diff --git a/.devcontainer/gcc11-cuda12.9/devcontainer.json b/.devcontainer/gcc11-cuda12.9/devcontainer.json
index 93e4bfc7f..8bb11d44b 100644
--- a/.devcontainer/gcc11-cuda12.9/devcontainer.json
+++ b/.devcontainer/gcc11-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-gcc11-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-gcc11-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/gcc12-cuda12.9/devcontainer.json b/.devcontainer/gcc12-cuda12.9/devcontainer.json
index 23f730cd7..bc0c2c5d4 100644
--- a/.devcontainer/gcc12-cuda12.9/devcontainer.json
+++ b/.devcontainer/gcc12-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-gcc12-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-gcc12-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/gcc13-cuda12.9/devcontainer.json b/.devcontainer/gcc13-cuda12.9/devcontainer.json
index bfa5181cc..76a707979 100644
--- a/.devcontainer/gcc13-cuda12.9/devcontainer.json
+++ b/.devcontainer/gcc13-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-gcc13-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-gcc13-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/gcc14-cuda12.9/devcontainer.json b/.devcontainer/gcc14-cuda12.9/devcontainer.json
index f75488e15..c245bbd79 100644
--- a/.devcontainer/gcc14-cuda12.9/devcontainer.json
+++ b/.devcontainer/gcc14-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-gcc14-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-gcc14-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/gcc15-cuda12.9/devcontainer.json b/.devcontainer/gcc15-cuda12.9/devcontainer.json
index 34e821a83..cdc71a127 100644
--- a/.devcontainer/gcc15-cuda12.9/devcontainer.json
+++ b/.devcontainer/gcc15-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-gcc15-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-gcc15-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/llvm16-cuda12.9/devcontainer.json b/.devcontainer/llvm16-cuda12.9/devcontainer.json
index bf6015371..ff84e9845 100644
--- a/.devcontainer/llvm16-cuda12.9/devcontainer.json
+++ b/.devcontainer/llvm16-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-llvm16-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-llvm16-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/llvm22-cuda12.0/devcontainer.json b/.devcontainer/llvm22-cuda12.0/devcontainer.json
index 169e88864..bb8b08ba3 100644
--- a/.devcontainer/llvm22-cuda12.0/devcontainer.json
+++ b/.devcontainer/llvm22-cuda12.0/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-llvm22-cuda12.0",
+  "image": "rapidsai/devcontainers:26.12-cpp-llvm22-cuda12.0",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/llvm22-cuda12.9/devcontainer.json b/.devcontainer/llvm22-cuda12.9/devcontainer.json
index 977d2f5cb..ef600841c 100644
--- a/.devcontainer/llvm22-cuda12.9/devcontainer.json
+++ b/.devcontainer/llvm22-cuda12.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-llvm22-cuda12.9",
+  "image": "rapidsai/devcontainers:26.12-cpp-llvm22-cuda12.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/llvm22-cuda13.2/devcontainer.json b/.devcontainer/llvm22-cuda13.2/devcontainer.json
index 9ce013676..80ab674fb 100644
--- a/.devcontainer/llvm22-cuda13.2/devcontainer.json
+++ b/.devcontainer/llvm22-cuda13.2/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-llvm22-cuda13.2",
+  "image": "rapidsai/devcontainers:26.12-cpp-llvm22-cuda13.2",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/nvhpc26.1/devcontainer.json b/.devcontainer/nvhpc26.5/devcontainer.json
similarity index 97%
rename from .devcontainer/nvhpc26.1/devcontainer.json
rename to .devcontainer/nvhpc26.5/devcontainer.json
index 13a4fb8b3..094ca060a 100644
--- a/.devcontainer/nvhpc26.1/devcontainer.json
+++ b/.devcontainer/nvhpc26.5/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-nvhpc26.1",
+  "image": "rapidsai/devcontainers:26.12-cpp-nvhpc26.5",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.devcontainer/nvhpc26.3/devcontainer.json b/.devcontainer/nvhpc26.9/devcontainer.json
similarity index 97%
rename from .devcontainer/nvhpc26.3/devcontainer.json
rename to .devcontainer/nvhpc26.9/devcontainer.json
index a208e4fcb..530e8eae0 100644
--- a/.devcontainer/nvhpc26.3/devcontainer.json
+++ b/.devcontainer/nvhpc26.9/devcontainer.json
@@ -1,5 +1,5 @@
 {
-  "image": "rapidsai/devcontainers:26.08-cpp-nvhpc26.3",
+  "image": "rapidsai/devcontainers:26.12-cpp-nvhpc26.9",
 
