Kong raised a $175M Series E in November 2024 at a $2B post-money valuation, led by Tiger Global and co-led by Balderton Capital, with participation from Ontario Teachers' Pension Plan, 137 Ventures, Andreessen Horowitz, Index Ventures, CRV, Sapphire Ventures, Notable Capital, and StepStone.

Before the Series E, Kong raised a $100M Series D in February 2021 at a $1.4B valuation, also led by Tiger Global. Earlier rounds included a $43M Series C in March 2019 led by Index Ventures, an $18M Series B in March 2017 led by Andreessen Horowitz with participation from CRV and Index, a $8.5M Series A in November 2013, and a $1M seed in 2011.

Kong has raised $345M in total primary equity across its full funding history.

Kong sits between an organization's applications and the systems those applications call, including backend APIs, microservices, Kafka event streams, LLM providers, MCP tool servers, and other AI agents. Requests pass through Kong, which applies authentication, rate limits, routing rules, cost controls, and audit logging before the call reaches its destination.

At the core is Kong Konnect, a cloud-hosted control plane where platform engineering teams configure policy, manage gateways, publish API catalogs, and monitor traffic across the organization. Konnect supports three deployment modes: fully self-managed gateways where the customer hosts everything, a hybrid model where Kong runs the control plane but the customer runs the data plane inside their own infrastructure, and Dedicated Cloud Gateways where Kong operates single-tenant managed runtimes on AWS, Azure, or GCP. This deployment flexibility is relevant in regulated industries where data cannot leave a specific environment.

In a typical workflow, a platform team creates a control plane in Konnect, deploys gateway instances near its services, defines which services are reachable and under what conditions, and publishes those services to an internal Service Catalog or an external Developer Portal. Developers consuming those APIs authenticate through Kong, are rate-limited through Kong, and have their traffic logged through Kong, without embedding that logic inside the application itself.

Kong's AI Gateway applies the same pattern to LLM traffic. Rather than having each application team integrate directly with OpenAI, Anthropic, Azure OpenAI, or a self-hosted model, LLM calls route through Kong, which applies token rate limiting, semantic caching, PII sanitization, prompt guardrails, and semantic routing, sending simpler prompts to cheaper models automatically. With 60% of enterprises now running seven or more LLMs internally, that routing layer can reduce cost materially: if 10,000 employees are sending mid-complexity queries, defaulting everything to a frontier model burns tokens unnecessarily, and a gateway that semantically routes straightforward tasks to a cheaper model can save tens of millions of dollars annually at scale.

The product also covers MCP tool governance through Kong's MCP Registry, agent-to-agent traffic through the Agent Gateway launched in April 2026, and Kafka event streams through the Event Gateway. Kong Context Mesh, currently in tech preview, uses the existing API catalog to transform APIs into governed, agent-consumable tools, allowing platform teams to expose enterprise systems to AI agents without hand-coding every integration.

Kong sells to enterprise platform engineering, API product, and AI infrastructure teams through a B2B model built around the Kong Konnect platform. The entry point is a self-serve Plus tier with per-gateway monthly pricing, while the primary revenue engine is a custom-priced Enterprise tier billed annually and sold through a direct sales motion with dedicated customer success and technical account management.

Pricing is based on platform consumption rather than seats. Enterprise contracts scale with the number of gateways, regions, traffic volume, and product modules in use. On the Plus tier, AI Gateway charges $100 per month per additional LLM model proxied beyond the included five, creating expansion revenue as enterprises add models to their portfolio. The OpenMeter acquisition added a usage-based billing layer that charges 0.4% of billing volume processed through Konnect Metering & Billing, adding a take-rate revenue stream tied to customer commercial activity rather than only infrastructure footprint.

The open-source flywheel is the main go-to-market engine. Kong Gateway has 43,500+ GitHub stars and is running in 37,000+ companies, generating inbound enterprise sales from organizations already running Kong in production. That installed base converts at roughly 2–3% to paid Konnect contracts, and those contracts are high-ACV: the average revenue per customer implied by the 2023 ARR milestone was around $167K, and expansion within accounts has historically driven net revenue retention above 130%.

The cost structure combines software and infrastructure. Hybrid deployments where customers run their own data planes carry software-like margins, while Dedicated Cloud Gateways add variable cloud infrastructure costs that Kong partially offsets through explicit per-control-plane and bandwidth pricing. AI governance features add no model inference cost to Kong's own COGS because Kong governs access to models rather than running them, making AI Gateway a structurally high-margin product layer on top of the existing gateway runtime.

Kong competes across three overlapping markets that are converging: classic API management, AI traffic governance, and an emerging agentic connectivity layer. The core competitive dynamic is that major hyperscalers and integration suites are adding AI gateway capabilities to existing platforms, shifting competition from gateway performance alone toward platform adjacency and procurement simplification.

Google Apigee, Microsoft Azure API Management, and AWS Bedrock AgentCore Gateway now offer AI routing, semantic caching, MCP tool exposure, and A2A governance as extensions of their existing cloud platforms. For enterprises already standardized on a single cloud, these tools can become the default governance layer at near-zero incremental cost, reducing the need for a neutral third-party platform.

