Kong's Multi-Cloud Advantage Shrinks

This points to bundling risk, not just feature competition. Kong wins when a company needs one control layer across AWS, Azure, Google Cloud, and on prem systems, but that pitch gets weaker when most traffic, identity, billing, and security reviews already live inside one cloud. In that setup, the cloud vendor can make its own gateway the default place to route models, expose tools, cache prompts, and enforce token limits.

• AWS, Google, and Microsoft now ship many of the same AI gateway building blocks inside their own stacks. Bedrock AgentCore Gateway turns APIs and Lambda functions into MCP tools with centralized governance. Apigee adds managed MCP support for governed APIs. Azure API Management supports token quotas and semantic caching for LLM APIs.
  2 amazon 3 amazon 4 google 5 microsoft 6 microsoft
• The economic problem for Kong is that native tools ride existing cloud contracts. If an enterprise already buys most compute from AWS or Azure, the security team already trusts that control plane, the platform team already uses its IAM model, and procurement can often add gateway features with far less friction than approving a separate vendor.
  1 sacra 2 amazon 4 google 5 microsoft
• Neutrality still matters in the pockets of the market that are genuinely mixed, such as large regulated companies with legacy systems across clouds, or teams switching among model providers. But even there, the neutral layer must do more than unify endpoints. It has to offer better policy controls, analytics, and workflow fit than the built in option.
  1 sacra 7 flexera 8 sacra

The direction of travel is clear. As agent infrastructure gets absorbed into the main hyperscaler control planes, standalone gateways will need to anchor on deeper cross cloud operations and stronger developer workflows, not on neutrality alone. Kong’s best path is to become the layer companies keep even after standardizing most new infrastructure on one cloud.