   "hostRequirements": { "gpu": "optional" },
 
diff --git a/.github/workflows/ci.gpu.yml b/.github/workflows/ci.gpu.yml
index 950c7e1d9..493d9cf76 100644
--- a/.github/workflows/ci.gpu.yml
+++ b/.github/workflows/ci.gpu.yml
@@ -21,14 +21,14 @@ jobs:
         include:
           - { name: "clang 22",    cuda: "12.0", cxx: "clang++", build: "Release", tag: "llvm22-cuda12.0", gpu: "v100", sm: "70", driver: "latest", arch: "amd64", flags: "-Wno-deprecated-declarations" }
           - { name: "clang 22",    cuda: "12.9", cxx: "clang++", build: "Release", tag: "llvm22-cuda12.9", gpu: "v100", sm: "70", driver: "latest", arch: "amd64", flags: "" }
-          - { name: "nvc++ 26.1",  cuda: "13.1", cxx: "mpic++",  build: "Release", tag: "nvhpc26.1",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
-          - { name: "nvc++ 26.1",  cuda: "13.1", cxx: "mpic++",  build: "Debug",   tag: "nvhpc26.1",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
-          - { name: "nvc++ 26.3",  cuda: "13.1", cxx: "mpic++",  build: "Release", tag: "nvhpc26.3",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
-          - { name: "nvc++ 26.3",  cuda: "13.1", cxx: "mpic++",  build: "Debug",   tag: "nvhpc26.3",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
+          - { name: "nvc++ 26.5",  cuda: "13.2", cxx: "mpic++",  build: "Release", tag: "nvhpc26.5",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
+          - { name: "nvc++ 26.5",  cuda: "13.2", cxx: "mpic++",  build: "Debug",   tag: "nvhpc26.5",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
+          - { name: "nvc++ 26.9",  cuda: "13.3", cxx: "mpic++",  build: "Release", tag: "nvhpc26.9",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
+          - { name: "nvc++ 26.9",  cuda: "13.3", cxx: "mpic++",  build: "Debug",   tag: "nvhpc26.9",       gpu: "l4",   sm: "75", driver: "latest", arch: "amd64", flags: "" }
     runs-on: linux-${{ matrix.arch }}-gpu-${{ matrix.gpu }}-${{ matrix.driver }}-1
     container:
       options: -u root
-      image: rapidsai/devcontainers:26.08-cpp-${{ matrix.tag }}
+      image: rapidsai/devcontainers:26.12-cpp-${{ matrix.tag }}
       env:
         NVIDIA_VISIBLE_DEVICES: "${{ env.NVIDIA_VISIBLE_DEVICES }}"
     permissions:
diff --git a/README.md b/README.md
index aa7cafbc1..44356a4ba 100644
--- a/README.md
+++ b/README.md
@@ -75,7 +75,7 @@ int main() {
 | Clang | 16 | |
 | MSVC | 14.43 | |
 | Xcode (Apple Clang) | 16 | |
-| nvc++ | 25.9 | required for [GPU support](#gpu-support) |
+| nvc++ | 26.5 | required for [GPU support](#gpu-support) |
 
 Requires `-std=c++20` or later.
 
diff --git a/include/stdexec/__detail/__execution_legacy.hpp b/include/stdexec/__detail/__execution_legacy.hpp
index e4a835f7e..e9574ba76 100644
--- a/include/stdexec/__detail/__execution_legacy.hpp
+++ b/include/stdexec/__detail/__execution_legacy.hpp
@@ -18,7 +18,17 @@
 #include "__config.hpp"
 
 #if !STDEXEC_NO_STDCPP_EXECUTION_POLICY()
-#  include <execution>  // IWYU pragma: keep
+// NVHPC's <execution> re-includes <stdexec/execution.hpp>. Including that from here, while
+// headers such as __bulk.hpp are only partially parsed, creates an include cycle that leaves
+// those headers' contents undefined for the headers pulled in by the cycle. Suppress NVHPC's
+// stdexec include; it is still honored the next time <execution> is included.
+#  if STDEXEC_NVHPC() && !defined(_NVHPC_STDEXEC_INCLUDE_IN_PROGRESS)
+#    define _NVHPC_STDEXEC_INCLUDE_IN_PROGRESS
+#    include <execution>  // IWYU pragma: keep
+#    undef _NVHPC_STDEXEC_INCLUDE_IN_PROGRESS
+#  else
+#    include <execution>  // IWYU pragma: keep
+#  endif
 #endif
 
 #include "__prologue.hpp"