NVIDIA · mchmarny · May 14, 2026 · May 13, 2026 · May 14, 2026 · May 14, 2026
@@ -259,9 +259,9 @@ print_dynamo_diagnostics() {
   kubectl_kind -n dynamo-system get events --sort-by='.lastTimestamp' 2>/dev/null | tail -30 || true
 }
 
-print_kgateway_diagnostics() {
-  echo "=== kgateway pods ==="
-  kubectl_kind -n kgateway-system get pods -o wide 2>/dev/null || true
+print_agentgateway_diagnostics() {
+  echo "=== agentgateway pods ==="
+  kubectl_kind -n agentgateway-system get pods -o wide 2>/dev/null || true
   echo "=== GatewayClass status ==="
   kubectl_kind get gatewayclass -o yaml 2>/dev/null || true
   echo "=== Gateway status ==="
@@ -280,9 +280,9 @@ case "${mode}" in
     print_kubeflow_diagnostics
     ;;
   inference)
-    print_h100_common_diagnostics dynamo-system kgateway-system
+    print_h100_common_diagnostics dynamo-system agentgateway-system
     print_dynamo_diagnostics
-    print_kgateway_diagnostics
+    print_agentgateway_diagnostics
     ;;
   *)
     echo "::error::unknown GPU_TEST_DIAGNOSTIC_MODE: ${mode}"

@@ -89,8 +89,8 @@ jobs:
               - 'recipes/components/nodewright-operator/**'
               - 'recipes/components/nvidia-dra-driver-gpu/**'
               - 'recipes/components/nvsentinel/**'
-              - 'recipes/components/kgateway/**'
-              - 'recipes/components/kgateway-crds/**'
+              - 'recipes/components/agentgateway/**'
+              - 'recipes/components/agentgateway-crds/**'
               - 'recipes/components/grove/**'
               - 'recipes/components/dynamo-platform/**'
               - 'recipes/components/prometheus-adapter/**'

@@ -42,8 +42,8 @@
   │    ├── gpu-operator/             (GPU driver, device-plugin, DCGM)     │
   │    ├── nvidia-dra-driver-gpu/    (Dynamic Resource Allocation)         │
   │    ├── kai-scheduler/            (gang scheduling)                     │
-  │    ├── kgateway-crds/            (Gateway API + inference CRDs)        │
-  │    ├── kgateway/                 (inference gateway controller)        │
+  │    ├── agentgateway-crds/        (Gateway API + inference CRDs)        │
+  │    ├── agentgateway/             (inference gateway controller)        │
   │    ├── nvsentinel/               (security/compliance)                 │
   │    ├── nodewright-operator/         (node configuration)                  │
   │    ├── nodewright-customizations/   (H100 tuning)                         │
@@ -60,13 +60,13 @@
   │  $ cd bundle && ./deploy.sh                                            │
   │                                                                        │
   │  cert-manager ──▶ kube-prometheus-stack ──▶ gpu-operator ──▶           │
-  │  kai-scheduler ──▶ kgateway ──▶ nvidia-dra-driver ──▶                  │
+  │  kai-scheduler ──▶ agentgateway ──▶ nvidia-dra-driver ──▶              │
   │  dynamo-platform ──▶ nodewright ──▶ nvsentinel ──▶ ...                 │
   │                                                                        │
   │  Result: Fully configured GPU cluster                                  │
   │    • 8x H100 GPUs advertised via DRA                                   │
   │    • Gang scheduling (KAI Scheduler)                                   │
-  │    • Inference gateway (kgateway)                                      │
+  │    • Inference gateway (agentgateway)                                  │
   │    • GPU metrics (DCGM → Prometheus → HPA)                             │
   │    • Dynamo inference platform                                         │
   └────────────────────────────────────────────────────────────────────────┘
@@ -114,8 +114,8 @@
 │  └── aws-efa                        │  └── aws-efa                        │
 │      │                              │      │                              │
 │  eks-training.yaml                  │  eks-inference.yaml                 │
-│  (no new components)                │  ├── kgateway-crds          ◀── NEW │
-│      │                              │  └── kgateway               ◀── NEW │
+│  (no new components)                │  ├── agentgateway-crds      ◀── NEW │
+│      │                              │  └── agentgateway           ◀── NEW │
