Azure AI Foundry — SKU Constellation Map

What a Head of AI/ML or CTO needs to know about Azure AI Foundry as a SKU — the three cost stacks (tokens, endpoints, training), responsible-AI tooling, and the positioning against Databricks Mosaic AI and Copilot Studio.

BusinessCapabilityTechnology
Compass
  • Businesspersona, use case, outcome
  • Capabilitywhat the org needs to do
  • Technologythe technology choices
Guided journey · Step 1 of 4

Azure AI Foundry — SKU Anchor

Model the three cost stacks separately — token, endpoint, training. Combining them hides the surprises; tracking them produces governable cost.

~ 3 weeks

Search any SKU, capability, risk, or source on this map.

Filter by type

Narrative intro

Azure AI Foundry is the pro-code AI development surface — Azure OpenAI, the model catalogue, agent service, prompt flow, AI Search, the Azure ML lifecycle. The procurement story is three independent cost stacks. The operating story is responsible AI as a build-time discipline. The strategic story is the boundary against Databricks Mosaic AI (lakehouse-native) and Copilot Studio (low-code). This map sits the SKU inside those decisions.

Key takeaways

  • Three cost stacks — token, endpoint, training — model and track separately
  • Responsible AI tooling belongs in the build workflow, not as post-launch audit
  • Foundry vs Mosaic AI is workload-fit, not vendor-preference
  • Foundry vs Copilot Studio is pro-code vs low-code — both can be valid in the same estate

Programme shape

Estimated duration
10–20 weeks
Estimated FTE
1 FTE AI/ML platform lead + part-time security and governance SMEs
Spend tier
significant
Risk level
elevated

Three-stack cost model is the dominant procurement risk. Responsible AI tooling needs to sit in the build workflow, not as a post-launch audit.

Back to all maps