diff --git a/.github/actions/locate-vcvarsall-and-setup-env/action.yml b/.github/actions/locate-vcvarsall-and-setup-env/action.yml
index fba855f14b487..6826f514ec558 100644
--- a/.github/actions/locate-vcvarsall-and-setup-env/action.yml
+++ b/.github/actions/locate-vcvarsall-and-setup-env/action.yml
@@ -5,6 +5,9 @@ inputs:
     description: 'Target architecture (x64 or x86)'
     required: true
     default: 'x64'
+  cache_key:
+    description: 'a workflow & job qualified unique key to distinguish a cache. used as a suffix for subcaches.'
+    required: true
 outputs:
   vcvarsall_path:
     description: "Path to vcvarsall.bat"
@@ -13,68 +16,106 @@ runs:
   using: "composite"
   steps:
 
-    - name: Setup VCPKG
-      uses: microsoft/onnxruntime-github-actions/setup-build-tools@v0.0.9
-      with:
-        vcpkg-version: '2025.08.27'
-        vcpkg-hash: '9a4b32849792e13bee1d24726f073b3881acae4165206ddf1a6378e44a4ddd05b3ee93f55ff46d8e8873b3cbcd06606212989e248f0bd615a5bf365070074079'
-        cmake-version: '3.31.6'
-        cmake-hash: '0f1584e8666cf4a65ec514bd02afe281caabf1d45d2c963f3151c41484f457386aa03273ab25776a670be02725354ce0b46f3a5121857416da37366342a833a0'
-        add-cmake-to-path: 'true'
-        disable-terrapin: 'false'
-
-    - name: Verify vcpkg setup
-      shell: pwsh # Use powershell to easily access env var
-      run: |
-          Write-Host "VCPKG_INSTALLATION_ROOT is set to: $env:VCPKG_INSTALLATION_ROOT"
-          & "$env:VCPKG_INSTALLATION_ROOT/vcpkg" version
-
-    - name: Find vcvarsall.bat
-      id: find-vcvarsall
-      shell: python  # Use Python shell
-      run: |
-        import os
-        import subprocess
-
-        vswhere_path = os.path.join(os.environ["ProgramFiles(x86)"], "Microsoft Visual Studio", "Installer", "vswhere.exe")
-
-        try:
-            process = subprocess.run([vswhere_path, "-latest", "-property", "installationPath"], capture_output=True, text=True, check=True)
-            vs_install_path = process.stdout.strip()
-            vcvarsall_path = os.path.join(vs_install_path, "VC", "Auxiliary", "Build", "vcvarsall.bat")
-
-            if os.path.exists(vcvarsall_path):
-                print(f"vcvarsall found at: {vcvarsall_path}")
-                # Use GITHUB_OUTPUT environment variable
-                with open(os.environ['GITHUB_OUTPUT'], 'a') as f:
-                    f.write(f"vcvarsall_path={vcvarsall_path}\n")
-            else:
-                print(f"vcvarsall.bat not found at expected path: {vcvarsall_path}")
-                # Use 'exit(1)' for Python to properly signal failure to GitHub Actions
-                exit(1)
-
-
-        except subprocess.CalledProcessError as e:
-            print(f"Error running vswhere.exe: {e}")
-            print(f"vswhere output: {e.stdout}")
-            print(f"vswhere stderr: {e.stderr}")
-            exit(1) # Exit with a non-zero code on error
-        except FileNotFoundError:
-            print(f"vswhere.exe not found at: {vswhere_path}")
-            exit(1)
-
-
-    - name: Setup Environment
-      shell: cmd
-      run: |
-        REM Get initial environment variables
-        set > initial_env.txt
-
-        REM Call vcvarsall.bat using the output from the previous step
-        call "${{ steps.find-vcvarsall.outputs.vcvarsall_path }}" ${{ inputs.architecture }}
-
-        REM Get environment variables after calling vcvarsall.bat
-        set > final_env.txt
-
-        REM Call the Python script to update the GitHub Actions environment
-        python ${{ github.action_path }}\update_environment.py
+  - name: Setup VCPKG
+    uses: microsoft/onnxruntime-github-actions/build/setup-build-tools@dev/sanaahamel/download-ninja
+    with:
+      ccache-version: 4.13.1
+      ccache-hash: fe678d8e82f35e1d5f13c7bcd86447c08b67b0831dec8c772ef1257090a0f51678e296bb6d3ea38433f472d0fded52da96445cbc6fa8f945ecf94351d2c149d8
+      cmake-version: '3.31.6'
+      cmake-hash: '0f1584e8666cf4a65ec514bd02afe281caabf1d45d2c963f3151c41484f457386aa03273ab25776a670be02725354ce0b46f3a5121857416da37366342a833a0'
+      ninja-version: 1.13.0
+      ninja-hash: cffea62711b5a89ec494b48ae5e8df32f24cab0e642363ac20cc250c7d7f75ad43258f8d3721c68b23413e401d42630945c534ff95355cfe664e5da1d36569bf
+      vcpkg-version: '2025.08.27'
+      vcpkg-hash: '9a4b32849792e13bee1d24726f073b3881acae4165206ddf1a6378e44a4ddd05b3ee93f55ff46d8e8873b3cbcd06606212989e248f0bd615a5bf365070074079'
+      add-cmake-to-path: 'true'
+      disable-terrapin: 'true' # ccache isn't on terrapin
+
+  - name: Verify vcpkg setup
+    shell: pwsh # Use powershell to easily access env var
+    run: |
+      Write-Host "VCPKG_INSTALLATION_ROOT is set to: $env:VCPKG_INSTALLATION_ROOT"
+      & "$env:VCPKG_INSTALLATION_ROOT/vcpkg" version
+
+  - name: Find vcvarsall.bat
+    id: find-vcvarsall
+    shell: python # Use Python shell
+    run: |
+      import os
+      import subprocess
+
+      vswhere_path = os.path.join(os.environ["ProgramFiles(x86)"], "Microsoft Visual Studio", "Installer", "vswhere.exe")
+
+      try:
+          process = subprocess.run([vswhere_path, "-latest", "-property", "installationPath"], capture_output=True, text=True, check=True)
+          vs_install_path = process.stdout.strip()
+          vcvarsall_path = os.path.join(vs_install_path, "VC", "Auxiliary", "Build", "vcvarsall.bat")
+
+          if os.path.exists(vcvarsall_path):
+              print(f"vcvarsall found at: {vcvarsall_path}")
+              # Use GITHUB_OUTPUT environment variable
+              with open(os.environ['GITHUB_OUTPUT'], 'a') as f:
+                  f.write(f"vcvarsall_path={vcvarsall_path}\n")
+          else:
+              print(f"vcvarsall.bat not found at expected path: {vcvarsall_path}")
+              # Use 'exit(1)' for Python to properly signal failure to GitHub Actions
+              exit(1)
+
+
+      except subprocess.CalledProcessError as e:
+          print(f"Error running vswhere.exe: {e}")
+          print(f"vswhere output: {e.stdout}")
+          print(f"vswhere stderr: {e.stderr}")
+          exit(1) # Exit with a non-zero code on error
+      except FileNotFoundError:
+          print(f"vswhere.exe not found at: {vswhere_path}")
+          exit(1)
+
+  - name: Setup Environment
+    shell: cmd
+    run: |
+      REM Get initial environment variables
+      set > initial_env.txt
+
+      REM Call vcvarsall.bat using the output from the previous step
+      call "${{ steps.find-vcvarsall.outputs.vcvarsall_path }}" ${{ inputs.architecture }}
+
+      REM Get environment variables after calling vcvarsall.bat
+      set > final_env.txt
+
+      REM Call the Python script to update the GitHub Actions environment
+      python ${{ github.action_path }}\update_environment.py
+
+  - name: Set cache dir env vars
+    env:
+      github_runner_temp: ${{ runner.temp }}
+    shell: pwsh
+    run: |
+      echo "CCACHE_DIR=${env:github_runner_temp}/.ccache" >> "${env:GITHUB_ENV}"
+      echo "VCPKG_DEFAULT_BINARY_CACHE=${env:github_runner_temp}/.vcpkg-cache" >> "${env:GITHUB_ENV}"
+
+      # vcpkg will error/exit if `VCPKG_DEFAULT_BINARY_CACHE` does not exist as a directory.
+      # c.f.  ccache happily creates the cache dir if it does not exist.
+      #       Don't create ccache's dir so that `action/cache` skips if ccache isn't populated.
+      New-Item -ItemType Directory -Path "${env:github_runner_temp}/.vcpkg-cache" -Force
+
+      # HACK: `actions/cache` will somehow pick up `tar` from the git installation. This bin dir is not on the PATH.
+      #       `tar` tries to find `gzip` (which is right next to it), and fails b/c the git bin dir isn't on the PATH.
+      #       This breaks attempts to download/upload caches.
+      #       Add git's bin dir to the PATH so that `actions/cache`'s invoke of `tar` can find `gzip`.
+      echo "PATH=${env:PATH};C:\\Program Files\\Git\\usr\\bin" >> "${env:GITHUB_ENV}"
+
+  # n.b. Caller is responsible for building w/ ccache enabled via `--use_cache`
+  #      We could set the env var `ORT_BUILD_WITH_CACHE` to indirectly enable it, but this hack seems deprecated.
+  #      Ideally ccache should be auto-enabled by the build system if ccache/sccache is detected on the PATH.
+  - name: Setup CCache
+    uses: actions/cache@v5
+    with:
+      # Fully qualify by `cache_key`. `actions/cache` does not isolate by workflow, unlike ADO cache actions.
+      key: ccache | ${{ inputs.cache_key }}
+      path: ${{ runner.temp }}/.ccache # ensure this is coordinated/matches w/ env-vars step above
+
+  - name: Setup VCPKG Cache
+    uses: actions/cache@v5
+    with:
+      key: vcpkg-cache | ${{ inputs.cache_key }}
+      path: ${{ runner.temp }}/.vcpkg-cache
diff --git a/.github/workflows/linux-wasm-ci-build-and-test-workflow.yml b/.github/workflows/linux-wasm-ci-build-and-test-workflow.yml
index c53c61242e6bc..6d2dac4068da6 100644
--- a/.github/workflows/linux-wasm-ci-build-and-test-workflow.yml
+++ b/.github/workflows/linux-wasm-ci-build-and-test-workflow.yml
@@ -45,21 +45,22 @@ jobs:
     env:
       buildArch: x64
       common_build_args: >-
-        --parallel
-        --use_cache
-        ${{ inputs.use_vcpkg == true && '--use_vcpkg --use_vcpkg_ms_internal_asset_cache' || '' }}
-        --config ${{ inputs.build_config }}
-        --skip_submodule_sync
         --build_wasm
+        --config '${{ inputs.build_config }}'
         --enable_wasm_simd
+        --parallel
+        --skip_submodule_sync
+        --use_cache
         ${{ inputs.enable_wasm_threads == true && '--enable_wasm_threads' || '' }}
+        ${{ inputs.use_vcpkg == true && '--use_vcpkg --use_vcpkg_ms_internal_asset_cache' || '' }}
         ${{ inputs.extra_build_args }}
+
       reduced_size_build_args: >-
-        --disable_ml_ops
         --disable_generation_ops
+        --disable_ml_ops
         --disable_types string float4 float8 optional sparsetensor
-        --include_ops_by_config onnxruntime/wasm/reduced_types.config
         --enable_reduced_operator_type_support
+        --include_ops_by_config onnxruntime/wasm/reduced_types.config
 
     steps:
       - name: Checkout code
@@ -106,20 +107,24 @@ jobs:
           disable-terrapin: 'true'
 
       - name: Build (simd + threads)
+        shell: bash
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py \
             ${{ env.common_build_args }} \
-            --build_dir ${{ github.workspace }}/build/wasm_inferencing \
+            --build_dir "${GITHUB_WORKSPACE}/build/wasm_inferencing" \
             ${{ inputs.build_config == 'Release' && '--enable_wasm_api_exception_catching' || '' }} \
             --skip_tests
         working-directory: ${{ github.workspace }}
 
