Platform Economics, Chargeback, and Executive Controls

Quantum platforms can fail economically even when the science is interesting. Queue reservations, provider subscriptions, simulator GPU time, staff review hours, and repeated reruns can create cost without decision value. A serious program needs economic controls that connect usage to outcomes.

FinOps chargeback and allocation practices provide a useful pattern: costs should map to owners, cost centers, products, or value streams, and finance needs a source of truth for invoice reconciliation and allocation [R226]. Quantum operations should extend that model with quantum-specific cost units.

Diagram loads as you read

Platform Economics, Chargeback, and Executive Controls · Figure 1

View diagram source

flowchart LR
    Usage[Quantum usage] --> Meter[Metering]
    Meter --> Allocate[Allocation]
    Allocate --> Showback[Showback]
    Showback --> Chargeback[Chargeback]
    Chargeback --> Portfolio[Portfolio decision]
    Portfolio --> Budget[Budget adjustment]

Quantum cost units

Cost unit	Why it matters
reservation hour	scarce QPU access commitment
shot	direct execution volume
circuit-second	rough hardware occupancy proxy
logical operation estimate	fault-tolerant planning unit
simulator GPU-hour	classical validation cost
review hour	governance and domain expertise
revalidation cycle	hidden cost of weak evidence

Diagram loads as you read

Quantum cost units · Figure 2

View diagram source

flowchart TB
    TotalCost[Total platform cost] --> QPU[QPU access]
    TotalCost --> Simulation[Simulation]
    TotalCost --> Storage[Storage and evidence]
    TotalCost --> People[Review and operations]
    TotalCost --> Tooling[Tooling and CI]
    TotalCost --> Risk[Risk and compliance]

Allocation model

The cost ledger should carry workload, owner, project, provider, backend, reservation, evidence tier, and claim ID.

Diagram loads as you read

Allocation model · Figure 3

View diagram source

classDiagram
    class CostRecord {
      cost_record_id
      workload_id
      owner
      cost_center
      provider
      backend
      cost_unit
      amount
      claim_id
    }
    class Budget {
      budget_id
      owner
      period
      limit
      threshold
    }
    class DecisionRecord {
      decision_id
      value_claim
      kill_criteria
    }
    CostRecord --> Budget
    CostRecord --> DecisionRecord

Budget gates

Diagram loads as you read

Budget gates · Figure 4

View diagram source

flowchart LR
    Request[Execution request] --> Estimate[Cost estimate]
    Estimate --> Budget{Budget available?}
    Budget -- no --> Reject[Reject or escalate]
    Budget -- yes --> Value{Value case present?}
    Value -- no --> Sandbox[Sandbox limits]
    Value -- yes --> Run[Run with meter]

Budget gates should not punish exploration, but they should prevent uncontrolled iteration loops. Exploratory workloads can use small sandbox budgets. Production-adjacent claims require explicit value cases and kill criteria.

Executive dashboard

Executives need fewer quantum metrics, not more. The dashboard should connect spend to learning, capability, and decisions.

Diagram loads as you read

Executive dashboard · Figure 5

View diagram source

flowchart TB
    Spend[Spend] --> UnitCost[Unit cost]
    Spend --> Outcomes[Validated outcomes]
    Outcomes --> Decisions[Decisions enabled]
    Outcomes --> Claims[Claims retired or advanced]
    UnitCost --> Efficiency[Efficiency trend]
    Decisions --> Portfolio[Portfolio allocation]

Useful executive questions:

Question	Metric
Are we learning faster?	time from idea to validated result
Are costs controlled?	cost per accepted evidence package
Are vendors improving?	accepted-result rate by provider
Are claims converging?	claims advanced, held, or killed
Are we reducing risk?	revalidation burden and defect impact

Kill criteria

Quantum programs need explicit kill criteria because technical optimism can otherwise become a budget sink.

Diagram loads as you read

Kill criteria · Figure 6

View diagram source

stateDiagram-v2
    [*] --> Hypothesis
    Hypothesis --> Sandbox
    Sandbox --> EvidenceTier1
    EvidenceTier1 --> EvidenceTier2
    EvidenceTier2 --> ProductionCandidate
    EvidenceTier1 --> Kill: cost or confidence fail
    EvidenceTier2 --> Kill: classical baseline wins
    ProductionCandidate --> Kill: no adoption path

Kill criteria should compare against classical baselines, opportunity cost, data availability, evidence confidence, and customer value.

Chargeback caveat

Chargeback can create bad incentives if teams avoid necessary validation to save budget. Showback may be better early. Chargeback should be introduced only when cost attribution is reliable and governance does not discourage scientific rigor.

Diagram loads as you read

Chargeback caveat · Figure 7

View diagram source

flowchart LR
    Immature[Immature metering] --> Showback[Showback]
    Showback --> Improve[Improve allocation]
    Improve --> ChargebackReady{Chargeback ready?}
    ChargebackReady -- no --> Showback
    ChargebackReady -- yes --> Chargeback[Chargeback]

Operating rule

Spend is not justified by quantum activity. Spend is justified by bounded learning, retired uncertainty, validated capability, or decision value. Cost records should therefore connect to evidence and portfolio decisions, not only provider invoices.