diff --git a/ci/test_cudf_polars_polars_tests.sh b/ci/test_cudf_polars_polars_tests.sh
index f8365c1dc5b..72b683fc459 100755
--- a/ci/test_cudf_polars_polars_tests.sh
+++ b/ci/test_cudf_polars_polars_tests.sh
@@ -13,14 +13,24 @@ CUDF_POLARS_WHEELHOUSE=$(RAPIDS_PY_WHEEL_NAME="cudf_polars_${RAPIDS_PY_CUDA_SUFF
 LIBCUDF_WHEELHOUSE=$(RAPIDS_PY_WHEEL_NAME="libcudf_${RAPIDS_PY_CUDA_SUFFIX}" rapids-download-wheels-from-github cpp)
 PYLIBCUDF_WHEELHOUSE=$(rapids-download-from-github "$(rapids-package-name "wheel_python" pylibcudf --stable --cuda "$RAPIDS_CUDA_VERSION")")
 
+# generate constraints (possibly pinning to oldest support versions of dependencies)
+rapids-generate-pip-constraints py_test_cudf_polars "${PIP_CONSTRAINT}"
+
 rapids-logger "Install libcudf, pylibcudf and cudf_polars"
+
+# notes:
+#
+#   * echo to expand wildcard before adding `[test]` requires for pip
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
+#
 rapids-pip-retry install \
     -v \
+    --constraint "${PIP_CONSTRAINT}" \
     "$(echo "${CUDF_POLARS_WHEELHOUSE}"/cudf_polars_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)[test]" \
     "$(echo "${LIBCUDF_WHEELHOUSE}"/libcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)" \
     "$(echo "${PYLIBCUDF_WHEELHOUSE}"/pylibcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)"
 
-
 TAG=$(python -c 'import polars; print(f"py-{polars.__version__}")')
 rapids-logger "Clone polars to ${TAG}"
 git clone https://github.com/pola-rs/polars.git --branch "${TAG}" --depth 1
@@ -55,7 +65,17 @@ sed -i 's/^pandas-stubs/pandas-stubs<3/' polars/py-polars/requirements-dev.txt
 # existing dependency to rewrite.
 echo "pyparsing>=3.0.0,<3.3.0" >> polars/py-polars/requirements-dev.txt
 
-rapids-pip-retry install -r polars/py-polars/requirements-dev.txt -r polars/py-polars/requirements-ci.txt
+# notes:
+#
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
+#
+rapids-pip-retry install \
+    -v \
+    --prefer-binary \
+    --constraint "${PIP_CONSTRAINT}" \
+    -r polars/py-polars/requirements-dev.txt \
+    -r polars/py-polars/requirements-ci.txt
 
 # shellcheck disable=SC2317
 function set_exitcode()
diff --git a/ci/test_cudf_polars_with_rapidsmpf.sh b/ci/test_cudf_polars_with_rapidsmpf.sh
index 6e9ad45afe3..5072373e536 100755
--- a/ci/test_cudf_polars_with_rapidsmpf.sh
+++ b/ci/test_cudf_polars_with_rapidsmpf.sh
@@ -15,8 +15,18 @@ PYLIBCUDF_WHEELHOUSE=$(rapids-download-from-github "$(rapids-package-name "wheel
 
 rapids-logger "Installing cudf_polars and its dependencies (including rapidsmpf)"
 
+# generate constraints (possibly pinning to oldest support versions of dependencies)
+rapids-generate-pip-constraints py_test_cudf_polars "${PIP_CONSTRAINT}"
+
+# notes:
+#
+#   * echo to expand wildcard before adding `[test]` requires for pip
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
+#
 rapids-pip-retry install \
     -v \
+    --prefer-binary \
     --constraint "${PIP_CONSTRAINT}" \
     "$(echo "${CUDF_POLARS_WHEELHOUSE}"/cudf_polars_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)[test,experimental,rapidsmpf]" \
     "$(echo "${LIBCUDF_WHEELHOUSE}"/libcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)" \
diff --git a/ci/test_wheel_cudf.sh b/ci/test_wheel_cudf.sh
index a890c11f88d..3c8e3b84a1f 100755
--- a/ci/test_wheel_cudf.sh
+++ b/ci/test_wheel_cudf.sh
@@ -16,7 +16,7 @@ PYLIBCUDF_WHEELHOUSE=$(rapids-download-from-github "$(rapids-package-name "wheel
 rapids-logger "Install pylibcudf and its basic dependencies in a virtual environment"
 
 # generate constraints (possibly pinning to oldest support versions of dependencies)
-rapids-generate-pip-constraints py_test_cudf ./constraints.txt
+rapids-generate-pip-constraints py_test_cudf "${PIP_CONSTRAINT}"
 
