0.35]
    PQC migration: [0.85, 0.90]
    Public benchmark: [0.65, 0.70]

The portfolio should include near-term operationally valuable work, quantum-safe security work, strategic learning, and carefully bounded research. It should not be a collection of demos.

Advantage statement template

Illustrative listing · text

For workload class <X>, against baseline <Y>, under constraints <Z>,
the quantum-enabled workflow provides <value dimension> with evidence level <E>,
at estimated cost <C>, by date or maturity trigger <T>.

Operating rule

Quantum advantage is a managed claim. Define the baseline, unit economics, evidence threshold, and kill criteria before expanding investment.

Additional technical sources: [R270].