       - name: Build (simd + threads + JSEP)
         if: ${{ inputs.build_jsep == true }}
+        shell: bash
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py \
             ${{ env.common_build_args }} \
-            --build_dir ${{ github.workspace }}/build/wasm_inferencing_jsep \
+            --build_dir "${GITHUB_WORKSPACE}/build/wasm_inferencing_jsep" \
             --use_jsep \
             --use_webnn \
             --target onnxruntime_webassembly \
@@ -129,10 +134,12 @@ jobs:
 
       - name: Build (simd + threads + WebGPU experimental)
         if: ${{ inputs.build_webgpu == true }}
+        shell: bash
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py \
             ${{ env.common_build_args }} \
-            --build_dir ${{ github.workspace }}/build/wasm_inferencing_webgpu \
+            --build_dir "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu" \
             --use_webgpu \
             --use_webnn \
             --target onnxruntime_webassembly \
@@ -143,10 +150,12 @@ jobs:
 
       - name: Build (simd + threads + WebGPU experimental, JSPI)
         if: ${{ inputs.build_webgpu == true }}
+        shell: bash
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py \
             ${{ env.common_build_args }} \
-            --build_dir ${{ github.workspace }}/build/wasm_inferencing_webgpu_jspi \
+            --build_dir "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu_jspi" \
             --use_webgpu \
             --use_webnn \
             --enable_wasm_jspi \
@@ -157,21 +166,25 @@ jobs:
 
       - name: Create Artifacts
         if: ${{ inputs.skip_publish != true }}
+        shell: bash
+        env:
+          build_cfg: ${{ inputs.build_config }}
         run: |
-          mkdir -p ${{ github.workspace }}/artifacts/wasm/
-          cp ${{ github.workspace }}/build/wasm_inferencing/${{ inputs.build_config }}/ort-wasm-simd-threaded.wasm ${{ github.workspace }}/artifacts/wasm/
-          cp ${{ github.workspace }}/build/wasm_inferencing/${{ inputs.build_config }}/ort-wasm-simd-threaded.mjs ${{ github.workspace }}/artifacts/wasm/
-          if [ -d ${{ github.workspace }}/build/wasm_inferencing_jsep ]; then
-            cp ${{ github.workspace }}/build/wasm_inferencing_jsep/${{ inputs.build_config }}/ort-wasm-simd-threaded.jsep.wasm ${{ github.workspace }}/artifacts/wasm/
-            cp ${{ github.workspace }}/build/wasm_inferencing_jsep/${{ inputs.build_config }}/ort-wasm-simd-threaded.jsep.mjs ${{ github.workspace }}/artifacts/wasm/
+          set -euo pipefail
+          mkdir -p "${GITHUB_WORKSPACE}/artifacts/wasm/"
+          cp "${GITHUB_WORKSPACE}/build/wasm_inferencing/${build_cfg}/ort-wasm-simd-threaded.wasm" "${GITHUB_WORKSPACE}/artifacts/wasm/"
+          cp "${GITHUB_WORKSPACE}/build/wasm_inferencing/${build_cfg}/ort-wasm-simd-threaded.mjs" "${GITHUB_WORKSPACE}/artifacts/wasm/"
+          if [ -d "${GITHUB_WORKSPACE}/build/wasm_inferencing_jsep" ]; then
+            cp "${GITHUB_WORKSPACE}/build/wasm_inferencing_jsep/${build_cfg}/ort-wasm-simd-threaded.jsep.wasm" "${GITHUB_WORKSPACE}/artifacts/wasm/"
+            cp "${GITHUB_WORKSPACE}/build/wasm_inferencing_jsep/${build_cfg}/ort-wasm-simd-threaded.jsep.mjs" "${GITHUB_WORKSPACE}/artifacts/wasm/"
           fi
-          if [ -d ${{ github.workspace }}/build/wasm_inferencing_webgpu ]; then
-            cp ${{ github.workspace }}/build/wasm_inferencing_webgpu/${{ inputs.build_config }}/ort-wasm-simd-threaded.asyncify.wasm ${{ github.workspace }}/artifacts/wasm/
-            cp ${{ github.workspace }}/build/wasm_inferencing_webgpu/${{ inputs.build_config }}/ort-wasm-simd-threaded.asyncify.mjs ${{ github.workspace }}/artifacts/wasm/
+          if [ -d "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu" ]; then
+            cp "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu/${build_cfg}/ort-wasm-simd-threaded.asyncify.wasm" "${GITHUB_WORKSPACE}/artifacts/wasm/"
+            cp "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu/${build_cfg}/ort-wasm-simd-threaded.asyncify.mjs" "${GITHUB_WORKSPACE}/artifacts/wasm/"
           fi
-          if [ -d ${{ github.workspace }}/build/wasm_inferencing_webgpu_jspi ]; then
-            cp ${{ github.workspace }}/build/wasm_inferencing_webgpu_jspi/${{ inputs.build_config }}/ort-wasm-simd-threaded.jspi.wasm ${{ github.workspace }}/artifacts/wasm/
-            cp ${{ github.workspace }}/build/wasm_inferencing_webgpu_jspi/${{ inputs.build_config }}/ort-wasm-simd-threaded.jspi.mjs ${{ github.workspace }}/artifacts/wasm/
+          if [ -d "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu_jspi" ]; then
+            cp "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu_jspi/${build_cfg}/ort-wasm-simd-threaded.jspi.wasm" "${GITHUB_WORKSPACE}/artifacts/wasm/"
+            cp "${GITHUB_WORKSPACE}/build/wasm_inferencing_webgpu_jspi/${build_cfg}/ort-wasm-simd-threaded.jspi.mjs" "${GITHUB_WORKSPACE}/artifacts/wasm/"
           fi
 
       - name: Upload WASM artifacts
@@ -184,20 +197,24 @@ jobs:
       - name: Test (Node.js) (simd + threads)
         # unit tests are currently only supported in Debug builds because they require exceptions, which are disabled in Release builds.
         if: ${{ inputs.build_config == 'Debug' }}
+        shell: bash
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py \
             ${{ env.common_build_args }} \
-            --build_dir ${{ github.workspace }}/build/wasm_inferencing \
+            --build_dir build/wasm_inferencing \
             --test
         working-directory: ${{ github.workspace }}
 
       - name: Test (browser) (simd + threads)
         # unit tests are currently only supported in Debug builds because they require exceptions, which are disabled in Release builds.
         if: ${{ inputs.build_config == 'Debug' }}
+        shell: bash
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py \
             ${{ env.common_build_args }} \
-            --build_dir ${{ github.workspace }}/build/wasm_inferencing \
+            --build_dir build/wasm_inferencing \
             --wasm_run_tests_in_browser \
             --targets onnxruntime_test_all onnxruntime_provider_test \
             --update --build --test
diff --git a/.github/workflows/macos-ci-build-and-test-workflow.yml b/.github/workflows/macos-ci-build-and-test-workflow.yml
index 76198c7f5c1ce..34bf5321e63fa 100644
--- a/.github/workflows/macos-ci-build-and-test-workflow.yml
+++ b/.github/workflows/macos-ci-build-and-test-workflow.yml
@@ -52,7 +52,8 @@ jobs:
     # see also: https://github.com/actions/runner-images/blob/main/README.md
     runs-on: ${{ matrix.machine == 'x86_64' && 'macos-13' || 'macos-15' }}
     env:
-      build_flags: >
+      CCACHE_DIR: ~/.cache/ccache # be explicit. must match `action/cache` below.
+      build_flags: >-
         --build_dir ./build
         --skip_submodule_sync
         --parallel
@@ -66,7 +67,9 @@ jobs:
         ${{ inputs.use_webgpu && '--use_webgpu' || '' }}
         ${{ inputs.use_xnnpack && '--use_xnnpack' || '' }}
         ${{ inputs.use_coreml && '--use_coreml --skip_onnx_tests' || '' }}
-        --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+        --use_cache
+        --use_vcpkg
+        --use_vcpkg_ms_internal_asset_cache
         --config ${{ matrix.build_config }}
         --osx_arch ${{ matrix.target }}
 
@@ -76,13 +79,15 @@ jobs:
     steps:
       - name: Checkout code
         uses: actions/checkout@v6
-      - uses: microsoft/onnxruntime-github-actions/setup-build-tools@v0.0.9
+
+      - uses: microsoft/onnxruntime-github-actions/setup-build-tools@8bad63a3c05d448311dfa8e5f531171c97471aa1 # v0.0.12
         with:
+          ccache-version: 4.13.1
+          ccache-hash: d831cbd84ac1efb32441c5bfbc7cc662830c0c22e96c5b5b27a6ce4746c81d6076261cf61217fb67ff4585fbd6cbd91ea817aa4ed904b81fcba621bb60881f56
           vcpkg-version: '2025.08.27'
           vcpkg-hash: 9a4b32849792e13bee1d24726f073b3881acae4165206ddf1a6378e44a4ddd05b3ee93f55ff46d8e8873b3cbcd06606212989e248f0bd615a5bf365070074079
           cmake-version: '3.31.8'
           cmake-hash: 99cc9c63ae49f21253efb5921de2ba84ce136018abf08632c92c060ba91d552e0f6acc214e9ba8123dee0cf6d1cf089ca389e321879fd9d719a60d975bcffcc8
-          add-cmake-to-path: 'true'
           disable-terrapin: 'true'
 
       - name: macOS CI pipeline prepare steps
@@ -93,6 +98,18 @@ jobs:
           xcode_version: ${{ env.xcode_version }}
           use_cache: true
 
+      - name: Setup CCache
+        uses: actions/cache@v5
+        with:
+          key: ccache | macos-ci-build-and-test-workflow.yml | build-and-test | ${{ matrix.machine }}-${{ matrix.target }}-${{ matrix.build_config }}
+          path: ~/.cache/ccache
+
+      - name: Setup VCPKG Cache
+        uses: actions/cache@v5
+        with:
+          key: vcpkg-cache | macos-ci-build-and-test-workflow.yml | build-and-test | ${{ matrix.machine }}-${{ matrix.target }}-${{ matrix.build_config }}
+          path: ~/.cache/vcpkg
+
       - uses: actions/cache@v5
         id: onnx-node-tests-cache
         with:
@@ -102,7 +119,7 @@ jobs:
       - name: Setup environment variables
         shell: bash
         run: |
-          set -e -x
+          set -xeuo pipefail
           export PATH=${{ github.workspace }}/build/installed/bin:$PATH
           export ONNX_ML=1
           export CMAKE_ARGS="-DONNX_GEN_PB_TYPE_STUBS=ON -DONNX_WERROR=OFF"
@@ -112,6 +129,8 @@ jobs:
         shell: bash
         working-directory: ${{ github.workspace }}
         run: |
+          set -euo pipefail
+          ccache --zero-stats
           rm -rf ${{ github.workspace }}/build/${{ matrix.build_config }}
           python ./tools/ci_build/build.py --update ${{ env.build_flags }}
 
@@ -119,12 +138,20 @@ jobs:
         shell: bash
         working-directory: ${{ github.workspace }}
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py --build ${{ env.build_flags }}
 
+      - name: CCache stats
+        shell: bash
+        working-directory: ${{ github.workspace }}
+        run: |
+          set -euo pipefail
+          ccache --show-stats -vv
+
       - name: Install
         shell: bash
         run: |
-          set -e -x
+          set -xeuo pipefail
           rm -rf ${{ github.workspace }}/build/installed
           cd ${{ github.workspace }}/build/${{ matrix.build_config }}
           make install DESTDIR=${{ github.workspace }}/build/installed
@@ -135,4 +162,5 @@ jobs:
         shell: bash
         working-directory: ${{ github.workspace }}
         run: |
+          set -euo pipefail
           python ./tools/ci_build/build.py --test ${{ env.build_flags }}
diff --git a/.github/workflows/windows_build_x64_asan.yml b/.github/workflows/windows_build_x64_asan.yml
index 265ea2bf8dcab..3dddb6e5e38d5 100644
--- a/.github/workflows/windows_build_x64_asan.yml
+++ b/.github/workflows/windows_build_x64_asan.yml
@@ -37,11 +37,30 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env # Use the composite action
         with:
           architecture: x64
+          cache_key: windows_build_x64_asan.yml | build_x64
 
