MicrosoftGreat announcement and very interesting direction for governing models and MCP servers.
One question regarding adoption: for organizations that already have Azure API Management instances in production, is AI Gateway intended to be deployed as a separate resource, or is there a planned path to leverage existing APIM instances for AI Gateway workloads?
Understanding the recommended coexistence and migration model would help platform teams evaluate adoption strategies and operational ownership.
MicrosoftExisting tiers of Azure API Management already offer AI gateway capabilities. AI Gateway tier is another option, exclusively focused and optimized for AI workloads.
I believe the AI Gateway tier is for the scenarios where you don't need other advanced API management capabilities. Other tiers of APIM currently support the AI gateway features as well including adding llm APIs, MCP servers, token metrics etc. So it's not going to be that you have two APIM instances running one with AI gateway tier and one with another. You select based on your use case. I don't see a reason why a medium to big size enterprise won't just use the current v2 SKU with ai gateway features. Or probably you need to separate your AI gateway with you enterprise API gateway.
https://learn.microsoft.com/en-us/azure/api-management/genai-gateway-capabilities
https://iaminazure.substack.com/p/azure-api-management-ai-gateway-capabilities
Tried capturing the difference
I doubt AI Gateway with it's limited capabilities in a position to warrant a migration, both going to be needed.
APIM has robust capabilities and i wish, many of them find their way to AI gateway so all combinations possible; APIM+/- AI Gateway .
MicrosoftThat's the plan. Thanks!
are there any timelines available when cost/policies (other than available now within the gatewya portal) etc are added to the AI Gateway so something like budgeting/limiting per team can be incorporated. The preview is limited in the sense that everybody having a key to the gateway can use all models available for that gateway.
So a bit what
From Multi-Model Chaos to a Governed AI Gateway: Cost Optimization on Azure | Microsoft Community Hub is trying to achieve.