 │      │                              │      │                              │
 │  h100-eks-training.yaml             │  h100-eks-inference.yaml            │
 │  ├── gpu-operator (CDI, gdrcopy)    │  └── nodewright-customizations         │
@@ -130,7 +130,8 @@
 │                                     │  └── dynamo-platform         ◀─ NEW │
 │                                     │                                     │
 ├─────────────────────────────────────┼─────────────────────────────────────┤
-│  Unique: kubeflow-trainer           │  Unique: kgateway-crds, kgateway,   │
+│  Unique: kubeflow-trainer           │  Unique: agentgateway-crds,         │
+│                                     │          agentgateway,              │
 │                                     │    dynamo-crds, dynamo-platform     │
 ├─────────────────────────────────────┴─────────────────────────────────────┤
 │  Shared (base + eks): cert-manager, kube-prometheus-stack, gpu-operator,  │
@@ -253,7 +254,7 @@ http://127.0.0.1:9090/chat.html
 │   │                            │   toolkit, DCGM, validator)              │         │
 │ 4 │ accelerator_metrics        │ gpu-operator (DCGM exporter)             │ base    │
 │ 5 │ ai_service_metrics         │ kube-prometheus-stack, prometheus-adapter│ base    │
-│ 6 │ ai_inference               │ kgateway-crds, kgateway                  │ eks-inf │
+│ 6 │ ai_inference               │ agentgateway-crds, agentgateway          │ eks-inf │
 │ 7 │ robust_controller          │ dynamo-crds, dynamo-platform             │ dynamo  │
 │ 8 │ pod_autoscaling            │ prometheus-adapter + HPA                 │ base    │
 │ 9 │ cluster_autoscaling        │ EKS Auto Scaling Group (ASG)             │ infra   │
@@ -263,7 +264,7 @@ http://127.0.0.1:9090/chat.html
 │    DRA, gang scheduling, secure access, accelerator metrics,                        │
 │    AI service metrics, pod autoscaling                                              │
 │                                                                                     │
-│  eks-inference layer (+1):  inference gateway (kgateway)                            │
+│  eks-inference layer (+1):  inference gateway (agentgateway)                        │
 │  dynamo layer (+1):         robust controller (Dynamo operator)                     │
 │  infra layer (+1):          cluster autoscaling (EKS ASG)                           │
 │                                                                                     │

@@ -85,7 +85,7 @@ kubectl get dynamographdeployments -n dynamo-workload
 kubectl get pods -n dynamo-workload -o wide -w
 
 # Verify the inference gateway routes to the workload
-kubectl get gateway inference-gateway -n kgateway-system
+kubectl get gateway inference-gateway -n agentgateway-system
 kubectl get inferencepool -n dynamo-workload
 ```
 

@@ -83,7 +83,7 @@ kubectl get dynamographdeployments -n dynamo-workload
 kubectl get pods -n dynamo-workload -o wide -w
 
 # Verify the inference gateway routes to the workload
-kubectl get gateway inference-gateway -n kgateway-system
+kubectl get gateway inference-gateway -n agentgateway-system
 kubectl get inferencepool -n dynamo-workload
 ```
 

@@ -82,8 +82,8 @@ Visual: Single input forking into two divergent paths
 │  Unique:              │       │  Unique:              │
 │    kubeflow-trainer   │       │    dynamo-crds        │
 │                       │       │    dynamo-platform    │
-│  GPU Operator:        │       │    kgateway-crds      │
-│    CDI=true           │       │    kgateway           │
+│  GPU Operator:        │       │    agentgateway-crds  │
+│    CDI=true           │       │    agentgateway       │
 │    gdrcopy=true       │       │                       │
 │                       │       │  DRA driver:          │
 │                       │       │    gpuResources=true  │

@@ -129,7 +129,7 @@ aicr query --service eks --accelerator h100 --intent inference --os ubuntu \
 diff /tmp/training.txt /tmp/inference.txt
 ```
 
-> `> kgateway` and `> kgateway-crds` — the Inference Gateway is added only
+> `> agentgateway` and `> agentgateway-crds` — the Inference Gateway is added only
 > when `--intent inference`.
