Centralized Hub AI

A single-region shared AI platform hosted in the Hub landing zone, consumed by multiple application spokes. All AI services centrally managed, governed, and shared across business units.

$10,000 – $15,000/mo*
13
Azure Services
$10K–$15K
Monthly Est.*
99.9%
Composite SLA
Private
Network Access

Overview

The Centralized Hub AI pattern deploys all AI services — Azure OpenAI, AI Search, Machine Learning, and supporting data/security services — in a single Hub VNet. Application spokes connect via VNet Peering to consume these shared services through Private Endpoints.

This pattern is ideal for enterprises that want centralized governance, shared model management, and cost-efficient reuse of expensive AI services like Azure OpenAI across many applications.

Key Characteristics
  • Hub-and-Spoke Topology — Shared AI platform in the hub, application workloads in spokes
  • Private Endpoints — All AI services accessible only via private network
  • Centralized Governance — Single point of management for models, data, secrets
  • VNet Peering — Low-latency, private connectivity between hub and spokes
  • Shared Quota — GPT-4o (30K TPM) + Embeddings (120K TPM) shared across spokes
OpenAI Models
GPT-4o30K TPM
text-embedding-3-large120K TPM
Region

Canada Central

Storage Containers
raw processed models embeddings

Architecture Diagram

graph TB
    subgraph Spokes["Application Spokes"]
        S1["Spoke 1<br/>AKS / App Service"]
        S2["Spoke 2<br/>AKS / App Service"]
        SN["Spoke N<br/>AKS / App Service"]
    end
    subgraph Hub["Hub VNet — Platform Landing Zone"]
        AOAI["Azure OpenAI<br/>GPT-4o + Embeddings"]
        AIS["Azure AI Search<br/>Vector + Semantic"]
        AML["Azure Machine Learning<br/>Model Registry"]
        ADLS["ADLS Gen2<br/>Data Lake"]
        KV["Azure Key Vault<br/>Secrets"]
        MON["Log Analytics +<br/>App Insights"]
    end
    subgraph Security["Network Security"]
        PE["Private Endpoints"]
        DNS["Private DNS Zones"]
        NSG["NSG — DenyAllInbound"]
    end
    S1 -->|"VNet Peering"| AOAI
    S2 -->|"VNet Peering"| AIS
    SN -->|"VNet Peering"| AOAI
    AOAI --> AIS
    AML --> ADLS
    AML --> KV
    AOAI --> MON
    AIS --> MON
    PE --- AOAI
    PE --- AIS
    PE --- KV
    DNS --- PE
    NSG --- Hub
          

Bill of Materials

#ServiceResource NameSKU / TierPurposeMonthly Cost
1Azure OpenAI{base}-{env}-openaiS0GPT-4o + Embeddings for all spokes~$2,750
2Azure AI Search{base}-{env}-searchStandard S1Vector + semantic search indexes$245.28
3Azure Machine Learning{base}-{env}-amlWorkspaceModel registry, experiments, pipelines$0*
4ADLS Gen2{base}{env}adlsStandard LRS, HNSData lake for models, embeddings, raw data~$50
5Azure Key Vault{base}-{env}-kvStandardSecrets, keys, certificates (RBAC auth)~$5
6Log Analytics{base}-{env}-lawPerGB2018Centralized logging (90-day retention)~$23
7Application Insights{base}-{env}-aiWorkspace-basedAPM telemetry for AI servicesIncl.
8Hub VNet{base}-{env}-hub-vnet/16, 4 subnetsNetwork isolation backboneFree
9Spoke VNets (Ã-N){base}-{env}-spoke-{n}-vnet/16, 2 subnetsApplication landing zonesFree
10VNet PeeringHub↔SpokeBidirectionalPrivate connectivity hub ↔ spokes~$10/TB
11Private DNS Zones6 zonesGlobalName resolution for private endpoints~$3
12Private EndpointsPer AI servicePrivate network access to PaaS~$22
13NSG{base}-{env}-hub-nsgDenyAllInboundMicro-segmentation on hub subnetsFree
Estimated Range (Moderate Production)$10,000 – $15,000/mo

* AML workspace is free; compute instances billed separately. Token costs based on ~500M tokens/month.

