Skip to main content

Economic & Resource Model

AI Products must declare their economic and resource consumption model.
This ensures consumers understand the cost, efficiency, and sustainability implications of adoption.

Why This Matters​

  • Transparency → Clear view of costs avoids hidden risks.
  • Governance → Resource-heavy products must justify economic trade-offs.
  • Sustainability → AI Products consume significant compute and energy, which must be managed responsibly.
  • Fairness → Supports equitable access across consumers, tenants, or organizations.

Resource Model Requirements​

AI Products must declare:

  • Compute requirements → CPU, GPU, TPU, memory, storage.
  • Elasticity → how resource use scales with load (see Inference & Scaling).
  • Energy usage → estimated carbon footprint or efficiency indicators.
  • Network requirements → bandwidth, latency sensitivity.
  • External dependencies → data sources, orchestration frameworks, specialized hardware.

Economic Model Requirements​

AI Products must declare their cost structure, including:

  • Consumption-based pricing → cost per request, per token, per inference, etc.
  • Provisioned capacity → reserved quotas with predictable cost.
  • Tiered access → free, premium, or enterprise plans.
  • Shared vs Dedicated resources → multi-tenant vs single-tenant cost models.
  • Sustainability disclosures → energy and environmental impact reporting.

Example​

LLM Product

  • Resource Model: Requires 1 GPU (A100 class) per replica, 20 GB memory.
  • Elasticity: Scales horizontally up to 50 replicas.
  • Energy Usage: Estimated 0.4 kWh per 1K tokens generated.
  • Economic Model:
    • $0.002 per token (consumption-based).
    • Enterprise tier with reserved GPU capacity for flat monthly cost.
    • Sustainability report published quarterly.

Governance Integration​

  • Costs must align with organizational economic and sustainability policies.
  • Products must declare resource and cost models in the catalog/marketplace entry (see Discoverability).
  • Economic reporting may be audited for fairness and transparency.

Summary​

  • AI Products must declare resource and economic models.
  • Consumers need clarity on compute, storage, energy use, and cost structures.
  • Transparency in economics and sustainability is part of being a true AI Product.

Principle: An AI Product without a transparent economic and resource model risks being unsustainable, inequitable, or untrustworthy.