       - name: Build and Test (Combined)
-        shell: cmd
+        shell: pwsh
         run: |
-          @echo off
-          echo %PATH%
-          python -m pip install -r "%GITHUB_WORKSPACE%\tools\ci_build/github/windows\python\requirements.txt"
-          python "%GITHUB_WORKSPACE%\tools\ci_build\build.py" --config Debug --build_dir "%RUNNER_TEMP%\build" --skip_submodule_sync --parallel --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_generator "Visual Studio 17 2022" --disable_memleak_checker --enable_address_sanitizer
+          ccache --zero-stats
+
+          echo ${env:PATH}
+          python -m pip install -r "${env:GITHUB_WORKSPACE}\tools\ci_build\github\windows\python\requirements.txt"
+          python "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
+            --build_dir "${env:RUNNER_TEMP}\build" `
+            --cmake_generator "Ninja" `
+            --config Debug `
+            --disable_memleak_checker `
+            --enable_address_sanitizer `
+            --parallel `
+            --skip_submodule_sync `
+            --use_cache `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
+          if ($LASTEXITCODE -ne 0) {
+            exit $LASTEXITCODE
+          }
+
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
diff --git a/.github/workflows/windows_cuda.yml b/.github/workflows/windows_cuda.yml
index 852d0164083c4..28c2e03c93137 100644
--- a/.github/workflows/windows_cuda.yml
+++ b/.github/workflows/windows_cuda.yml
@@ -39,6 +39,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_cuda.yml | build
 
       - name: Install python modules
         run: python -m pip install -r .\tools\ci_build\github\windows\python\requirements.txt
@@ -79,7 +80,7 @@ jobs:
       - name: API Documentation Check and generate
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ runner.temp }}\build\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -109,19 +110,45 @@ jobs:
 
       - name: Build and Clean Binaries
         working-directory: ${{ runner.temp }}
+        shell: pwsh
         run: |
+          ccache --zero-stats
+
           npm install -g typescript
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
+
           # Execute the build process
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --update --build --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel --nvcc_threads 1 --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --build_wheel --build_java --use_cuda --cuda_home="$env:RUNNER_TEMP\v12.8" --enable_cuda_profiling --use_vcpkg --use_vcpkg_ms_internal_asset_cache --enable_transformers_tool_test --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build `
+            --build_csharp `
+            --build_dir build `
+            --build_java `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86 `
+            --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --cuda_home="$env:RUNNER_TEMP\v12.8" `
+            --enable_cuda_profiling `
+            --enable_transformers_tool_test `
+            --nvcc_threads 1 `
+            --parallel `
+            --skip_submodule_sync `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_cuda `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
 
           # Clean up the output directory before uploading artifacts
-          $outputDir = "${{ runner.temp }}\build\RelWithDebInfo"
+          $outputDir = "$env:RUNNER_TEMP\build\RelWithDebInfo"
           Write-Host "Cleaning up files from $outputDir..."
 
           Remove-Item -Path "$outputDir\onnxruntime" -Recurse -Force -ErrorAction SilentlyContinue
@@ -133,7 +160,10 @@ jobs:
           Remove-Item -Path "$outputDir\CMakeFiles" -Recurse -Force -ErrorAction SilentlyContinue
           # Remove intermediate object files as in the original script
           Remove-Item -Path $outputDir -Include "*.obj" -Recurse
+
+      - name: CCache Stats
         shell: pwsh
+        run: ccache --show-stats -vv
 
       - name: Upload build artifacts
         uses: actions/upload-artifact@v6
@@ -190,6 +220,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_cuda.yml | build # reuse build's caches. we do not expect this stage to modify them.
 
       - name: Install python modules
         run: python -m pip install -r .\tools\ci_build\github\windows\python\requirements.txt
@@ -220,9 +251,10 @@ jobs:
       - name: Install ONNX Runtime Wheel
         uses: ./.github/actions/install-onnxruntime-wheel
         with:
-          whl-directory: ${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo\dist
+          whl-directory: ${{ runner.temp }}\build\RelWithDebInfo\dist
 
       - name: Run Tests
+        shell: pwsh
         working-directory: ${{ runner.temp }}
         run: |
           npm install -g typescript
@@ -230,16 +262,36 @@ jobs:
             exit $lastExitCode
           }
 
-          python.exe ${{ github.workspace }}\tools\python\update_ctest_path.py   "${{ runner.temp }}\build\RelWithDebInfo\CTestTestfile.cmake" "${{ runner.temp }}\build\RelWithDebInfo"
+          python.exe ${env:GITHUB_WORKSPACE}\tools\python\update_ctest_path.py "${env:RUNNER_TEMP}\build\RelWithDebInfo\CTestTestfile.cmake" "${env:RUNNER_TEMP}\build\RelWithDebInfo"
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
 
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --test --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel --nvcc_threads 1 --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --build_wheel --build_java --use_cuda --cuda_home="$env:RUNNER_TEMP\v12.8" --enable_cuda_profiling --use_vcpkg --use_vcpkg_ms_internal_asset_cache --enable_transformers_tool_test --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build_csharp `
+            --build_dir build `
+            --build_java `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86  `
+            --cmake_extra_defines onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --cuda_home "${env:RUNNER_TEMP}\v12.8" `
+            --enable_cuda_profiling `
+            --enable_transformers_tool_test `
+            --nvcc_threads 1 `
+            --parallel `
+            --skip_submodule_sync `
+            --test `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_cuda `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
-        shell: pwsh
 
       - name: Validate C# native delegates
         run: python tools\ValidateNativeDelegateAttributes.py
diff --git a/.github/workflows/windows_dml.yml b/.github/workflows/windows_dml.yml
index de5eebdc86da0..c6d010a716c2c 100644
--- a/.github/workflows/windows_dml.yml
+++ b/.github/workflows/windows_dml.yml
@@ -45,6 +45,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_dml.yml | build_x64_RelWithDebInfo
 
       - name: Install python modules
         run: python -m pip install -r .\tools\ci_build\github\windows\python\requirements.txt
@@ -64,7 +65,7 @@ jobs:
       - name: API Documentation Check and generate
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -93,14 +94,35 @@ jobs:
           echo "OnnxRuntimeBuildDirectory=$buildDir" >> $env:GITHUB_ENV
 
       - name: Build and Test
+        shell: pwsh
         working-directory: ${{ runner.temp }}
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --build_dir build --skip_submodule_sync --build_csharp --parallel  --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --skip_onnx_tests --build_wheel --use_dml --enable_wcos --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+          ccache --zero-stats
+
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build_csharp `
+            --build_dir build `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --enable_wcos `
+            --parallel `
+            --skip_onnx_tests `
+            --skip_submodule_sync `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_dml `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
-          Remove-Item "${{ github.workspace }}\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\RelWithDebInfo" -Include "*.obj" -Recurse
+
+      - name: CCache Stats
         shell: pwsh
+        run: ccache --show-stats -vv
 
       - name: Validate C# native delegates
         run: python tools\ValidateNativeDelegateAttributes.py
@@ -110,4 +132,4 @@ jobs:
       - name: Install ONNX Runtime Wheel
         uses: ./.github/actions/install-onnxruntime-wheel
         with:
-          whl-directory: ${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo\dist
+          whl-directory: ${{ runner.temp }}\build\RelWithDebInfo\dist
diff --git a/.github/workflows/windows_openvino.yml b/.github/workflows/windows_openvino.yml
index 8ff7a7071a755..040293a125380 100644
--- a/.github/workflows/windows_openvino.yml
+++ b/.github/workflows/windows_openvino.yml
@@ -50,6 +50,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_openvino.yml | BUILD_OPENVINO_EP
 
       - name: Download OpenVINO Toolkit v2025.4.1
         env:
@@ -142,28 +143,62 @@ jobs:
 
 
 
-      - name: Generate onnxruntime.sln
+      - name: Generate build files
         shell: pwsh
+        env:
+          build_cfg: ${{ env.BuildConfig }}
         working-directory: ${{ runner.temp }}
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --config ${{ env.BuildConfig }} --build_dir build --cmake_generator "Visual Studio 17 2022" --build_shared_lib --use_openvino CPU --use_binskim_compliant_compile_flags --update --parallel
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build_dir build `
+            --build_shared_lib `
+            --cmake_generator "Ninja" `
+            --config ${env:build_cfg} `
+            --parallel `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_openvino CPU
 
       - name: Build
         shell: pwsh
+        env:
+          build_cfg: ${{ env.BuildConfig }}
         working-directory: ${{ runner.temp }}
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --config ${{ env.BuildConfig }} --build_dir build --cmake_generator "Visual Studio 17 2022" --build_shared_lib --use_openvino CPU --use_binskim_compliant_compile_flags --build --parallel
+          ccache --zero-stats
+
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build `
+            --build_dir build `
+            --build_shared_lib `
+            --cmake_generator "Ninja" `
+            --config ${env:build_cfg} `
+            --parallel `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_openvino CPU
+          if ($LASTEXITCODE -ne 0) {
+            exit $LASTEXITCODE
+          }
 
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
 
       - name: Run unit tests
         shell: pwsh
+        env:
+          build_cfg: ${{ env.BuildConfig }}
         working-directory: ${{ runner.temp }}
         run: |
-          python "${{ github.workspace }}\tools\ci_build\build.py" `
-            --config "${{ env.BuildConfig }}" `
+          python "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
             --build_dir build `
-            --cmake_generator "Visual Studio 17 2022" `
             --build_shared_lib `
-            --use_openvino CPU `
+            --cmake_generator "Ninja" `
+            --config "${env:build_cfg}" `
+            --enable_onnx_tests `
+            --test `
             --use_binskim_compliant_compile_flags `
-            --test --enable_onnx_tests
+            --use_cache `
+            --use_openvino CPU
diff --git a/.github/workflows/windows_qnn_x64.yml b/.github/workflows/windows_qnn_x64.yml
index 7175b81e227c9..ca44de881a50c 100644
--- a/.github/workflows/windows_qnn_x64.yml
+++ b/.github/workflows/windows_qnn_x64.yml
@@ -47,6 +47,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_qnn_x64.yml | build_test_qnn_ep_${{ matrix.QnnLibKind }}
 
       - name: Download QNN SDK
         working-directory: ${{ runner.temp }}
@@ -64,19 +65,44 @@ jobs:
           dir $qnn_sdk_path
 
       - name: Build and Test
-        shell: cmd
+        shell: pwsh
+        env:
+          qnn_kind: ${{ matrix.QnnLibKind }}
         run: |
-          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --build_dir ${{ runner.temp }}\build --cmake_generator "Visual Studio 17 2022" --build_java --build_shared_lib --use_qnn ${{ matrix.QnnLibKind }} --qnn_home %QNN_SDK_ROOT% --use_binskim_compliant_compile_flags --update --build --test --enable_onnx_tests --parallel
+          ccache --zero-stats
+
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build `
+            --build_dir "${env:RUNNER_TEMP}\build" `
+            --build_java `
+            --build_shared_lib `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --enable_onnx_tests `
+            --parallel `
+            --qnn_home "${env:QNN_SDK_ROOT}" `
+            --test `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_qnn ${env:qnn_kind}
+          if ($LASTEXITCODE -ne 0) {
+            exit $LASTEXITCODE
+          }
+
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
 
       - name: Run ONNX Tests
         shell: cmd
-        working-directory: ${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo
+        working-directory: ${{ runner.temp }}\build\RelWithDebInfo
         run: |
           .\onnx_test_runner -j 1 -e qnn -i "backend_path|%QNN_SDK_ROOT%\lib\x86_64-windows-msvc\QnnCpu.dll" ${{ github.workspace }}\cmake\external\onnx\onnx\backend\test\data\node
 
       - name: Run float32 model tests
         shell: cmd
-        working-directory: ${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo
+        working-directory: ${{ runner.temp }}\build\RelWithDebInfo
         run: |
           rem This step assumes the model data exists at C:\data\float32_models on the runner
           if exist C:\data\float32_models (
diff --git a/.github/workflows/windows_tensorrt.yml b/.github/workflows/windows_tensorrt.yml
index 0a47f46aa8516..185a721e2fde6 100644
--- a/.github/workflows/windows_tensorrt.yml
+++ b/.github/workflows/windows_tensorrt.yml
@@ -39,6 +39,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_tensorrt.yml | build
 
       - name: Install python modules
         run: python -m pip install -r .\tools\ci_build\github\windows\python\requirements.txt
@@ -85,7 +86,7 @@ jobs:
       - name: API Documentation Check and generate
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ runner.temp }}\build\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -115,19 +116,40 @@ jobs:
 
       - name: Build and Clean Binaries
         working-directory: ${{ runner.temp }}
+        shell: pwsh
         run: |
+          ccache --zero-stats
+
           npm install -g typescript
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
           # Execute the build process
-          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --parallel --nvcc_threads 1 --use_binskim_compliant_compile_flags --build_dir build --skip_submodule_sync --build_shared_lib --build --update --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests --use_tensorrt --tensorrt_home="${{ runner.temp }}\TensorRT-10.14.1.48.Windows.win10.cuda-12.9" --cuda_home="${{ runner.temp }}\v12.8" --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
+          python ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build `
+            --build_dir "${env:RUNNER_TEMP}\build" `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86 `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --cuda_home "${env:RUNNER_TEMP}\v12.8" `
+            --enable_onnx_tests `
+            --nvcc_threads 1 `
+            --parallel `
+            --skip_submodule_sync `
+            --tensorrt_home "${env:RUNNER_TEMP}\TensorRT-10.14.1.48.Windows.win10.cuda-12.9" `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_tensorrt `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
 
           # Clean up the output directory before uploading artifacts
-          $outputDir = "${{ runner.temp }}\build\RelWithDebInfo"
+          $outputDir = "${env:RUNNER_TEMP}\build\RelWithDebInfo"
           Write-Host "Cleaning up files from $outputDir..."
 
