Cost Model Breakdown¶
Overview¶
TenantZero AI deploys Azure-native services in the customer's own subscription. All costs are incurred directly on the customer's Azure billing, with no vendor markup or SaaS fees.
Cost Components by Service¶
Networking (Fixed)¶
| Resource | SKU | Cost Driver |
|---|---|---|
| Virtual Network | Free | No charge for VNet |
| Subnets | Free | No charge for subnets |
| NSGs | Free | No charge for NSGs |
| Private DNS Zones | Per zone/month | ~$0.50/zone/month |
| Private Endpoints | Per endpoint/month | ~$0.01/hour per PE |
Estimated Monthly: $25–$50 (4 PE + 4 DNS zones)
AI Services (Variable)¶
| Resource | SKU | Cost Driver |
|---|---|---|
| Azure OpenAI | S0 | Per 1K tokens (input/output pricing varies by model) |
| Azure AI Search | Standard | Base + per replica + per partition |
OpenAI cost is consumption-driven. A gpt-4o-mini deployment at capacity 1 incurs costs only on token usage. Capacity units control throughput limits, not billing.
AI Search cost scales with SKU tier, replica count, and partition count: - Standard: ~$250/month base (1 replica, 1 partition) - Each additional replica: ~$250/month - Each additional partition: ~$250/month
Estimated Monthly (baseline): $250–$500 (Search) + variable (OpenAI)
Data Services (Variable)¶
| Resource | SKU | Cost Driver |
|---|---|---|
| Cosmos DB | Standard (RU) | Provisioned throughput + storage |
| PostgreSQL Flexible | Standard_D2s_v3 | Compute + storage + backup |
Cosmos DB: Minimum ~$25/month (400 RU/s serverless) to ~$200/month (provisioned). Scales with throughput and storage.
PostgreSQL Flexible Server: ~$125/month for Standard_D2s_v3 + $0.115/GB/month storage.
Estimated Monthly: $125–$200 (one data profile active)
Security and Identity (Fixed)¶
| Resource | SKU | Cost Driver |
|---|---|---|
| Key Vault | Standard | Per operation + per key/secret stored |
| Managed Identities | Free | No charge |
Estimated Monthly: $5–$15
Observability (Variable)¶
| Resource | SKU | Cost Driver |
|---|---|---|
| Log Analytics | PerGB2018 | Per GB ingested + retention beyond free tier |
First 5 GB/month free. Beyond that, ~$2.76/GB ingested. Retention beyond 31 days: ~$0.10/GB/month.
Estimated Monthly: $20–$100 (depends on log volume)
Governance (Fixed)¶
| Resource | Cost |
|---|---|
| Azure Policy | Free (built-in and custom definitions) |
| Policy Assignments | Free |
Optional Compute (Variable)¶
| Resource | SKU | Cost Driver |
|---|---|---|
| AKS (private) | Standard_D4s_v5 | Per node + managed disk |
AKS: ~$140/month per Standard_D4s_v5 node. Default 3 nodes = ~$420/month. AKS control plane is free for Standard tier.
Estimated Monthly (if enabled): $420–$600
Total Estimated Monthly Cost¶
| Tier | Range | Notes |
|---|---|---|
| Foundation | $400–$900/month | Core services, no AKS |
| Enterprise | $450–$1,000/month | + diagnostics, policy (minimal overhead) |
| Regulated | $900–$1,600/month | + AKS, CMK, extended retention |
These are baseline estimates. OpenAI token consumption is the primary variable cost and is entirely usage-dependent.
Cost Optimization Levers¶
- AI Search: Reduce replicas/partitions in dev (minimum 1/1)
- OpenAI: Use lower-cost models (gpt-4o-mini vs gpt-4o) and lower capacity units
- Cosmos DB: Use serverless mode for dev/staging; provisioned for prod
- PostgreSQL: Use B-series burstable SKUs for dev (e.g., Standard_B2s at ~$50/month)
- Log Analytics: Reduce retention in non-prod, use daily cap to limit ingestion
- AKS: Disable in dev/staging if not needed; use spot node pools for cost savings
Cost Allocation¶
All resources are tagged with:
- client: Customer identifier
- environment: dev/staging/prod
- managed_by: terraform
- workload: tenantzero-ai
- cost-center: Configurable per deployment
Use Azure Cost Management with tag-based filtering for chargeback and showback reporting.