 
 CDI defaults also flip:

@@ -359,8 +359,8 @@ Bundler names correspond to component names in [`recipes/registry.yaml`](https:/
 | `kai-scheduler` | DRA-aware gang scheduler with topology-aware placement |
 | `grove` | Dynamo pod lifecycle management |
 | `dynamo-platform` | NVIDIA Dynamo inference serving platform |
-| `kgateway-crds` | Kubernetes Gateway API CRDs |
-| `kgateway` | Kubernetes Gateway API implementation |
+| `agentgateway-crds` | Kubernetes Gateway API CRDs for AI/ML inference (Gateway API + Inference Extension) |
+| `agentgateway` | Kubernetes Gateway API implementation for AI/ML inference (InferencePool routing) |
 | `k8s-nim-operator` | NVIDIA NIM Operator for inference microservice deployments |
 | `kueue` | Kubernetes-native job queuing for batch and AI workloads |
 | `kubeflow-trainer` | Kubeflow Training Operator for distributed training |

@@ -1111,7 +1111,7 @@ aicr bundle --recipe recipe.yaml \
 This results in:
 - **GPU daemonsets** (driver, device-plugin, toolkit, dcgm): `nodeSelector=nodeGroup=gpu-worker` + tolerations for `dedicated=worker-workload` with both `NoSchedule` and `NoExecute`
 - **NFD workers**: no nodeSelector (runs on all nodes) + tolerations for `dedicated=worker-workload` with both `NoSchedule` and `NoExecute`
-- **System components** (gpu-operator controller, NFD gc/master, dynamo grove, kgateway proxy): `nodeSelector=nodeGroup=system-worker` + tolerations for `dedicated=system-workload` with both `NoSchedule` and `NoExecute`
+- **System components** (gpu-operator controller, NFD gc/master, dynamo grove, agentgateway proxy): `nodeSelector=nodeGroup=system-worker` + tolerations for `dedicated=system-workload` with both `NoSchedule` and `NoExecute`
 
 **Behavior:**
 - All components from the recipe are bundled automatically

@@ -27,8 +27,8 @@ The source of truth is [`recipes/registry.yaml`](https://github.com/NVIDIA/aicr/
 | **kai-scheduler** | DRA-aware gang scheduler with hierarchical queues and topology-aware placement. Ensures distributed training jobs land on nodes with optimal interconnect topology. | [KAI Scheduler](https://github.com/kai-scheduler/KAI-Scheduler) |
 | **grove** | Pod lifecycle management for Dynamo inference platform. Installed as a standalone component. | [Grove](https://github.com/ai-dynamo/grove) |
 | **dynamo-platform** | NVIDIA Dynamo inference serving platform with bundled CRDs. Distributed inference with prefix-cache-aware routing and disaggregated prefill/decode. | [Dynamo](https://github.com/ai-dynamo/dynamo) |
-| **kgateway-crds** | Custom Resource Definitions for kgateway (Kubernetes Gateway API implementation). | [kgateway](https://github.com/kgateway-dev/kgateway) |
-| **kgateway** | Kubernetes Gateway API implementation. Provides model-aware ingress routing for inference workloads. | [kgateway](https://github.com/kgateway-dev/kgateway) |
+| **agentgateway-crds** | Custom Resource Definitions for agentgateway (Kubernetes Gateway API implementation for AI/ML inference). | [agentgateway](https://github.com/agentgateway/agentgateway) |
+| **agentgateway** | Kubernetes Gateway API implementation for AI/ML inference. Implements the Gateway API Inference Extension for model-aware ingress routing to InferencePool backends. | [agentgateway](https://github.com/agentgateway/agentgateway) |
 | **k8s-nim-operator** | NVIDIA NIM Operator for managing NIM (NVIDIA Inference Microservices) deployments on Kubernetes. | [K8s NIM Operator](https://github.com/NVIDIA/k8s-nim-operator) |
 | **kueue** | Kubernetes-native job queuing system. Manages quotas and admits jobs for batch and AI workloads. | [Kueue](https://github.com/kubernetes-sigs/kueue) |
 | **kubeflow-trainer** | Kubeflow Training Operator for distributed training jobs (PyTorch, etc.). Manages multi-node training job lifecycle with JobSet integration. | [Kubeflow Trainer](https://github.com/kubeflow/trainer) |
@@ -41,7 +41,7 @@ Not every component appears in every recipe. The recipe engine selects component
 
 - **Base components** (cert-manager, kube-prometheus-stack) appear in most recipes.