           Remove-Item -Path "$outputDir\onnxruntime" -Recurse -Force -ErrorAction SilentlyContinue
@@ -139,7 +161,10 @@ jobs:
           Remove-Item -Path "$outputDir\CMakeFiles" -Recurse -Force -ErrorAction SilentlyContinue
           # Remove intermediate object files as in the original script
           Remove-Item -Path $outputDir -Include "*.obj" -Recurse
+
+      - name: CCache Stats
         shell: pwsh
+        run: ccache --show-stats -vv
 
       - name: Upload build artifacts
         uses: actions/upload-artifact@v6
@@ -196,6 +221,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_tensorrt.yml | build # re-use build's caches, we do not expect to modify/extend them.
 
       - name: Install python modules
         run: python -m pip install -r .\tools\ci_build\github\windows\python\requirements.txt
@@ -232,26 +258,44 @@ jobs:
       - name: Install ONNX Runtime Wheel
         uses: ./.github/actions/install-onnxruntime-wheel
         with:
-          whl-directory: ${{ runner.temp }}\build\RelWithDebInfo\RelWithDebInfo\dist
+          whl-directory: ${{ runner.temp }}\build\RelWithDebInfo\dist
 
       - name: Run Tests
         working-directory: ${{ runner.temp }}
+        shell: pwsh
         run: |
           npm install -g typescript
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
 
-          python.exe ${{ github.workspace }}\tools\python\update_ctest_path.py   "${{ runner.temp }}\build\RelWithDebInfo\CTestTestfile.cmake" "${{ runner.temp }}\build\RelWithDebInfo"
+          python.exe ${env:GITHUB_WORKSPACE}\tools\python\update_ctest_path.py   "${env:RUNNER_TEMP}\build\RelWithDebInfo\CTestTestfile.cmake" "${env:RUNNER_TEMP}\build\RelWithDebInfo"
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
 
-          python ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --use_binskim_compliant_compile_flags --parallel --nvcc_threads 1 --build_dir build --skip_submodule_sync --build_shared_lib --test --cmake_generator "Visual Studio 17 2022" --build_wheel --enable_onnx_tests --use_tensorrt --tensorrt_home="${{ runner.temp }}\TensorRT-10.14.1.48.Windows.win10.cuda-12.9" --cuda_home="${{ runner.temp }}\v12.8" --use_vcpkg --use_vcpkg_ms_internal_asset_cache --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
+          python ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build_dir build `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86 `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --cuda_home "${env:RUNNER_TEMP}\v12.8" `
+            --enable_onnx_tests `
+            --nvcc_threads 1 `
+            --parallel `
+            --skip_submodule_sync `
+            --tensorrt_home "${env:RUNNER_TEMP}\TensorRT-10.14.1.48.Windows.win10.cuda-12.9" `
+            --test `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_tensorrt `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
-        shell: pwsh
 
       - name: Validate C# native delegates
         run: python tools\ValidateNativeDelegateAttributes.py
diff --git a/.github/workflows/windows_webgpu.yml b/.github/workflows/windows_webgpu.yml
index e67eda41d2e0e..4904ecc79e365 100644
--- a/.github/workflows/windows_webgpu.yml
+++ b/.github/workflows/windows_webgpu.yml
@@ -53,6 +53,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_webgpu.yml | webgpu_build_x64_RelWithDebInfo_${{ matrix.vcpkg_option }}_${{ matrix.wgsl_template }}
 
       - name: Install python modules
         run: python -m pip install -r tools\ci_build\github\windows\python\requirements.txt
@@ -74,7 +75,7 @@ jobs:
       - name: API Documentation Check and generate
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -107,29 +108,31 @@ jobs:
 
       - name: Build and Test
         shell: pwsh
+        env:
+          wgsl_template: ${{ matrix.wgsl_template }}
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py `
-            --config RelWithDebInfo `
-            --build_dir ${{ github.workspace }} `
-            --skip_submodule_sync `
+          python.exe "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
             --build_csharp `
-            --parallel `
-            --use_binskim_compliant_compile_flags `
-            --cmake_generator "Visual Studio 17 2022" `
+            --build_dir "${env:GITHUB_WORKSPACE}" `
+            --build_java `
+            --build_nodejs `
             --build_shared_lib `
+            --cmake_extra_defines onnxruntime_BUILD_DAWN_SHARED_LIBRARY=ON `
+            --cmake_extra_defines onnxruntime_BUILD_UNIT_TESTS=ON `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
             --enable_onnx_tests `
-            --build_nodejs `
-            --build_java `
+            --parallel `
+            --skip_submodule_sync `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
             --use_webgpu `
-            --wgsl_template ${{ matrix.wgsl_template }} `
-            ${{ matrix.vcpkg_option == 'vcpkg' && '--use_vcpkg --use_vcpkg_ms_internal_asset_cache' || '' }} `
-            --cmake_extra_defines `
-            onnxruntime_BUILD_UNIT_TESTS=ON `
-            onnxruntime_BUILD_DAWN_SHARED_LIBRARY=ON
+            --wgsl_template "${env:wgsl_template}" `
+            ${{ matrix.vcpkg_option == 'vcpkg' && '--use_vcpkg --use_vcpkg_ms_internal_asset_cache' || '' }}
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
-          Remove-Item "${{ github.workspace }}\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\RelWithDebInfo" -Include "*.obj" -Recurse
 
       - name: Run tests (onnxruntime_test_all, onnxruntime_provider_test) with verbose logging
         shell: pwsh
@@ -137,17 +140,17 @@ jobs:
           $env:ORT_UNIT_TEST_MAIN_LOG_LEVEL = "0"
           .\onnxruntime_test_all.exe 2> .\onnxruntime_test_stderr.log
           .\onnxruntime_provider_test.exe 2>> .\onnxruntime_test_stderr.log
-        working-directory: ${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo
+        working-directory: ${{ github.workspace }}\RelWithDebInfo
 
       - name: Check log file
         shell: cmd
         run: |
-          dir ${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo\onnxruntime_test_stderr.log
+          dir ${{ github.workspace }}\RelWithDebInfo\onnxruntime_test_stderr.log
 
       - name: Validate shader keys
         uses: ./.github/actions/webgpu-validate-shader-key
         with:
-          log_file_path: ${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo\onnxruntime_test_stderr.log
+          log_file_path: ${{ github.workspace }}\RelWithDebInfo\onnxruntime_test_stderr.log
 
       - name: Validate C# native delegates
         run: python tools\ValidateNativeDelegateAttributes.py
@@ -186,6 +189,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_webgpu.yml | webgpu_plugin_build_x64_RelWithDebInfo
 
       - name: Install python modules
         run: python -m pip install -r tools\ci_build\github\windows\python\requirements.txt
@@ -212,7 +216,7 @@ jobs:
             --skip_submodule_sync `
             --parallel `
             --use_binskim_compliant_compile_flags `
-            --cmake_generator "Visual Studio 17 2022" `
+            --cmake_generator "Ninja" `
             --enable_onnx_tests `
             --use_webgpu shared_lib `
             --wgsl_template static `
@@ -259,31 +263,60 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_webgpu.yml | webgpu_external_dawn_build_x64_RelWithDebInfo
 
       - name: Install python modules
         run: python -m pip install -r tools\ci_build\github\windows\python\requirements.txt
         shell: cmd
         working-directory: ${{ github.workspace }}
 
-      - name: Generate onnxruntime.sln
+      - name: Generate build files
         shell: pwsh
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --build_dir ${{ github.workspace }} --skip_submodule_sync --update --parallel --cmake_generator "Visual Studio 17 2022" --use_webgpu --use_external_dawn --skip_tests --target onnxruntime_webgpu_external_dawn_test
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build_dir ${env:GITHUB_WORKSPACE} `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --parallel `
+            --skip_submodule_sync `
+            --skip_tests `
+            --target onnxruntime_webgpu_external_dawn_test `
+            --update `
+            --use_cache `
+            --use_external_dawn `
+            --use_webgpu
 
       - name: Build
         shell: pwsh
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --build_dir ${{ github.workspace }} --skip_submodule_sync --build --parallel --cmake_generator "Visual Studio 17 2022" --use_webgpu --use_external_dawn --skip_tests --target onnxruntime_webgpu_external_dawn_test
+          ccache --zero-stats
+
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build `
+            --build_dir ${env:GITHUB_WORKSPACE} `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --parallel `
+            --skip_submodule_sync `
+            --skip_tests `
+            --target onnxruntime_webgpu_external_dawn_test `
+            --use_cache `
+            --use_external_dawn `
+            --use_webgpu
+
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
 
       - name: Run tests (onnxruntime_webgpu_external_dawn_test)
         run: onnxruntime_webgpu_external_dawn_test.exe
         shell: cmd
-        working-directory: ${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo
+        working-directory: ${{ github.workspace }}\RelWithDebInfo
 
       - name: Run tests (onnxruntime_webgpu_external_dawn_test) - no_proc_table
         run: onnxruntime_webgpu_external_dawn_test.exe --no_proc_table
         shell: cmd
-        working-directory: ${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo
+        working-directory: ${{ github.workspace }}\RelWithDebInfo
 
   webgpu_minimal_build_edge_build_x64_RelWithDebInfo:
     runs-on: [
@@ -316,6 +349,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_webgpu.yml | webgpu_minimal_build_edge_build_x64_RelWithDebInfo
 
       - name: Install python modules
         run: python -m pip install -r tools\ci_build\github\windows\python\requirements.txt
@@ -337,7 +371,7 @@ jobs:
       - name: API Documentation Check and generate
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -365,11 +399,32 @@ jobs:
       - name: Build
         shell: pwsh
         run: |
-          python.exe ${{ github.workspace }}\tools\ci_build\build.py --config RelWithDebInfo --build_dir ${{ github.workspace }} --skip_submodule_sync --build_csharp --parallel  --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --update --build --build_shared_lib --disable_exceptions --disable_rtti --enable_msvc_static_runtime --enable_reduced_operator_type_support --skip_tests --use_binskim_compliant_compile_flags --cmake_extra_defines onnxruntime_BUILD_UNIT_TESTS=OFF onnxruntime_DISABLE_SPARSE_TENSORS=ON onnxruntime_DISABLE_OPTIONAL_TYPE=ON --minimal_build extended --use_webgpu
+          python.exe ${env:GITHUB_WORKSPACE}\tools\ci_build\build.py `
+            --build `
+            --build_csharp `
+            --build_dir ${env:GITHUB_WORKSPACE} `
+            --build_shared_lib `
+            --cmake_extra_defines onnxruntime_BUILD_UNIT_TESTS=OFF `
+            --cmake_extra_defines onnxruntime_DISABLE_OPTIONAL_TYPE=ON `
+            --cmake_extra_defines onnxruntime_DISABLE_SPARSE_TENSORS=ON `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --disable_exceptions `
+            --disable_rtti `
+            --enable_msvc_static_runtime `
+            --enable_reduced_operator_type_support `
+            --minimal_build extended `
+            --parallel `
+            --skip_submodule_sync `
+            --skip_tests `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_webgpu
           if ($lastExitCode -ne 0) {
             exit $lastExitCode
           }
-          Remove-Item "${{ github.workspace }}\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\RelWithDebInfo" -Include "*.obj" -Recurse
 