Kong's defense is multi-cloud neutrality and hybrid deployment flexibility. Regulated enterprises that cannot route all traffic through a single public cloud, or that need consistent policy enforcement across AWS, Azure, GCP, and on-premise environments simultaneously, have an architectural reason to choose Kong over a cloud-native alternative. That case weakens if cloud consolidation accelerates.

Salesforce MuleSoft, IBM API Connect, Gravitee, and WSO2 compete from the integration suite angle, where API management is bundled into a broader platform decision rather than evaluated on its own merits. MuleSoft's Agent Fabric and Omni Gateway now cover LLM proxying, MCP governance, and agent traffic alongside its existing integration estate, making it a credible alternative when the buying criterion is reducing vendor count rather than gateway performance.

Gravitee is the closest narrative competitor because it presents a unified runtime for APIs, events, and AI/agent traffic from a single control plane, similar to Kong's framing. WSO2 claims to federate across heterogeneous gateway estates including Kong itself, which would shift value from the runtime layer to the policy orchestration layer above it if adopted.

Purpose-built AI gateway vendors like Portkey attack from the opposite direction, starting with LLM routing and prompt observability rather than extending API management. Portkey processes 500B+ tokens per day across 24,000+ organizations and has raised $18M, offering a simpler buyer story for AI-native teams whose immediate problem is LLM productionization rather than enterprise API governance.

LiteLLM, the open-source AI gateway with 40,000+ GitHub stars, sets a floor for functionality and price. A March 2026 supply chain compromise that exposed enterprise customers' API keys and code strengthened Kong's enterprise security case against unmanaged open-source alternatives and created a migration opportunity. OpenRouter, which Sacra has covered as the Costco of LLMs processing 3T+ tokens per day at a ~5% take-rate, serves a different market, developer-side model discovery and price arbitrage rather than behind-the-firewall enterprise governance, but still appears in competitive evaluations when enterprises are deciding how to govern LLM traffic.

Kong's expansion logic is that API governance and AI governance are converging into the same control plane problem, and that the same platform that governs REST traffic today can govern LLM calls, MCP tool access, agent-to-agent communication, and event streams tomorrow. Each new traffic type is both a product extension and a separate budget line.

The AI Gateway launched in February 2024 and has gone through 14+ major releases since, extending coverage from basic LLM proxying to semantic routing, MCP governance, and A2A agent traffic. These additions map to new buyers: AI platform teams now have a distinct budget for LLM governance that did not exist in 2022, and regulated industries face compliance requirements that make centralized AI traffic management mandatory rather than optional.

Kong Context Mesh, currently in tech preview, is the next expansion vector: automatically discovering existing enterprise APIs and transforming them into governed, agent-consumable tools. As enterprises move from chatbot deployments to multi-step agentic workflows, consuming 10x more tokens per task through tool-calling loops and million-token context windows, the governance layer between agents and enterprise systems becomes a prerequisite for controlling AI spend.

The OpenMeter acquisition added Konnect Metering & Billing, extending Kong from a governance tool into revenue infrastructure. Enterprises can meter API and AI token usage, define pricing plans, enforce entitlements at the gateway layer, and generate invoices or internal chargebacks from the same control plane that governs their traffic.

This is a different TAM than API management. The API economy that Kong's customers are building, monetized developer platforms, AI-powered products, and partner ecosystems, represents a large pool of commercial activity. Kong's 0.4% take-rate on billing volume processed through Konnect scales with that activity rather than with Kong's own customer count, creating a revenue stream tied to growth in the API economy. WSO2's parallel acquisition of Moesif indicates competitors are targeting the same category.

Kong's installed base of 37,000+ companies running Kong OSS represents a large unconverted pipeline. The current paid conversion rate of roughly 2–3% leaves room for expansion, particularly as AI Gateway creates a new entry point for organizations that were not previously in the market for enterprise API management but now need LLM governance.

Geographically, Kong has expanded into Japan via a joint venture, opened offices in Bengaluru, Sydney, Budapest, and London, and is available on AWS Marketplace for procurement via cloud commitments. In Europe, the EU AI Act and GDPR create explicit demand for Kong's on-premise and hybrid deployment models, where data residency requirements make fully managed cloud alternatives less viable for regulated industries.

Hyperscaler absorption: As AWS, Google, and Microsoft integrate AI gateway capabilities, semantic caching, MCP tool exposure, token rate limiting, and A2A governance, directly into their cloud platforms at near-zero incremental cost, Kong's multi-cloud neutrality argument weakens as more enterprise infrastructure consolidates onto a single hyperscaler, which is the dominant direction of enterprise cloud procurement.

Lua architecture ceiling: Kong's plugin system requires Lua, while purpose-built AI gateway competitors such as Portkey, LiteLLM, and a growing field of AI-native alternatives are built in Python, the language many AI engineers use, which risks excluding Kong from AI-native team purchases even as it invests heavily in AI connectivity positioning.

Platform breadth vs. depth: Kong now spans API gateway, AI gateway, event streaming governance, MCP registry, agent-to-agent traffic, machine identity, and usage-based billing, but several newer products, including Context Mesh and Agent Gateway, remain in early availability while competitors such as Gravitee and WSO2 market similar platform consolidation stories to their installed bases.