Enterprise AIEnterprise
Microsoft Foundry
Microsoft's unified platform for enterprise AI agents
- Overview
- Microsoft Foundry (formerly Azure AI Foundry, formerly Azure AI Studio) is Microsoft's full-stack platform for building and deploying AI applications and agents on Azure. It provides a catalog of 11,000+ models from OpenAI, Anthropic, Meta, xAI, and others alongside a GA Foundry Agent Service with sandboxed hosted agents, MCP-enabled toolboxes, managed memory, and built-in observability. Enterprise security, data residency options (including EU and APAC Data Zones), and one-click publishing to Microsoft 365 Copilot and Teams make it the default AI platform for Microsoft-ecosystem organizations.
- Why it matters
- Microsoft Foundry's rebrand from Azure AI Foundry signals a deliberate repositioning: agents, not models, are now the primary unit of the platform. Regulated industries (finance, healthcare, public sector) are adopting it specifically for its EU and APAC Data Zone options and private networking, displacing custom-built agent infrastructure. The rapidly expanding multi-provider model catalog reduces OpenAI lock-in risk, but total cost of ownership frequently runs 15–40% above raw token estimates once Azure AI Search, storage, and networking are included. Teams evaluating the platform should budget carefully and watch for PTU commitment pricing, which can significantly alter unit economics at scale.
- In practice
- Enterprise teams most commonly deploy Foundry for RAG-based document Q&A over SharePoint or Fabric data sources, using the managed RAG pipeline builder to avoid custom orchestration code. More advanced teams use the Foundry Agent Service to build multi-step autopilot agents — for example, overnight ticket triage or daily reporting routines — published directly into Teams for end-user access. A real limitation: PromptFlow, the earlier orchestration layer, is being deprecated and migration to Microsoft Framework Workflows is required by April 2027, creating near-term upgrade work for teams that built on it.
- Best for
- Microsoft-ecosystem enterprises building production AI agents that require regulatory compliance, data residency controls, and deep integration with Microsoft 365 and Teams
- Pricing
- Pay-as-you-go token pricing for most models (e.g., GPT-4o at $2.50/M input tokens, $10.00/M output tokens); provisioned throughput units (PTUs) available for high-volume workloads at 30–50% savings. Regional surcharges apply: EU Data Zone is ~9% above global rates, APAC Data Zone ~20% above global, effective September 1, 2026. Separate charges for Azure AI Search, Blob Storage, Copilot Studio sessions, and networking egress.
Key features
- 11,000+ model catalog spanning OpenAI, Anthropic, Meta, xAI, Mistral, and Microsoft Phi
- Foundry Agent Service (GA): sandboxed hosted agents with managed memory, MCP toolboxes, and end-to-end tracing
- Enterprise data residency with EU and APAC Data Zones, private networking, and Entra RBAC
- One-click publishing of agents to Microsoft 365 Copilot and Teams
- Fully managed RAG pipeline builder with built-in Foundry IQ (Azure AI Search) integration
Quick answers
- What is Microsoft Foundry?
- Microsoft Foundry (formerly Azure AI Foundry, formerly Azure AI Studio) is Microsoft's full-stack platform for building and deploying AI applications and agents on Azure. It provides a catalog of 11,000+ models from OpenAI, Anthropic, Meta, xAI, and others alongside a GA Foundry Agent Service with sandboxed hosted agents, MCP-enabled toolboxes, managed memory, and built-in observability. Enterprise security, data residency options (including EU and APAC Data Zones), and one-click publishing to Microsoft 365 Copilot and Teams make it the default AI platform for Microsoft-ecosystem organizations.
- Why does Microsoft Foundry matter?
- Microsoft Foundry's rebrand from Azure AI Foundry signals a deliberate repositioning: agents, not models, are now the primary unit of the platform. Regulated industries (finance, healthcare, public sector) are adopting it specifically for its EU and APAC Data Zone options and private networking, displacing custom-built agent infrastructure. The rapidly expanding multi-provider model catalog reduces OpenAI lock-in risk, but total cost of ownership frequently runs 15–40% above raw token estimates once Azure AI Search, storage, and networking are included. Teams evaluating the platform should budget carefully and watch for PTU commitment pricing, which can significantly alter unit economics at scale.
- What is Microsoft Foundry best for?
- Microsoft-ecosystem enterprises building production AI agents that require regulatory compliance, data residency controls, and deep integration with Microsoft 365 and Teams
- How much does Microsoft Foundry cost?
- Pay-as-you-go token pricing for most models (e.g., GPT-4o at $2.50/M input tokens, $10.00/M output tokens); provisioned throughput units (PTUs) available for high-volume workloads at 30–50% savings. Regional surcharges apply: EU Data Zone is ~9% above global rates, APAC Data Zone ~20% above global, effective September 1, 2026. Separate charges for Azure AI Search, Blob Storage, Copilot Studio sessions, and networking egress.
Seen in recent stories
Where Microsoft Foundry showed up in the last 90 days of KeyNews editions.
Know the terms
Glossary entries that explain what Microsoft Foundry is built on.
Agentic orchestration
The architecture pattern of coordinating multiple AI agents to accomplish complex tasks, with a supervisor agent routing work, managing state, and combining results from specialized sub-agents.
RAG (Retrieval-Augmented Generation)
A technique that retrieves relevant documents from an external knowledge base and feeds them to a model alongside the user's query. RAG reduces hallucination and keeps responses grounded in current, factual data.
Hyperscaler
A cloud computing provider operating at massive scale, primarily Microsoft Azure, Amazon AWS, and Google Cloud. Hyperscalers provide the GPU infrastructure, managed AI services, and global data center networks that power most AI deployments.
MCP (Model Context Protocol)
An open standard (created by Anthropic) that lets AI models connect to external tools, data sources, and services through a unified interface. MCP is becoming the USB-C of AI integrations.
Models behind it
The models Microsoft Foundry runs on, offers, or competes with.
More enterprise ai tools
New tools drop every week.
ONE BRIEFING · EVERY FRIDAY · FREE
Free. Unsubscribe anytime.