       - name: Validate C# native delegates
         run: python tools\ValidateNativeDelegateAttributes.py
diff --git a/.github/workflows/windows_x64_debug_build_x64_debug.yml b/.github/workflows/windows_x64_debug_build_x64_debug.yml
index 134326b8f57de..19ef393dada15 100644
--- a/.github/workflows/windows_x64_debug_build_x64_debug.yml
+++ b/.github/workflows/windows_x64_debug_build_x64_debug.yml
@@ -36,6 +36,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env # Use the composite action
         with:
           architecture: x64
+          cache_key: windows_x64_debug_build_x64_debug.yml | build_x64_debug
 
       - name: Install python modules
         shell: cmd
@@ -57,7 +58,7 @@ jobs:
         shell: cmd
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\build\Debug\Debug
+          set ORT_DOXY_OUT=${{ github.workspace }}\build\Debug
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -89,16 +90,33 @@ jobs:
       - name: Build and Test
         shell: pwsh
         run: |
-          python.exe "${{ github.workspace }}\tools\ci_build\build.py" --config Debug --build_dir "${{ github.workspace }}\build" --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --enable_onnx_tests --build_java --build_nodejs --build_wheel --disable_memleak_checker --msbuild_extra_options "IncludeMobileTargets=false" --build_nuget --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+          python.exe "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
+            --build_csharp `
+            --build_dir "${env:GITHUB_WORKSPACE}\build" `
+            --build_java `
+            --build_nodejs `
+            --build_nuget `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_generator "Ninja" `
+            --config Debug `
+            --disable_memleak_checker `
+            --enable_onnx_tests `
+            --msbuild_extra_options "IncludeMobileTargets=false" `
+            --parallel `
+            --skip_submodule_sync `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($LASTEXITCODE -ne 0) {
             exit $LASTEXITCODE
           }
-          Remove-Item "${{ github.workspace }}\build\Debug" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\build\Debug" -Include "*.obj" -Recurse
         env: # Set environment variables here, applies to this step only
           ALLOW_RELEASED_ONNX_OPSET_ONLY: '0'
           DocUpdateNeeded: 'false' # Can be set dynamically based on build output if needed
 
-
       - name: Validate C# native delegates
         shell: cmd
         run: python tools\ValidateNativeDelegateAttributes.py
@@ -109,7 +127,7 @@ jobs:
         run: |
           python -m pip uninstall -y onnxruntime onnxruntime-gpu onnxruntime-training onnxruntime-directml -qq
           Get-ChildItem -Path dist/*.whl | foreach {pip --disable-pip-version-check install --upgrade $_.fullname}
-        working-directory: "${{ github.workspace }}\\build\\Debug\\Debug"
+        working-directory: "${{ github.workspace }}\\build\\Debug"
 
     # Publish artifacts only on failure and if DocUpdateNeeded is true (example)
       - name: Publish OperatorKernels.md (Conditional)
diff --git a/.github/workflows/windows_x64_release_build_x64_release.yml b/.github/workflows/windows_x64_release_build_x64_release.yml
index 7acbdcd404ad3..6977d4493f8b2 100644
--- a/.github/workflows/windows_x64_release_build_x64_release.yml
+++ b/.github/workflows/windows_x64_release_build_x64_release.yml
@@ -36,6 +36,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_x64_release_build_x64_release.yml | build_x64_release
 
       - name: Install python modules
         shell: cmd
@@ -57,7 +58,7 @@ jobs:
         shell: cmd
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -89,15 +90,38 @@ jobs:
       - name: Build and Test
         shell: pwsh
         run: |
-          python.exe "${{ github.workspace }}\tools\ci_build\build.py" --config RelWithDebInfo --build_dir "${{ github.workspace }}\build" --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --enable_onnx_tests --build_wheel --build_java --build_nodejs --msbuild_extra_options "IncludeMobileTargets=false" --build_nuget --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+          ccache --zero-stats
+
+          python.exe "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
+            --build_csharp `
+            --build_dir "${env:GITHUB_WORKSPACE}\build" `
+            --build_java `
+            --build_nodejs `
+            --build_nuget `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --enable_onnx_tests `
+            --msbuild_extra_options "IncludeMobileTargets=false" `
+            --parallel `
+            --skip_submodule_sync `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($LASTEXITCODE -ne 0) {
             exit $LASTEXITCODE
           }
-          Remove-Item "${{ github.workspace }}\build\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\build\RelWithDebInfo" -Include "*.obj" -Recurse
         env:
           ALLOW_RELEASED_ONNX_OPSET_ONLY: '0'
           DocUpdateNeeded: 'false'
 
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
+
       - name: Run onnxruntime_provider_test with example_plugin_ep
         shell: pwsh
         run: |
@@ -124,7 +148,7 @@ jobs:
           "@
 
           .\onnxruntime_provider_test.exe
-        working-directory: ${{ github.workspace }}\build\RelWithDebInfo\RelWithDebInfo
+        working-directory: ${{ github.workspace }}\build\RelWithDebInfo
 
       - name: Validate C# native delegates
         shell: cmd
@@ -136,7 +160,7 @@ jobs:
         run: |
           python -m pip uninstall -y onnxruntime onnxruntime-gpu onnxruntime-training onnxruntime-directml -qq
           Get-ChildItem -Path dist/*.whl | foreach {pip --disable-pip-version-check install --upgrade $_.fullname}
-        working-directory: "${{ github.workspace }}\\build\\RelWithDebInfo\\RelWithDebInfo"
+        working-directory: "${{ github.workspace }}\\build\\RelWithDebInfo"
 
       - name: Publish OperatorKernels.md (Conditional)
         uses: actions/upload-artifact@v6
diff --git a/.github/workflows/windows_x64_release_ep_generic_interface_build_x64_release_ep_generic_interface.yml b/.github/workflows/windows_x64_release_ep_generic_interface_build_x64_release_ep_generic_interface.yml
index 8cb5bad664921..74fb24274d046 100644
--- a/.github/workflows/windows_x64_release_ep_generic_interface_build_x64_release_ep_generic_interface.yml
+++ b/.github/workflows/windows_x64_release_ep_generic_interface_build_x64_release_ep_generic_interface.yml
@@ -36,6 +36,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_x64_release_ep_generic_interface_build_x64_release_ep_generic_interface.yml | build_x64_release_ep_generic_interface
 
       - name: Install python modules
         shell: cmd
@@ -57,7 +58,7 @@ jobs:
         shell: cmd
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -83,14 +84,35 @@ jobs:
       - name: Build and test
         shell: pwsh
         run: |
-          python.exe "${{ github.workspace }}\tools\ci_build\build.py" --config RelWithDebInfo --build_dir "${{ github.workspace }}\build" --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --update --build --test --enable_generic_interface --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+          ccache --zero-stats
+
+          python.exe "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
+            --build `
+            --build_csharp `
+            --build_dir "${env:GITHUB_WORKSPACE}\build" `
+            --build_shared_lib `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --enable_generic_interface `
+            --parallel `
+            --skip_submodule_sync `
+            --test `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache
           if ($LASTEXITCODE -ne 0) {
             exit $LASTEXITCODE
           }
-          Remove-Item "${{ github.workspace }}\build\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\build\RelWithDebInfo" -Include "*.obj" -Recurse
         env:
           ALLOW_RELEASED_ONNX_OPSET_ONLY: '0'
 
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
+
       - name: Validate C# native delegates
         shell: cmd
         run: python tools\ValidateNativeDelegateAttributes.py
diff --git a/.github/workflows/windows_x64_release_vitisai_build_x64_release.yml b/.github/workflows/windows_x64_release_vitisai_build_x64_release.yml
index a3200b21cf658..c433aa9f64973 100644
--- a/.github/workflows/windows_x64_release_vitisai_build_x64_release.yml
+++ b/.github/workflows/windows_x64_release_vitisai_build_x64_release.yml
@@ -57,7 +57,7 @@ jobs:
         shell: cmd
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -83,14 +83,35 @@ jobs:
       - name: Build
         shell: pwsh
         run: |
-          python.exe "${{ github.workspace }}\tools\ci_build\build.py" --config RelWithDebInfo --build_dir "${{ github.workspace }}\build" --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --update --build --build_wheel --use_vitisai --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+          ccache --zero-stats
+
+          python.exe "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
+            --build `
+            --build_csharp `
+            --build_dir "${env:GITHUB_WORKSPACE}\build" `
+            --build_shared_lib `
+            --build_wheel `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --parallel `
+            --skip_submodule_sync `
+            --update `
+            --use_binskim_compliant_compile_flags `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache `
+            --use_vitisai
           if ($LASTEXITCODE -ne 0) {
             exit $LASTEXITCODE
           }
-          Remove-Item "${{ github.workspace }}\build\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\build\RelWithDebInfo" -Include "*.obj" -Recurse
+          ccache --show-stats -vv
         env:
           ALLOW_RELEASED_ONNX_OPSET_ONLY: '0'
 
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
+
       - name: Validate C# native delegates
         shell: cmd
         run: python tools\ValidateNativeDelegateAttributes.py
@@ -104,7 +125,7 @@ jobs:
             exit $LASTEXITCODE
           }
           Get-ChildItem -Path dist/*.whl | foreach {pip --disable-pip-version-check install --upgrade $_.fullname}
-        working-directory: "${{ github.workspace }}\\build\\RelWithDebInfo\\RelWithDebInfo"
+        working-directory: "${{ github.workspace }}\\build\\RelWithDebInfo"
     env:
       OrtPackageId: Microsoft.ML.OnnxRuntime
       OnnxRuntimeBuildDirectory: ${{ github.workspace }}\build
diff --git a/.github/workflows/windows_x64_release_xnnpack.yml b/.github/workflows/windows_x64_release_xnnpack.yml
index 1e55bd376a7e5..503dc7dad0694 100644
--- a/.github/workflows/windows_x64_release_xnnpack.yml
+++ b/.github/workflows/windows_x64_release_xnnpack.yml
@@ -36,6 +36,7 @@ jobs:
         uses: ./.github/actions/locate-vcvarsall-and-setup-env
         with:
           architecture: x64
+          cache_key: windows_x64_release_xnnpack.yml | build_x64_release_xnnpack
 
       - name: Install python modules
         shell: cmd
@@ -57,7 +58,7 @@ jobs:
         shell: cmd
         run: |
           set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo\RelWithDebInfo
+          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo
           mkdir %ORT_DOXY_SRC%
           mkdir %ORT_DOXY_OUT%
           "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
@@ -83,15 +84,37 @@ jobs:
       - name: Build and Test
         shell: pwsh
         run: |
-          python.exe "${{ github.workspace }}\tools\ci_build\build.py" --use_xnnpack --config RelWithDebInfo --build_dir "${{ github.workspace }}\build" --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --enable_onnx_tests --disable_rtti --msbuild_extra_options "IncludeMobileTargets=false" --build_nuget --use_vcpkg --use_vcpkg_ms_internal_asset_cache
+          ccache --zero-stats
+
+          python.exe "${env:GITHUB_WORKSPACE}\tools\ci_build\build.py" `
+            --build_csharp `
+            --build_dir "${env:GITHUB_WORKSPACE}\build" `
+            --build_nuget `
+            --build_shared_lib `
+            --cmake_generator "Ninja" `
+            --config RelWithDebInfo `
+            --disable_rtti `
+            --enable_onnx_tests `
+            --msbuild_extra_options "IncludeMobileTargets=false" `
+            --parallel `
+            --skip_submodule_sync `
+            --use_binskim_compliant_compile_flags `
+            --use_cache `
+            --use_vcpkg `
+            --use_vcpkg_ms_internal_asset_cache `
+            --use_xnnpack
           if ($LASTEXITCODE -ne 0) {
             exit $LASTEXITCODE
           }
-          Remove-Item "${{ github.workspace }}\build\RelWithDebInfo" -Include "*.obj" -Recurse
+          Remove-Item "${env:GITHUB_WORKSPACE}\build\RelWithDebInfo" -Include "*.obj" -Recurse
         env:
           ALLOW_RELEASED_ONNX_OPSET_ONLY: '0'
           DocUpdateNeeded: 'false'
 