 RESULTS_DIR=${RAPIDS_TESTS_DIR:-"$(mktemp -d)"}
 RAPIDS_TESTS_DIR=${RAPIDS_TESTS_DIR:-"${RESULTS_DIR}/test-results"}/
@@ -25,9 +25,16 @@ mkdir -p "${RAPIDS_TESTS_DIR}"
 # To test pylibcudf without its optional dependencies, we create a virtual environment
 python -m venv env
 . env/bin/activate
+
+# notes:
+#
+#   * echo to expand wildcard before adding `[test]` requires for pip
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
+#
 rapids-pip-retry install \
     -v \
-    --constraint ./constraints.txt \
+    --prefer-binary \
     --constraint "${PIP_CONSTRAINT}" \
     "$(echo "${LIBCUDF_WHEELHOUSE}"/libcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)" \
     "$(echo "${PYLIBCUDF_WHEELHOUSE}"/pylibcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)[test]"
@@ -48,12 +55,12 @@ rapids-logger "Install cudf, pylibcudf, and test requirements"
 # notes:
 #
 #   * echo to expand wildcard before adding `[test]` requires for pip
-#   * need to provide --constraint="${PIP_CONSTRAINT}" because that environment variable is
-#     ignored if any other --constraint are passed via the CLI
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
 #
 rapids-pip-retry install \
     -v \
-    --constraint ./constraints.txt \
+    --prefer-binary \
     --constraint "${PIP_CONSTRAINT}" \
     "$(echo "${CUDF_WHEELHOUSE}"/cudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)[test]" \
     "$(echo "${LIBCUDF_WHEELHOUSE}"/libcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)" \
diff --git a/ci/test_wheel_cudf_polars.sh b/ci/test_wheel_cudf_polars.sh
index abdaeebd165..b67f8fd9f7b 100755
--- a/ci/test_wheel_cudf_polars.sh
+++ b/ci/test_wheel_cudf_polars.sh
@@ -18,17 +18,17 @@ PYLIBCUDF_WHEELHOUSE=$(rapids-download-from-github "$(rapids-package-name "wheel
 rapids-logger "Installing cudf_polars and its dependencies"
 
 # generate constraints (possibly pinning to oldest support versions of dependencies)
-rapids-generate-pip-constraints py_test_cudf_polars ./constraints.txt
+rapids-generate-pip-constraints py_test_cudf_polars "${PIP_CONSTRAINT}"
 
 # notes:
 #
-#   * echo to expand wildcard before adding `[test,experimental]` requires for pip
-#   * need to provide --constraint="${PIP_CONSTRAINT}" because that environment variable is
-#     ignored if any other --constraint are passed via the CLI
+#   * echo to expand wildcard before adding `[test]` requires for pip
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
 #
 rapids-pip-retry install \
     -v \
-    --constraint ./constraints.txt \
+    --prefer-binary \
     --constraint "${PIP_CONSTRAINT}" \
     "$(echo "${CUDF_POLARS_WHEELHOUSE}"/cudf_polars_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)[test,experimental]" \
     "$(echo "${LIBCUDF_WHEELHOUSE}"/libcudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)" \
diff --git a/ci/test_wheel_dask_cudf.sh b/ci/test_wheel_dask_cudf.sh
index 636ced077d7..8930d85eab3 100755
--- a/ci/test_wheel_dask_cudf.sh
+++ b/ci/test_wheel_dask_cudf.sh
@@ -17,17 +17,17 @@ PYLIBCUDF_WHEELHOUSE=$(rapids-download-from-github "$(rapids-package-name "wheel
 rapids-logger "Install dask_cudf, cudf, pylibcudf, and test requirements"
 
 # generate constraints (possibly pinning to oldest support versions of dependencies)
-rapids-generate-pip-constraints py_test_dask_cudf ./constraints.txt
+rapids-generate-pip-constraints py_test_dask_cudf "${PIP_CONSTRAINT}"
 
