Gateway Competition Moves To Platform Adjacency
Kong
The real battle is moving from who has the fastest gateway to who already owns the surrounding workflow and budget. Once AWS, Google, and Cloudflare bundle AI routing, tool exposure, caching, and governance into platforms teams already use, the gateway stops being a standalone speed test and becomes part of a broader buying decision about vendor count, security review, and whether one contract can cover models, APIs, and agents.
-
Google is turning Apigee into a full AI control layer, not just an API proxy. It now adds MCP support, semantic caching, API hub cataloging, tool monetization, and multi agent orchestration, which means a team already on Google Cloud can extend its current control plane instead of buying a separate gateway product.
-
AWS is making the same bundling move with Bedrock AgentCore Gateway. It can turn APIs and Lambda functions into MCP compatible tools, front other agents and HTTP services, route across model providers, and plug into SaaS tools like Salesforce and Jira. That makes procurement easier because the gateway arrives inside the existing AWS estate.
-
Kong is responding by widening the surface area it governs. Its recent launches add MCP connectivity and agent to agent traffic on top of API and LLM traffic, pushing Kong from a narrow gateway toward a neutral control point for mixed environments. That matters most for enterprises that do not want AI governance tied to a single cloud.
The next step is that gateway products will increasingly be judged by how much surrounding infrastructure they collapse into one control plane. Vendors that can govern APIs, models, tools, and agents across clouds will keep an opening, while single cloud suites will keep winning accounts where buying convenience matters more than best of breed gateway performance.