+      - name: CCache Stats
+        shell: pwsh
+        run: ccache --show-stats -vv
+
       - name: Validate C# native delegates
         shell: cmd
         run: python tools\ValidateNativeDelegateAttributes.py
diff --git a/.github/workflows/windows_x86.yml b/.github/workflows/windows_x86.yml
deleted file mode 100644
index 897b4f712ffdd..0000000000000
--- a/.github/workflows/windows_x86.yml
+++ /dev/null
@@ -1,164 +0,0 @@
-name: Windows CPU CI Pipeline
-
-on:
-  push:
-    branches: [main, 'rel-*']
-  pull_request:
-    branches: [main]
-  workflow_dispatch:
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && github.ref || github.sha }}
-  cancel-in-progress: true
-
-jobs:
-  build_x86_release:
-    runs-on: [
-      "self-hosted",
-      "1ES.Pool=onnxruntime-github-vs2022-latest",
-      "JobId=windows-x86-release-${{ github.run_id }}-${{ github.run_number }}-${{ github.run_attempt }}"
-      ]
-    timeout-minutes: 300
-
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@v6
-        with:
-          submodules: false
-
-      - name: Setup Python
-        uses: actions/setup-python@v6
-        with:
-          python-version: '3.12'
-          architecture: x86 # x86 Python
-
-      - name: Locate vcvarsall and Setup Env
-        uses: ./.github/actions/locate-vcvarsall-and-setup-env
-        with:
-          architecture: x86 # x86 architecture for vcvarsall
-
-      - name: Install python modules
-        shell: cmd
-        run: python -m pip install -r "${{ github.workspace }}\tools\ci_build\github\windows\python\requirements.txt"
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v6
-        with:
-          node-version: '20.x'
-          architecture: x86 #Add architecture
-
-      - name: Setup Java
-        uses: actions/setup-java@v5
-        with:
-          distribution: 'temurin'
-          java-version: '17'
-          architecture: x86 # x86 Java
-
-      - name: API Documentation Check and generate
-        shell: cmd
-        run: |
-          set ORT_DOXY_SRC=${{ github.workspace }}
-          set ORT_DOXY_OUT=${{ github.workspace }}\build\RelWithDebInfo\RelWithDebInfo
-          mkdir %ORT_DOXY_SRC%
-          mkdir %ORT_DOXY_OUT%
-          "C:\Program Files\doxygen\bin\doxygen.exe" ${{ github.workspace }}\tools\ci_build\github\Doxyfile_csharp.cfg
-        working-directory: ${{ github.workspace }}
-
-      - name: Use .NET 8.x
-        uses: actions/setup-dotnet@v5
-        with:
-          dotnet-version: '8.x'
-        env:
-          PROCESSOR_ARCHITECTURE: x86 # x86 .NET
-
-      - name: Prefer x86 dotnet on PATH
-        shell: pwsh
-        run: |
-          $x86DotnetDir = 'C:\Program Files (x86)\dotnet'
-          $x64DotnetDir = 'C:\Program Files\dotnet'
-          $machinePath = [Environment]::GetEnvironmentVariable('Path', 'Machine')
-          Write-Host "Machine PATH: $machinePath"
-
-          $pathEntries = @($env:PATH -split ';' | Where-Object { $_ })
-          $reorderedPathEntries = @(
-            $x86DotnetDir
-            $pathEntries | Where-Object { $_ -ne $x86DotnetDir }
-          )
-
-          $env:PATH = ($reorderedPathEntries -join ';')
-
-          # Only add the x86 dotnet directory to GITHUB_PATH if it is not already on PATH (after normalization)
-          $normalizedX86DotnetDir = $x86DotnetDir.TrimEnd('\')
-          $normalizedPathEntries = $pathEntries | ForEach-Object { $_.TrimEnd('\') }
-          if (-not ($normalizedPathEntries -contains $normalizedX86DotnetDir)) {
-            Add-Content -Path $env:GITHUB_PATH -Value $x86DotnetDir
-          }
-          $dotnetPaths = @(Get-Command dotnet -All | Select-Object -ExpandProperty Source -Unique)
-          Write-Host 'Resolved dotnet executables:'
-          $dotnetPaths | ForEach-Object { Write-Host $_ }
-
-          $x86DotnetExe = "$x86DotnetDir\dotnet.exe"
-          $x64DotnetExe = "$x64DotnetDir\dotnet.exe"
-          $x86Index = $dotnetPaths.IndexOf($x86DotnetExe)
-          $x64Index = $dotnetPaths.IndexOf($x64DotnetExe)
-
-          if ($x86Index -lt 0) {
-            throw "Expected x86 dotnet executable was not found: $x86DotnetExe"
-          }
-
-          if ($x64Index -ge 0 -and $x86Index -gt $x64Index) {
-            throw "x86 dotnet must appear before x64 dotnet on PATH. Found $x86DotnetExe after $x64DotnetExe."
-          }
-
-      - name: Use Nuget 6.x
-        uses: nuget/setup-nuget@v2
-        with:
-          nuget-version: '6.x'
-
-      - name: NuGet restore
-        shell: cmd
-        run: |
-          nuget restore ${{ github.workspace }}\packages.config -PackagesDirectory ${{ github.workspace }}\build\RelWithDebInfo -ConfigFile ${{ github.workspace }}\NuGet.config
-
-      - name: Build and Test
-        shell: pwsh
-        run: |
-          python.exe "${{ github.workspace }}\tools\ci_build\build.py" --config RelWithDebInfo --build_dir "${{ github.workspace }}\build" --skip_submodule_sync --build_csharp --parallel --use_binskim_compliant_compile_flags --cmake_generator "Visual Studio 17 2022" --build_shared_lib --enable_onnx_tests --build_wheel --msbuild_extra_options "IncludeMobileTargets=false" --build_nuget --compile_no_warning_as_error --use_vcpkg --use_vcpkg_ms_internal_asset_cache
-          if ($LASTEXITCODE -ne 0) {
-            exit $LASTEXITCODE
-          }
-          Remove-Item "${{ github.workspace }}\build\RelWithDebInfo" -Include "*.obj" -Recurse
-        env:
-          ALLOW_RELEASED_ONNX_OPSET_ONLY: '0'
-          DocUpdateNeeded: 'false'
-
-      - name: Validate C# native delegates
-        shell: cmd
-        run: python tools\ValidateNativeDelegateAttributes.py
-        working-directory: ${{ github.workspace }}\\csharp
-
-      - name: Install onnxruntime wheel
-        shell: pwsh
-        run: |
-          python -m pip uninstall -y onnxruntime onnxruntime-gpu onnxruntime-training onnxruntime-directml -qq
-          Get-ChildItem -Path dist/*.whl | foreach {pip --disable-pip-version-check install --upgrade $_.fullname}
-        working-directory: "${{ github.workspace }}\\build\\RelWithDebInfo\\RelWithDebInfo"
-
-      - name: Publish OperatorKernels.md (Conditional)
-        uses: actions/upload-artifact@v6
-        if: failure() && env.DocUpdateNeeded == 'true'
-        with:
-          name: OperatorKernels.md
-          path: ${{ github.workspace }}/docs/OperatorKernels.md
-
-      - name: Publish ContribOperators.md (Conditional)
-        uses: actions/upload-artifact@v6
-        if: failure() && env.DocUpdateNeeded == 'true'
-        with:
-          name: ContribOperators.md
-          path: ${{ github.workspace }}/docs/ContribOperators.md
-
-    env:
-      OrtPackageId: Microsoft.ML.OnnxRuntime
-      OnnxRuntimeBuildDirectory: ${{ github.workspace }}\build
-      DOTNET_SKIP_FIRST_TIME_EXPERIENCE: 'true'
diff --git a/cmake/onnxruntime.cmake b/cmake/onnxruntime.cmake
index ff06ae91bead9..fc285f528a413 100644
--- a/cmake/onnxruntime.cmake
+++ b/cmake/onnxruntime.cmake
@@ -268,6 +268,11 @@ if (onnxruntime_USE_EXTENSIONS)
   )
 endif()
 
+if(WIN32)
+  # required if we're delay loading anything
+  list(APPEND onnxruntime_EXTERNAL_LIBRARIES delayimp.lib)
+endif()
+
 # If you are linking a new library, please add it to the list onnxruntime_INTERNAL_LIBRARIES or onnxruntime_EXTERNAL_LIBRARIES,
 # Please do not add a library directly to the target_link_libraries command
 if (onnxruntime_BUILD_SHARED_LIB)
@@ -285,6 +290,7 @@ endif()
 if(WIN32)
   target_link_options(onnxruntime PRIVATE ${onnxruntime_DELAYLOAD_FLAGS})
 endif()
+
 #See: https://cmake.org/cmake/help/latest/prop_tgt/SOVERSION.html
 if(NOT APPLE AND NOT WIN32)
   if(CMAKE_SYSTEM_NAME MATCHES "AIX")
diff --git a/csharp/OnnxRuntime.CSharp.proj b/csharp/OnnxRuntime.CSharp.proj
index 6779fd60bcd0a..a3180c7d0e4a4 100644
--- a/csharp/OnnxRuntime.CSharp.proj
+++ b/csharp/OnnxRuntime.CSharp.proj
@@ -43,7 +43,7 @@ CMake creates a target to this project
     <!--internal build related properties for Windows -->
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">..\build\Windows</OnnxRuntimeBuildDirectory>
     <OnnxRuntimePackagesDirectory Condition="'$(OnnxRuntimePackagesDirectory)'==''">$(OnnxRuntimeBuildDirectory)\packages</OnnxRuntimePackagesDirectory>
-    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>
+    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
     <PythonExe>python</PythonExe>
   </PropertyGroup>
 
diff --git a/csharp/sample/InferenceSample/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp.csproj b/csharp/sample/InferenceSample/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp.csproj
index 2f81e52d0464d..b532583386c0b 100644
--- a/csharp/sample/InferenceSample/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp.csproj
+++ b/csharp/sample/InferenceSample/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp/Microsoft.ML.OnnxRuntime.InferenceSample.NetCoreApp.csproj
@@ -6,9 +6,9 @@
     <Platforms>AnyCPU;x86</Platforms>
     <OnnxRuntimeCsharpRoot>$(ProjectDir)..\..\..</OnnxRuntimeCsharpRoot>
     <Configurations>Debug;Release;RelWithDebInfo</Configurations>
-    <IsLinuxBuild Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::Linux)))' == 'true'">true</IsLinuxBuild> 
-    <IsWindowsBuild Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::Windows)))' == 'true'">true</IsWindowsBuild> 
-    <IsMacOSBuild Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::OSX)))' == 'true'">true</IsMacOSBuild> 
+    <IsLinuxBuild Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::Linux)))' == 'true'">true</IsLinuxBuild>
+    <IsWindowsBuild Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::Windows)))' == 'true'">true</IsWindowsBuild>
+    <IsMacOSBuild Condition="'$([System.Runtime.InteropServices.RuntimeInformation]::IsOSPlatform($([System.Runtime.InteropServices.OSPlatform]::OSX)))' == 'true'">true</IsMacOSBuild>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(IsLinuxBuild)'=='true'">
@@ -20,13 +20,13 @@
   <PropertyGroup Condition="'$(IsWindowsBuild)'=='true'">
     <!--internal build related properties for Windows -->
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">$(OnnxRuntimeCsharpRoot)\..\build\Windows</OnnxRuntimeBuildDirectory>
-    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>    
+    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(IsMacOSBuild)'=='true'">
     <!--internal build related properties for OSX -->
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">$(OnnxRuntimeCsharpRoot)\..\build\MacOS</OnnxRuntimeBuildDirectory>
-    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>    
+    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
   </PropertyGroup>
 
   <ItemGroup>
diff --git a/csharp/src/Microsoft.AI.MachineLearning.Interop/Microsoft.AI.MachineLearning.Interop.csproj b/csharp/src/Microsoft.AI.MachineLearning.Interop/Microsoft.AI.MachineLearning.Interop.csproj
index 5e727372d0276..db647abe6f5bf 100644
--- a/csharp/src/Microsoft.AI.MachineLearning.Interop/Microsoft.AI.MachineLearning.Interop.csproj
+++ b/csharp/src/Microsoft.AI.MachineLearning.Interop/Microsoft.AI.MachineLearning.Interop.csproj
@@ -9,7 +9,7 @@
     <Configuration Condition=" '$(Configuration)' == '' ">Debug</Configuration>
 