 # notes:
 #
 #   * echo to expand wildcard before adding `[test]` requires for pip
-#   * need to provide --constraint="${PIP_CONSTRAINT}" because that environment variable is
-#     ignored if any other --constraint are passed via the CLI
+#   * just providing --constraint="${PIP_CONSTRAINT}" to be explicit, and because
+#     that environment variable is ignored if any other --constraint are passed via the CLI
 #
 rapids-pip-retry install \
   -v \
-  --constraint ./constraints.txt \
+  --prefer-binary \
   --constraint "${PIP_CONSTRAINT}" \
   "$(echo "${CUDF_WHEELHOUSE}"/cudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)" \
   "$(echo "${DASK_CUDF_WHEELHOUSE}"/dask_cudf_"${RAPIDS_PY_CUDA_SUFFIX}"*.whl)[test]" \
diff --git a/conda/recipes/libcudf/recipe.yaml b/conda/recipes/libcudf/recipe.yaml
index 65a20c827ed..67f921d9a67 100644
--- a/conda/recipes/libcudf/recipe.yaml
+++ b/conda/recipes/libcudf/recipe.yaml
@@ -129,7 +129,6 @@ outputs:
       run:
         - ${{ pin_compatible("cuda-version", upper_bound="x", lower_bound="x") }}
         - cuda-nvrtc
-        - libnvjitlink >=${{ cuda_version }}
         - if: linux and x86_64
           then:
             - libcufile
@@ -154,7 +153,6 @@ outputs:
           - libcufile
           - libcurand
           - libkvikio
-          - libnvjitlink
           - librdkafka
           - libzlib
           - librmm
diff --git a/dependencies.yaml b/dependencies.yaml
index a241b951728..4c43cf6d1d4 100644
--- a/dependencies.yaml
+++ b/dependencies.yaml
@@ -204,6 +204,7 @@ files:
       table: project.optional-dependencies
       key: test
     includes:
+      - cuda_version
       - test_python_common
       - test_python_cudf_common
       - test_python_cudf
@@ -352,6 +353,7 @@ files:
       table: project.optional-dependencies
       key: test
     includes:
+      - cuda_version
       - depends_on_dask_cuda
       - numpy_run
       - test_python_common
@@ -390,6 +392,7 @@ files:
       table: project.optional-dependencies
       key: test
     includes:
+      - cuda_version
       - depends_on_dask_cuda
       - test_python_common
       - test_python_cudf_common
@@ -563,6 +566,8 @@ dependencies:
       - output_types: [requirements, pyproject]
         packages:
           - pyarrow>=19.0.0
+  # 'cuda_version' intentionally does not contain fallback entries... we want
+  # a loud error if an unsupported 'cuda' value is passed
   cuda_version:
     specific:
       - output_types: conda
@@ -591,6 +596,51 @@ dependencies:
               cuda: "13.1"
             packages:
               - cuda-version=13.1
+      - output_types: requirements
+        matrices:
+          # if use_cuda_wheels=false is provided, do not add dependencies on any CUDA wheels
+          # (e.g. for DLFW and pip devcontainers)
+          - matrix:
+              use_cuda_wheels: "false"
+            packages:
+          - matrix:
+              arch: aarch64
+              cuda: "12.2"
+              use_cuda_wheels: "true"
+            packages:
+              # some components (like nvidia-cublas-cu12 and nvidia-cuda-nvcc-cu12) didn't have
+              # aarch64 wheels until CTK 12.3, so allow a slightly looser bound here
+              - cuda-toolkit>=12.2,<12.4
+          - matrix:
+              cuda: "12.2"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.2.*
+          - matrix:
+              cuda: "12.5"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.5.*
+          - matrix:
+              cuda: "12.8"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.8.*
+          - matrix:
+              cuda: "12.9"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.9.*
+          - matrix:
+              cuda: "13.0"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==13.0.*
+          - matrix:
+              cuda: "13.1"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==13.1.*
   cuda:
     common:
       - output_types: [conda]
@@ -604,14 +654,30 @@ dependencies:
     specific:
       - output_types: [requirements, pyproject]
         matrices:
+          # if use_cuda_wheels=false is provided, do not add dependencies on any CUDA wheels
+          # (e.g. for DLFW and pip devcontainers)
           - matrix:
-              cuda: "12.*"
+              use_cuda_wheels: "false"
             packages:
-              - nvidia-nvjitlink-cu12>=12.9
-          # fallback to CUDA 13 versions if 'cuda' is '13.*' or not provided
+          # We always want nvJitLink >= whatever minor version was available in the build environment
+          # nvJitLink tends to follow the CTK's minor versions, but check at https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html
+          #
+          # ref: https://github.com/rapidsai/cudf/pull/20873
+          - matrix:
+              cuda: "12.9"
+              use_cuda_wheels: "true"
+            packages:
+              - nvidia-nvjitlink-cu12>=12.9,<13
+          - matrix:
+              cuda: "13.*"
+              use_cuda_wheels: "true"
+            packages:
+              - &nvjitlink_cu13 nvidia-nvjitlink>=13.0,<14