Service Breakdown

Azure OpenAI
S0 ~$2,750/mo

Provides GPT-4o (30K TPM) for chat completions and text-embedding-3-large (120K TPM) for vector embeddings. Shared across all spokes as the central LLM endpoint.

  • GlobalStandard deployment
  • Private endpoint access only
  • Content filtering enabled
Azure AI Search
Standard S1 $245.28/mo

Vector and semantic search index for RAG patterns. Stores and retrieves embeddings for knowledge bases used by all spoke applications.

  • Semantic ranking enabled
  • Vector search capability
  • 25 GB storage included
Azure Machine Learning
Workspace Free*

Central model registry for versioning, experiment tracking, and ML pipeline orchestration. Linked to ADLS, Key Vault, and App Insights.

  • Model registry & versioning
  • Pipeline orchestration
  • Compute billed separately
ADLS Gen2
Standard LRS / HNS ~$50/mo

Hierarchical data lake for raw data, processed datasets, model artifacts, and generated embeddings. Containers: raw, processed, models, embeddings.

  • Hierarchical namespace enabled
  • TLS 1.2 minimum
  • Private endpoint access
Azure Key Vault
Standard ~$5/mo

Centralized secrets management for API keys, connection strings, and certificates. Uses RBAC authorization with purge protection enabled.

  • RBAC authorization (no access policies)
  • Purge protection enabled
  • Private endpoint access
Log Analytics + App Insights
PerGB2018 ~$23/mo

Centralized monitoring and APM. Collects diagnostics from all hub services with 90-day retention. App Insights provides request tracing and performance metrics.

  • 90-day data retention
  • Workspace-based App Insights
  • ~$2.30/GB ingestion
Hub VNet + Peering
/16 CIDR ~$10/TB

Hub VNet with 4 subnets provides the network backbone. VNet Peering connects spokes to hub bidirectionally for private, low-latency connectivity.

  • 4 subnets (AI, PE, Data, Mgmt)
  • Bidirectional peering
  • Same-region, minimal latency
Private Endpoints + DNS
6 zones ~$25/mo

Private Endpoints for each AI service ensure zero public internet exposure. Six Private DNS Zones (OpenAI, Search, KV, Blob, DFS, AML) handle private name resolution.

  • OpenAI, Search, KV, Blob, DFS, AML
  • $7.30/endpoint/month
  • $0.50/DNS zone/month
Network Security Groups
DenyAllInbound Free

NSGs on all hub subnets enforce micro-segmentation with DenyAllInbound as the default rule. Only VNet peering traffic and private endpoint traffic is permitted.

  • DenyAllInbound default
  • VNet-only allowed traffic
  • Per-subnet segmentation

Security & Networking

Network Security
  • All AI services accessed via Private Endpoints within Hub VNet
  • Spokes connect through bidirectional VNet Peering
  • Private DNS Zones linked to Hub VNet for name resolution
  • NSG with DenyAllInbound default rule on hub subnets
  • TLS 1.2 minimum on all storage and endpoints
Identity & Access
  • Key Vault uses RBAC authorization (no access policies)
  • Managed Identities for service-to-service authentication
  • Azure RBAC per-spoke for workload isolation
  • Service-level RBAC on Azure OpenAI deployments
  • Purge protection on Key Vault secrets

Use Cases

Enterprise Copilots

Shared GPT-4o across multiple applications — internal knowledge assistants, document Q&A, code review bots.

Centralized AI Governance

Single point of management for AI models, usage quotas, cost tracking, and compliance across business units.

Cost-Efficient Sharing

Amortize Azure OpenAI and AI Search costs across many consumers instead of per-workload duplication.

Standardized Model Management

AML Model Registry for versioning, A/B testing, and staged deployment across all spoke applications.

Constraints & Considerations

ConstraintMitigation
Shared quota limits (OpenAI TPM/RPM)Implement per-spoke rate limiting; request quota increases
Requires strong RBAC + isolation strategyUse Azure RBAC per-spoke, service-level RBAC on OpenAI
Single point of failure for AI servicesMonitor with App Insights; plan capacity headroom
Cross-spoke data leakage riskSeparate storage containers per spoke; enforce RBAC
Ready to Deploy This Pattern?

Generate a complete deployment spec sheet with GitHub Actions workflow, Bicep file structure, and prerequisite checklist.

Generate Deployment Spec Sheet