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">..\..\..\build\Windows</OnnxRuntimeBuildDirectory>
-    <BuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</BuildOutputDir>
+    <BuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</BuildOutputDir>
     <WindowsAIInteropOutputDir>$(BuildOutputDir)\Microsoft.AI.MachineLearning.Interop</WindowsAIInteropOutputDir>
     <OutputPath>$(WindowsAIInteropOutputDir)</OutputPath>
   </PropertyGroup>
diff --git a/csharp/src/Microsoft.ML.OnnxRuntime/Microsoft.ML.OnnxRuntime.csproj b/csharp/src/Microsoft.ML.OnnxRuntime/Microsoft.ML.OnnxRuntime.csproj
index 193be042ae420..e22b7fe1c62fb 100644
--- a/csharp/src/Microsoft.ML.OnnxRuntime/Microsoft.ML.OnnxRuntime.csproj
+++ b/csharp/src/Microsoft.ML.OnnxRuntime/Microsoft.ML.OnnxRuntime.csproj
@@ -136,7 +136,7 @@
     Properties that are used when creating the managed package using the Pack target.
   -->
   <PropertyGroup Condition="'$(IsWindowsBuild)'=='true'">
-      <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>
+      <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
   </PropertyGroup>
   <PropertyGroup Condition="'$(IsLinuxBuild)'=='true'">
       <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
diff --git a/csharp/test/Microsoft.ML.OnnxRuntime.EndToEndTests/Microsoft.ML.OnnxRuntime.EndToEndTests.csproj b/csharp/test/Microsoft.ML.OnnxRuntime.EndToEndTests/Microsoft.ML.OnnxRuntime.EndToEndTests.csproj
index 4da9b5ffae3e4..e3e9b9897da2c 100644
--- a/csharp/test/Microsoft.ML.OnnxRuntime.EndToEndTests/Microsoft.ML.OnnxRuntime.EndToEndTests.csproj
+++ b/csharp/test/Microsoft.ML.OnnxRuntime.EndToEndTests/Microsoft.ML.OnnxRuntime.EndToEndTests.csproj
@@ -27,7 +27,7 @@
   <PropertyGroup Condition="'$(IsWindowsBuild)'=='true'">
     <!--internal build related properties for Windows -->
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">$(OnnxRuntimeCsharpRoot)\..\build\Windows</OnnxRuntimeBuildDirectory>
-    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>
+    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(IsMacOSBuild)'=='true'">
diff --git a/csharp/test/Microsoft.ML.OnnxRuntime.Tests.Common/NativeLibraryInclude.props b/csharp/test/Microsoft.ML.OnnxRuntime.Tests.Common/NativeLibraryInclude.props
index 3daab21dbcbac..d9f4da4d7081e 100644
--- a/csharp/test/Microsoft.ML.OnnxRuntime.Tests.Common/NativeLibraryInclude.props
+++ b/csharp/test/Microsoft.ML.OnnxRuntime.Tests.Common/NativeLibraryInclude.props
@@ -66,7 +66,7 @@ test project only targets net8 and netstandard2.0.
         <When Condition="'$(UsePrebuiltNativePackage)' != 'true'">
             <PropertyGroup Condition="'$(IsWindowsBuild)'=='true' OR '$(IsWindowsTarget)'=='true'">
                 <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">$(OnnxRuntimeRoot)\build\Windows</OnnxRuntimeBuildDirectory>
-                <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>
+                <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
             </PropertyGroup>
 
             <PropertyGroup Condition="'$(IsLinuxBuild)'=='true'">
diff --git a/csharp/test/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp.csproj b/csharp/test/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp.csproj
index 54b9925710296..8497b92fe1fc6 100644
--- a/csharp/test/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp.csproj
+++ b/csharp/test/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp.csproj
@@ -45,7 +45,7 @@
   <PropertyGroup Condition="'$(IsWindowsBuild)'=='true'">
     <!--internal build related properties for Windows -->
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">$(OnnxRuntimeRoot)\build\Windows</OnnxRuntimeBuildDirectory>
-    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>
+    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(IsMacOSBuild)'=='true'">
diff --git a/csharp/tools/Microsoft.ML.OnnxRuntime.PerfTool/Microsoft.ML.OnnxRuntime.PerfTool.csproj b/csharp/tools/Microsoft.ML.OnnxRuntime.PerfTool/Microsoft.ML.OnnxRuntime.PerfTool.csproj
index f3e070044ad6f..582662e6cedbc 100644
--- a/csharp/tools/Microsoft.ML.OnnxRuntime.PerfTool/Microsoft.ML.OnnxRuntime.PerfTool.csproj
+++ b/csharp/tools/Microsoft.ML.OnnxRuntime.PerfTool/Microsoft.ML.OnnxRuntime.PerfTool.csproj
@@ -26,7 +26,7 @@
   <PropertyGroup Condition="'$(IsWindowsBuild)'=='true'">
     <!--internal build related properties for Windows -->
     <OnnxRuntimeBuildDirectory Condition="'$(OnnxRuntimeBuildDirectory)'==''">$(OnnxRuntimeCsharpRoot)\..\build\Windows</OnnxRuntimeBuildDirectory>
-    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)\$(Configuration)</NativeBuildOutputDir>
+    <NativeBuildOutputDir>$(OnnxRuntimeBuildDirectory)\$(Configuration)</NativeBuildOutputDir>
     <ProtocDirectory Condition="'$(ProtocDirectory)'==''">$(OnnxRuntimeBuildDirectory)\$(Configuration)\external\protobuf\cmake\$(Configuration)</ProtocDirectory>
     <ProtocExe>$(ProtocDirectory)\protoc.exe</ProtocExe>
   </PropertyGroup>
diff --git a/onnxruntime/python/tools/quantization/README.md b/onnxruntime/python/tools/quantization/README.md
index 16d6f3f2fd346..50c229566b4e2 100644
--- a/onnxruntime/python/tools/quantization/README.md
+++ b/onnxruntime/python/tools/quantization/README.md
@@ -57,7 +57,7 @@ Install the python tools built in onnxruntime
 cd work_dir
 python -m venv ort_env
 ort_env\Scripts\activate
-python -m pip install <path-to-built-folder>\RelWithDebInfo\RelWithDebInfo\dist\<name-of-the-wheel>.whl
+python -m pip install <path-to-built-folder>\RelWithDebInfo\dist\<name-of-the-wheel>.whl
 
 # The following command yields model_quant.onnx under the same directory "resnet18-v1-7"
 python -m onnxruntime.quantization.static_quantize_runner -i resnet18-v1-7\model.onnx -o resnet18-v1-7\model_quant.onnx
diff --git a/onnxruntime/test/platform/windows/logging/HowToValidateEtwSinkOutput.md b/onnxruntime/test/platform/windows/logging/HowToValidateEtwSinkOutput.md
index 309b474c016c9..d496636d0c13b 100644
--- a/onnxruntime/test/platform/windows/logging/HowToValidateEtwSinkOutput.md
+++ b/onnxruntime/test/platform/windows/logging/HowToValidateEtwSinkOutput.md
@@ -29,7 +29,7 @@ Run the ETW sink unit tests
  * Run the tests in etw_sink_test.cc and then stop WPR.
   * This can be done by executing the specific tests in Visual Studio,
   * or by running the exe with all the tests from the platform library
-    * Assuming debug build on Windows run `<path to repo>\build\Windows\Debug\Debug\onnxruntime_test_framework.exe`
+    * Assuming debug build on Windows run `<path to repo>\build\Windows\Debug\onnxruntime_test_framework.exe`
 
 Stop the ETW tracing
     `<path to repo>\onnxruntime\test\platform\windows\logging> wpr -stop TraceCaptureFile.etl EtwSinkTest`
diff --git a/tools/ci_build/build.py b/tools/ci_build/build.py
index 7eeb9cb59d0b2..04ea1f518ec6e 100644
--- a/tools/ci_build/build.py
+++ b/tools/ci_build/build.py
@@ -722,7 +722,12 @@ def generate_build_tree(
     if args.use_cuda:
         nvcc_threads = number_of_nvcc_threads(args)
         cmake_args.append("-Donnxruntime_NVCC_THREADS=" + str(nvcc_threads))
-        cmake_args.append(f"-DCMAKE_CUDA_COMPILER={cuda_home}/bin/nvcc")
+        if is_windows():
+            # be explicit and use the entire filename name
+            cmake_args.append(f"-DCMAKE_CUDA_COMPILER={cuda_home}/bin/nvcc.exe")
+        else:
+            cmake_args.append(f"-DCMAKE_CUDA_COMPILER={cuda_home}/bin/nvcc")
+
         add_default_definition(cmake_extra_defines, "onnxruntime_USE_CUDA", "ON")
         if args.cuda_version:
             add_default_definition(cmake_extra_defines, "onnxruntime_CUDA_VERSION", args.cuda_version)
@@ -1745,7 +1750,7 @@ def run_onnxruntime_tests(args, source_dir, ctest_path, build_dir, configs):
             if is_reduced_ops_build(args) or args.minimal_build is not None:
                 return
 
-            if is_windows():
+            if is_windows() and args.cmake_generator != "Ninja":
                 cwd = os.path.join(cwd, config)
 
             if not args.skip_pip_install and args.enable_transformers_tool_test and not args.disable_contrib_ops:
diff --git a/tools/ci_build/github/azure-pipelines/dml-nuget-packaging.yml b/tools/ci_build/github/azure-pipelines/dml-nuget-packaging.yml
index 3cf28655c36e7..17d0151992fb7 100644
--- a/tools/ci_build/github/azure-pipelines/dml-nuget-packaging.yml
+++ b/tools/ci_build/github/azure-pipelines/dml-nuget-packaging.yml
@@ -80,10 +80,10 @@ extends:
           if errorlevel 1 exit /b 1
           copy $(Build.SourcesDirectory)\csharp\src\Microsoft.ML.OnnxRuntime\bin\RelWithDebInfo\*.nupkg $(Build.ArtifactStagingDirectory)
           if errorlevel 1 exit /b 1
-          powershell -ExecutionPolicy Bypass -File $(Build.SourcesDirectory)\tools\ci_build\github\windows\select_dml_package.ps1 -SourceDir "$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo" -IsReleaseBuild "${{ parameters.IsReleaseBuild }}" -Action copy -DestinationDir "$(Build.ArtifactStagingDirectory)"
+          powershell -ExecutionPolicy Bypass -File $(Build.SourcesDirectory)\tools\ci_build\github\windows\select_dml_package.ps1 -SourceDir "$(Build.BinariesDirectory)\RelWithDebInfo" -IsReleaseBuild "${{ parameters.IsReleaseBuild }}" -Action copy -DestinationDir "$(Build.ArtifactStagingDirectory)"
           if errorlevel 1 exit /b 1
           mkdir $(Build.ArtifactStagingDirectory)\testdata
-          copy $(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\custom_op_library.* $(Build.ArtifactStagingDirectory)\testdata
+          copy $(Build.BinariesDirectory)\RelWithDebInfo\custom_op_library.* $(Build.ArtifactStagingDirectory)\testdata
 
     - template: nuget/templates/dml-vs-2022.yml
       parameters:
@@ -101,12 +101,12 @@ extends:
           python -m pip install setuptools
           msbuild $(Build.SourcesDirectory)\csharp\OnnxRuntime.CSharp.proj /p:Configuration=RelWithDebInfo /p:TargetArchitecture=arm64 /t:CreatePackage /p:OrtPackageId=Microsoft.ML.OnnxRuntime.DirectML /p:IsReleaseBuild=${{ parameters.IsReleaseBuild }}
           if errorlevel 1 exit /b 1
-          powershell -ExecutionPolicy Bypass -File $(Build.SourcesDirectory)\tools\ci_build\github\windows\select_dml_package.ps1 -SourceDir "$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo" -IsReleaseBuild "${{ parameters.IsReleaseBuild }}" -Action rename -NewName "win-dml-arm64.zip"
+          powershell -ExecutionPolicy Bypass -File $(Build.SourcesDirectory)\tools\ci_build\github\windows\select_dml_package.ps1 -SourceDir "$(Build.BinariesDirectory)\RelWithDebInfo" -IsReleaseBuild "${{ parameters.IsReleaseBuild }}" -Action rename -NewName "win-dml-arm64.zip"
           if errorlevel 1 exit /b 1
-          copy $(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\win-dml-arm64.zip $(Build.ArtifactStagingDirectory)
+          copy $(Build.BinariesDirectory)\RelWithDebInfo\win-dml-arm64.zip $(Build.ArtifactStagingDirectory)
           if errorlevel 1 exit /b 1
           mkdir $(Build.ArtifactStagingDirectory)\testdata
-          copy $(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\custom_op_library.* $(Build.ArtifactStagingDirectory)\testdata
+          copy $(Build.BinariesDirectory)\RelWithDebInfo\custom_op_library.* $(Build.ArtifactStagingDirectory)\testdata
 