 - **Cloud-specific components** (aws-efa, aws-ebs-csi-driver) are added when the service matches.
-- **Intent-specific components** (kgateway, kgateway-crds) are added based on workload intent (e.g., inference recipes include the inference gateway).
+- **Intent-specific components** (agentgateway, agentgateway-crds) are added based on workload intent (e.g., inference recipes include the inference gateway).
 - **Platform-specific components** (slinky-slurm-operator, kubeflow-trainer, dynamo-platform) are added when the recipe selects a matching `--platform`.
 - **Accelerator/OS-specific tuning** (nodewright-customizations, nvidia-dra-driver-gpu) varies by hardware and OS combination.
 

@@ -23,12 +23,14 @@ A machine-readable **CycloneDX 1.6 JSON** companion to this page is produced by
 - Unique images: **71**
 - Distinct registries: **11**
 
-Registries: `602401143452.dkr.ecr.us-west-2.amazonaws.com`, `cr.kgateway.dev`, `docker.io`, `gcr.io`, `ghcr.io`, `gke.gcr.io`, `nvcr.io`, `public.ecr.aws`, `quay.io`, `registry.k8s.io`, `us-docker.pkg.dev`
+Registries: `602401143452.dkr.ecr.us-west-2.amazonaws.com`, `cr.agentgateway.dev`, `docker.io`, `gcr.io`, `ghcr.io`, `gke.gcr.io`, `nvcr.io`, `public.ecr.aws`, `quay.io`, `registry.k8s.io`, `us-docker.pkg.dev`
 
 ## Components
 
 | Component | Type | Chart | Pinned Version | Images |
 |-----------|------|-------|----------------|--------|
+| agentgateway | helm | agentgateway | v2.2.1 | 1 |
+| agentgateway-crds | helm | agentgateway-crds | v2.2.1 | 0 |
 | aws-ebs-csi-driver | helm | aws-ebs-csi-driver/aws-ebs-csi-driver | 2.59.0 | 6 |
 | aws-efa | helm | aws-efa-k8s-device-plugin | v0.5.26 | 1 |
 | cert-manager | helm | jetstack/cert-manager | v1.20.2 | 4 |
@@ -39,8 +41,6 @@ Registries: `602401143452.dkr.ecr.us-west-2.amazonaws.com`, `cr.kgateway.dev`, `
 | k8s-ephemeral-storage-metrics | helm | k8s-ephemeral-storage-metrics/k8s-ephemeral-storage-metrics | 1.19.2 | 1 |
 | k8s-nim-operator | helm | k8s-nim-operator | 3.1.0 | 1 |
 | kai-scheduler | helm | kai-scheduler | v0.14.1 | 2 |
-| kgateway | helm | kgateway | v2.0.0 | 1 |
-| kgateway-crds | helm | kgateway-crds | v2.0.0 | 0 |
 | kube-prometheus-stack | helm | prometheus-community/kube-prometheus-stack | 84.4.0 | 8 |
 | kubeflow-trainer | helm | kubeflow-trainer | 2.2.0 | 3 |
 | kueue | helm | kueue | 0.17.1 | 1 |
@@ -56,6 +56,14 @@ Registries: `602401143452.dkr.ecr.us-west-2.amazonaws.com`, `cr.kgateway.dev`, `
 
 ## Images by component
 
+### agentgateway
+
+- `cr.agentgateway.dev/controller:v2.2.1`
+
+### agentgateway-crds
+
+_No images extracted._
+
 ### aws-ebs-csi-driver
 
 - `public.ecr.aws/csi-components/csi-attacher:v4.11.0-eksbuild.4`
@@ -121,14 +129,6 @@ Registries: `602401143452.dkr.ecr.us-west-2.amazonaws.com`, `cr.kgateway.dev`, `
 - `ghcr.io/kai-scheduler/kai-scheduler/crd-upgrader:v0.14.1`
 - `ghcr.io/kai-scheduler/kai-scheduler/operator:v0.14.1`
 
-### kgateway
-
-- `cr.kgateway.dev/kgateway-dev/kgateway:v2.0.0`
-
-### kgateway-crds
-
-_No images extracted._
-
 ### kube-prometheus-stack
 
 - `docker.io/grafana/grafana:13.0.1`
@@ -222,7 +222,7 @@ AICR pulls from a deliberately diverse set of registries:
 - **`public.ecr.aws`** — AWS public artifacts (aws-ebs-csi-driver).