+          # if no matching matrix selectors passed, list the CUDA 13 requirement
+          # (just as a source of documentation, as this populates pyproject.toml in source control)
           - matrix:
             packages:
-              - nvidia-nvjitlink>=13.0
+              - *nvjitlink_cu13
   develop:
     common:
       - output_types: [conda, requirements]
@@ -725,14 +791,21 @@ dependencies:
     specific:
       - output_types: [requirements, pyproject]
         matrices:
+          # if use_cuda_wheels=false is provided, do not add dependencies on any CUDA wheels
+          # (e.g. for DLFW and pip devcontainers)
+          - matrix:
+              use_cuda_wheels: "false"
+            packages:
           - matrix:
               cuda: "12.*"
               cuda_suffixed: "true"
+              use_cuda_wheels: "true"
             packages:
               - cuda-toolkit[nvcc,nvrtc]==12.*
           - matrix:
               cuda: "13.*"
               cuda_suffixed: "true"
+              use_cuda_wheels: "true"
             packages:
               - cuda-toolkit[nvcc,nvrtc]==13.*
           - {matrix: null, packages: []}
diff --git a/python/cudf/cudf_pandas_tests/third_party_integration_tests/dependencies.yaml b/python/cudf/cudf_pandas_tests/third_party_integration_tests/dependencies.yaml
index db3c55948c6..233c2e5b998 100644
--- a/python/cudf/cudf_pandas_tests/third_party_integration_tests/dependencies.yaml
+++ b/python/cudf/cudf_pandas_tests/third_party_integration_tests/dependencies.yaml
@@ -138,14 +138,12 @@ dependencies:
       - output_types: conda
         packages:
           - pre-commit
+  # 'cuda_version' intentionally does not contain fallback entries... we want
+  # a loud error if an unsupported 'cuda' value is passed
   cuda_version:
     specific:
       - output_types: conda
         matrices:
-          - matrix:
-              cuda: "12.0"
-            packages:
-              - cuda-version=12.0
           - matrix:
               cuda: "12.2"
             packages:
@@ -170,6 +168,51 @@ dependencies:
               cuda: "13.1"
             packages:
               - cuda-version=13.1
+      - output_types: requirements
+        matrices:
+          # if use_cuda_wheels=false is provided, do not add dependencies on any CUDA wheels
+          # (e.g. for DLFW and pip devcontainers)
+          - matrix:
+              use_cuda_wheels: "false"
+            packages:
+          - matrix:
+              arch: aarch64
+              cuda: "12.2"
+              use_cuda_wheels: "true"
+            packages:
+              # some components (like nvidia-cublas-cu12 and nvidia-cuda-nvcc-cu12) didn't have
+              # aarch64 wheels until CTK 12.3, so allow a slightly looser bound here
+              - cuda-toolkit>=12.2,<12.4
+          - matrix:
+              cuda: "12.2"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.2.*
+          - matrix:
+              cuda: "12.5"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.5.*
+          - matrix:
+              cuda: "12.8"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.8.*
+          - matrix:
+              cuda: "12.9"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==12.9.*
+          - matrix:
+              cuda: "13.0"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==13.0.*
+          - matrix:
+              cuda: "13.1"
+              use_cuda_wheels: "true"
+            packages:
+              - cuda-toolkit==13.1.*
   py_version:
     specific:
       - output_types: conda
diff --git a/python/cudf/pyproject.toml b/python/cudf/pyproject.toml
index 9810afa959c..29c34ddadd0 100644
--- a/python/cudf/pyproject.toml
+++ b/python/cudf/pyproject.toml
@@ -121,7 +121,7 @@ xfail_strict = true
 [tool.rapids-build-backend]
 build-backend = "scikit_build_core.build"
 dependencies-file = "../../dependencies.yaml"
-matrix-entry = "cuda_suffixed=true"
+matrix-entry = "cuda_suffixed=true;use_cuda_wheels=true"
 requires = [
     "cmake>=3.30.4",
     "cython>=3.2.2",
diff --git a/python/libcudf/pyproject.toml b/python/libcudf/pyproject.toml
index 173b3e36d24..f3d9ecd58be 100644
--- a/python/libcudf/pyproject.toml
+++ b/python/libcudf/pyproject.toml
@@ -73,6 +73,6 @@ requires = [
     "librmm==26.4.*,>=0.0.0a0",
     "ninja",
     "nvidia-libnvcomp==5.1.0.21",
-    "nvidia-nvjitlink>=13.0",
+    "nvidia-nvjitlink>=13.0,<14",
     "rapids-logger==0.2.*,>=0.0.0a0",
 ] # This list was generated by `rapids-dependency-file-generator`. To make changes, edit ../../dependencies.yaml and run `rapids-dependency-file-generator`.
diff --git a/python/pylibcudf/pyproject.toml b/python/pylibcudf/pyproject.toml
index 79cfa5c9e14..d3ecdb6fa4d 100644
--- a/python/pylibcudf/pyproject.toml
+++ b/python/pylibcudf/pyproject.toml
@@ -118,7 +118,7 @@ xfail_strict = true
 [tool.rapids-build-backend]
 build-backend = "scikit_build_core.build"
 dependencies-file = "../../dependencies.yaml"
-matrix-entry = "cuda_suffixed=true"
+matrix-entry = "cuda_suffixed=true;use_cuda_wheels=true"
 requires = [
     "cmake>=3.30.4",
     "cython>=3.2.2",