     - template: stages/nuget_dml_packaging_stage.yml
       parameters:
diff --git a/tools/ci_build/github/azure-pipelines/nuget/templates/dml-vs-2022.yml b/tools/ci_build/github/azure-pipelines/nuget/templates/dml-vs-2022.yml
index 2548eebeb9d42..6aea5c372d40b 100644
--- a/tools/ci_build/github/azure-pipelines/nuget/templates/dml-vs-2022.yml
+++ b/tools/ci_build/github/azure-pipelines/nuget/templates/dml-vs-2022.yml
@@ -138,5 +138,5 @@ stages:
               symbolProject: 'ONNX Runtime'
               subscription: 'OnnxrunTimeCodeSign_20240611'
               searchPattern: |
-                $(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\onnxruntime.pdb
-                $(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\onnxruntime_providers_*.pdb
+                $(Build.BinariesDirectory)\RelWithDebInfo\onnxruntime.pdb
+                $(Build.BinariesDirectory)\RelWithDebInfo\onnxruntime_providers_*.pdb
diff --git a/tools/ci_build/github/azure-pipelines/stages/nuget-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/nuget-cuda-packaging-stage.yml
index a0f578e22f910..ca47d12c0a659 100644
--- a/tools/ci_build/github/azure-pipelines/stages/nuget-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/nuget-cuda-packaging-stage.yml
@@ -149,19 +149,19 @@ stages:
       displayName: 'Add TensorRT header file to the native nuGet package'
       inputs:
         filename: $(Build.SourcesDirectory)\tools\ci_build\github\windows\bundle_nuget_with_native_headers.bat
-        workingFolder: $(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo
+        workingFolder: $(Build.BinariesDirectory)\RelWithDebInfo
 
     - task: CopyFiles@2
       displayName: 'Copy nuget packages to: $(Build.ArtifactStagingDirectory)'
       inputs:
-        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo'
         Contents: '*.snupkg'
         TargetFolder: '$(Build.ArtifactStagingDirectory)'
 
     - task: CopyFiles@2
       displayName: 'Copy nuget packages to: $(Build.ArtifactStagingDirectory)'
       inputs:
-        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo'
         Contents: '*.nupkg'
         TargetFolder: '$(Build.ArtifactStagingDirectory)'
 
diff --git a/tools/ci_build/github/azure-pipelines/templates/c-api-cpu.yml b/tools/ci_build/github/azure-pipelines/templates/c-api-cpu.yml
index 448dbafcaaaac..5a20564606d47 100644
--- a/tools/ci_build/github/azure-pipelines/templates/c-api-cpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/c-api-cpu.yml
@@ -371,14 +371,14 @@ stages:
     - task: CopyFiles@2
       displayName: 'Copy native nuget package to: $(Build.ArtifactStagingDirectory)'
       inputs:
-        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo'
         Contents: '*.nupkg'
         TargetFolder: '$(Build.ArtifactStagingDirectory)'
 
     - task: CopyFiles@2
       displayName: 'Copy native nuget symbols package to: $(Build.ArtifactStagingDirectory)'
       inputs:
-        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+        SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo'
         Contents: '*.snupkg'
         TargetFolder: '$(Build.ArtifactStagingDirectory)'
 
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-arm64-qnn.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-arm64-qnn.yml
index 50229d91fbf03..fc1437e2c1699 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-arm64-qnn.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-arm64-qnn.yml
@@ -111,7 +111,7 @@ jobs:
   # Esrp signing
   - template: win-esrp-dll.yml
     parameters:
-      FolderPath: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\onnxruntime\capi'
+      FolderPath: '$(Build.BinariesDirectory)\RelWithDebInfo\onnxruntime\capi'
       DisplayName: 'ESRP - Sign Native dlls'
       DoEsrp: true
       Pattern: '*.pyd'
@@ -121,11 +121,11 @@ jobs:
     inputs:
       scriptPath: '$(Build.SourcesDirectory)\setup.py'
       arguments: 'bdist_wheel $(NightlyBuildOption) --wheel_name_suffix=qnn'
-      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo'
 
   - task: CopyFiles@2
     displayName: 'Copy Python Wheel to: $(Build.ArtifactStagingDirectory)'
     inputs:
-      SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\dist'
+      SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\dist'
       Contents: '*.whl'
       TargetFolder: '$(Build.ArtifactStagingDirectory)'
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-arm64ec-qnn.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-arm64ec-qnn.yml
index c350c56758a3b..97f8b4517f3ce 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-arm64ec-qnn.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-arm64ec-qnn.yml
@@ -91,7 +91,7 @@ jobs:
   # Esrp signing
   - template: win-esrp-dll.yml
     parameters:
-      FolderPath: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\onnxruntime\capi'
+      FolderPath: '$(Build.BinariesDirectory)\RelWithDebInfo\onnxruntime\capi'
       DisplayName: 'ESRP - Sign Native dlls'
       DoEsrp: true
       Pattern: '*.pyd'
@@ -101,12 +101,12 @@ jobs:
     inputs:
       scriptPath: '$(Build.SourcesDirectory)\setup.py'
       arguments: 'bdist_wheel $(NightlyBuildOption) --wheel_name_suffix=qnn'
-      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo'
 
   - task: CopyFiles@2
     displayName: 'Copy Python Wheel to: $(Build.ArtifactStagingDirectory)'
     inputs:
-      SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\dist'
+      SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\dist'
       Contents: '*.whl'
       TargetFolder: '$(Build.ArtifactStagingDirectory)'
 
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-x64-qnn.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-x64-qnn.yml
index 0363f2a5ffa0d..6cb1d1c9a2ab3 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-x64-qnn.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-x64-qnn.yml
@@ -94,7 +94,7 @@ jobs:
       # Esrp signing
       - template: win-esrp-dll.yml
         parameters:
-          FolderPath: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\onnxruntime\capi'
+          FolderPath: '$(Build.BinariesDirectory)\RelWithDebInfo\onnxruntime\capi'
           DisplayName: 'ESRP - Sign Native dlls'
           DoEsrp: true
           Pattern: '*.pyd'
@@ -104,12 +104,12 @@ jobs:
         inputs:
           scriptPath: '$(Build.SourcesDirectory)\setup.py'
           arguments: 'bdist_wheel $(NightlyBuildOption) --wheel_name_suffix=qnn'
-          workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+          workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo'
 
       - task: CopyFiles@2
         displayName: 'Copy Python Wheel to: $(Build.ArtifactStagingDirectory)'
         inputs:
-          SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo\dist'
+          SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\dist'
           Contents: '*.whl'
           TargetFolder: '$(Build.ArtifactStagingDirectory)'
 
diff --git a/tools/ci_build/github/azure-pipelines/templates/win-ci.yml b/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
index 8a5584c111525..52e9253926d02 100644
--- a/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
@@ -279,6 +279,6 @@ stages:
         displayName: 'Copy custom_op_library to: $(Build.ArtifactStagingDirectory)'
         condition: and(succeeded(), eq('${{ parameters.packageName}}', 'x64'))
         inputs:
-          SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+          SourceFolder: '$(Build.BinariesDirectory)\RelWithDebInfo'
           Contents: 'custom_op_library.dll'
           TargetFolder: '$(Build.ArtifactStagingDirectory)/testdata'
diff --git a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-cuda-minimal-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-cuda-minimal-ci-pipeline.yml
index 6aba9d3d9ac63..ac38accada797 100644
--- a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-cuda-minimal-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-cuda-minimal-ci-pipeline.yml
@@ -88,4 +88,4 @@ jobs:
     inputs:
       scriptPath: '$(Build.SourcesDirectory)\setup.py'
       arguments: 'bdist_wheel'
-      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo\RelWithDebInfo'
+      workingDirectory: '$(Build.BinariesDirectory)\RelWithDebInfo'
diff --git a/tools/ci_build/github/windows/extract_nuget_files.ps1 b/tools/ci_build/github/windows/extract_nuget_files.ps1
index 20d6c1f2b63a5..e551b288df079 100644
--- a/tools/ci_build/github/windows/extract_nuget_files.ps1
+++ b/tools/ci_build/github/windows/extract_nuget_files.ps1
@@ -4,7 +4,7 @@
 # This file is used by Zip-Nuget-Java Packaging Pipeline
 
 # Define the directory for NuGet artifacts.
-$nuget_artifacts_dir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\RelWithDebInfo\nuget-artifacts"
+$nuget_artifacts_dir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\nuget-artifacts"
 # Create the directory if it doesn't exist.
 New-Item -Path $nuget_artifacts_dir -ItemType directory -ErrorAction SilentlyContinue
 
@@ -92,7 +92,7 @@ if ($is_training_pipeline) {
 }
 
 # Copy onnxruntime and protoc binaries required by tests.
-$destinationDir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\RelWithDebInfo"
+$destinationDir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo"
 if ($is_training_pipeline) {
     Copy-Item -Path "$nuget_artifacts_dir\onnxruntime-training-win-x64-*\lib\*" -Destination $destinationDir -Recurse
 }
@@ -116,4 +116,4 @@ Get-ChildItem -Directory -Path "$nuget_artifacts_dir\onnxruntime-*" | ForEach-Ob
 
 # List the final artifacts.
 Write-Output "Post-copy artifacts:"
-Get-ChildItem -Recurse $nuget_artifacts_dir
\ No newline at end of file
+Get-ChildItem -Recurse $nuget_artifacts_dir
diff --git a/tools/ci_build/github/windows/extract_nuget_files_gpu.ps1 b/tools/ci_build/github/windows/extract_nuget_files_gpu.ps1
index 29946dcb73f8a..c3e473ae18e3f 100644
--- a/tools/ci_build/github/windows/extract_nuget_files_gpu.ps1
+++ b/tools/ci_build/github/windows/extract_nuget_files_gpu.ps1
@@ -3,7 +3,7 @@
 
 # This file is used by Zip-Nuget-Java Packaging Pipeline
 # Define the directory for NuGet artifacts.
-$nuget_artifacts_dir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\RelWithDebInfo\nuget-artifacts"
+$nuget_artifacts_dir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\nuget-artifacts"
 # Create the directory if it doesn't exist.
 New-Item -Path $nuget_artifacts_dir -ItemType directory -ErrorAction SilentlyContinue
 
@@ -48,7 +48,7 @@ Foreach-Object {
 New-Item -Path "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\_deps\protobuf-build\RelWithDebInfo" -ItemType directory -ErrorAction SilentlyContinue
 
 # Copy CUDA libraries.
-Copy-Item -Path "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\RelWithDebInfo\nuget-artifacts\onnxruntime-win-x64-cuda-*\lib\*" -Destination "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\RelWithDebInfo"
+Copy-Item -Path "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\nuget-artifacts\onnxruntime-win-x64-cuda-*\lib\*" -Destination "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo"
 
 # Install protoc via dotnet.
 $protocInstallDir = "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\_deps\protobuf-build"
@@ -69,7 +69,7 @@ else {
 }
 
 # Rename onnxruntime directories to a generic format.
-$ort_dirs = Get-ChildItem -Path "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\RelWithDebInfo\nuget-artifacts\onnxruntime-*" -Directory
+$ort_dirs = Get-ChildItem -Path "$Env:BUILD_BINARIESDIRECTORY\RelWithDebInfo\nuget-artifacts\onnxruntime-*" -Directory
 foreach ($ort_dir in $ort_dirs) {
     $dirname = Split-Path -Path $ort_dir -Leaf
     $lastHyphenIndex = $dirname.LastIndexOf('-')