diff --git a/docker/Dockerfile.cpu b/docker/Dockerfile.cpu
index c2ce0e88f296..e185c00cb2fa 100644
--- a/docker/Dockerfile.cpu
+++ b/docker/Dockerfile.cpu
@@ -165,6 +165,23 @@ RUN --mount=type=cache,target=/root/.cache/uv \
     --mount=type=cache,target=/vllm-workspace/.deps,sharing=locked \
     VLLM_TARGET_DEVICE=cpu python3 setup.py bdist_wheel --dist-dir=dist --py-limited-api=cp38
 
+######################### TRITON-CPU BUILD IMAGE #########################
+FROM base AS vllm-triton-cpu-build
+
+WORKDIR /vllm-workspace
+
+RUN mkdir dist
+
+RUN --mount=type=cache,target=/root/.cache/uv \
+    --mount=type=cache,target=/root/.cache/ccache \
+    --mount=type=cache,target=/vllm-workspace/.deps,sharing=locked \
+    if [ "$TARGETARCH" = "amd64" ] || [ "$VLLM_CPU_X86" != "0" ]; then \
+        git clone --recurse-submodules "https://github.com/triton-lang/triton-cpu.git"; \
+        cd triton-cpu; \
+        git checkout "270e696d"; \
+        uv build --wheel --out-dir=../dist; \
+    fi
+
 ######################### TEST DEPS #########################
 FROM base AS vllm-test-deps
 
@@ -257,7 +274,14 @@ WORKDIR /vllm-workspace
 RUN --mount=type=cache,target=/root/.cache/uv \
     --mount=type=cache,target=/root/.cache/ccache \
     --mount=type=bind,from=vllm-build,src=/vllm-workspace/dist,target=dist \
-    uv pip install "$(realpath dist/*.whl)[audio,triton-cpu]"
+    uv pip install "$(realpath dist/*.whl)[audio]"
+
+RUN --mount=type=cache,target=/root/.cache/uv \
+    --mount=type=cache,target=/root/.cache/ccache \
+    --mount=type=bind,from=vllm-triton-cpu-build,src=/vllm-workspace/dist,target=dist \
+    if [ "$TARGETARCH" = "amd64" ] || [ "$VLLM_CPU_X86" != "0" ]; then \
+        uv pip install "$(realpath dist/*.whl)"; \
+    fi
 
 # Add labels to document build configuration
 LABEL org.opencontainers.image.title="vLLM CPU"
diff --git a/setup.py b/setup.py
index a95ee3451b58..3221c1d1bc0e 100644
--- a/setup.py
+++ b/setup.py
@@ -1195,11 +1195,6 @@ def _read_requirements(filename: str) -> list[str]:
             "opentelemetry-exporter-otlp>=1.26.0",
             "opentelemetry-semantic-conventions-ai>=0.4.1",
         ],
-        "triton-cpu": [
-            "triton @ "
-            "git+https://github.com/triton-lang/triton-cpu.git@270e696d ; "
-            "platform_machine == 'x86_64'",
-        ],  # Remove after stable release
     },
     cmdclass=cmdclass,
     package_data=package_data,