 - **Regional ECR** (`<account>.dkr.ecr.<region>.amazonaws.com`) — EKS-internal add-ons. The `aws-efa` entry below shows `us-west-2` because that is the in-tree default; deployments in other regions override `awsefa:image.repository` at bundle or install time. See [Regional registry overrides](../integrator/recipe-development.md#regional-registry-overrides) for the pattern.
 - **`gcr.io`, `gke.gcr.io`, `us-docker.pkg.dev`** — GCP/GKE add-ons (gke-nccl-tcpxo).
-- **`cr.kgateway.dev`** — kgateway.
+- **`cr.agentgateway.dev`** — agentgateway (AI inference gateway).
 - **`docker.io`** — assorted upstream images (`busybox`, `pytorch`, etc.).
 
 Customers running in air-gapped or private-registry environments need to mirror every registry above. A dedicated mirroring guide is tracked under [#743](https://github.com/NVIDIA/aicr/issues/743).

@@ -1428,14 +1428,14 @@ func TestUndeployScript_PreflightSkipListCoversManifestDeletedReleases(t *testin
 			}{Version: "v0.1.0"},
 			ComponentRefs: []recipe.ComponentRef{
 				{Name: "cert-manager", Namespace: "cert-manager", Chart: "cert-manager", Version: "v1.17.2", Source: "https://charts.jetstack.io"},
-				{Name: "kgateway", Namespace: "kgateway-system", Chart: "kgateway", Version: "v0.1.0", Source: "https://example.invalid/charts"},
+				{Name: "agentgateway", Namespace: "agentgateway-system", Chart: "agentgateway", Version: "v0.1.0", Source: "https://example.invalid/charts"},
 				{Name: "nodewright-operator", Namespace: "skyhook", Chart: "nodewright-operator", Version: "v0.1.0", Source: "https://example.invalid/charts"},
 			},
-			DeploymentOrder: []string{"cert-manager", "kgateway", "nodewright-operator"},
+			DeploymentOrder: []string{"cert-manager", "agentgateway", "nodewright-operator"},
 		},
 		ComponentValues: map[string]map[string]any{
 			"cert-manager":        {},
-			"kgateway":            {},
+			"agentgateway":        {},
 			"nodewright-operator": {},
 		},
 		Version: "v1.0.0",
@@ -1449,7 +1449,7 @@ func TestUndeployScript_PreflightSkipListCoversManifestDeletedReleases(t *testin
         snippet=$(sed -n '/^skip_preflight_for_release()/,/^}/p' "$UNDEPLOY")
         eval "$snippet"
         skip_preflight_for_release "nodewright-operator" && echo "skip:nodewright-operator"
-        skip_preflight_for_release "kgateway" && echo "skip:kgateway"
+        skip_preflight_for_release "agentgateway" && echo "skip:agentgateway"
         if skip_preflight_for_release "cert-manager"; then
             echo "unexpected:cert-manager"
             exit 1
@@ -1470,7 +1470,7 @@ func TestUndeployScript_PreflightSkipListCoversManifestDeletedReleases(t *testin
 	}
 
 	out := stdout.String()
-	for _, want := range []string{"skip:nodewright-operator", "skip:kgateway", "check:cert-manager"} {
+	for _, want := range []string{"skip:nodewright-operator", "skip:agentgateway", "check:cert-manager"} {
 		if !strings.Contains(out, want) {
 			t.Errorf("expected %q in output; stdout=%q stderr=%q", want, out, stderr.String())
 		}

@@ -248,7 +248,7 @@ extra_crds_for_release() {
 # deleted from manifests before the controller is uninstalled.
 skip_preflight_for_release() {
   case "$1" in
-    nodewright-operator|kgateway) return 0 ;;
+    nodewright-operator|agentgateway) return 0 ;;
     *) return 1 ;;
   esac
 }

@@ -248,7 +248,7 @@ extra_crds_for_release() {
 # deleted from manifests before the controller is uninstalled.
 skip_preflight_for_release() {
   case "$1" in
-    nodewright-operator|kgateway) return 0 ;;
+    nodewright-operator|agentgateway) return 0 ;;
     *) return 1 ;;
   esac
 }