diff --git a/pkgs/development/libraries/nvidia-thrust/default.nix b/pkgs/development/libraries/nvidia-thrust/default.nix
deleted file mode 100644
index f68b57f193b7..000000000000
--- a/pkgs/development/libraries/nvidia-thrust/default.nix
+++ /dev/null
@@ -1,102 +0,0 @@
-{ lib
-, config
-, fetchFromGitHub
-, stdenv
-, cmake
-, pkg-config
-, cudaPackages ? { }
-, symlinkJoin
-, tbb
-, hostSystem ? "CPP"
-, deviceSystem ? if config.cudaSupport then "CUDA" else "OMP"
-}:
-
-# Policy for device_vector<T>
-assert builtins.elem deviceSystem [
-  "CPP" # Serial on CPU
-  "OMP" # Parallel with OpenMP
-  "TBB" # Parallel with Intel TBB
-  "CUDA" # Parallel on GPU
-];
-
-# Policy for host_vector<T>
-# Always lives on CPU, but execution can be made parallel
-assert builtins.elem hostSystem [ "CPP" "OMP" "TBB" ];
-
-let
-  pname = "nvidia-thrust";
-  version = "1.16.0";
-
-  inherit (cudaPackages) backendStdenv cudaFlags;
-  cudaCapabilities = map cudaFlags.dropDot cudaFlags.cudaCapabilities;
-
-  tbbSupport = builtins.elem "TBB" [ deviceSystem hostSystem ];
-  cudaSupport = deviceSystem == "CUDA";
-
-  # TODO: Would like to use this:
-  cudaJoined = symlinkJoin {
-    name = "cuda-packages-unsplit";
-    paths = with cudaPackages; [
-      cuda_nvcc
-      cuda_nvrtc # symbols: cudaLaunchDevice, &c; notice postBuild
-      cuda_cudart # cuda_runtime.h
-      libcublas
-    ];
-    postBuild = ''
-      ln -s $out/lib $out/lib64
-    '';
-  };
-in
-stdenv.mkDerivation {
-  inherit pname version;
-
-  src = fetchFromGitHub {
-    owner = "NVIDIA";
-    repo = "thrust";
-    rev = version;
-    fetchSubmodules = true;
-    hash = "sha256-/EyznxWKuHuvHNjq+SQg27IaRbtkjXR2zlo2YgCWmUQ=";
-  };
-
-  # NVIDIA's "compiler hacks" seem like work-arounds for legacy toolchains and
-  # cause us errors such as:
-  # > Thrust's test harness uses CMAKE_CXX_COMPILER for the CUDA host compiler.
-  # > Refusing to overwrite specified CMAKE_CUDA_HOST_COMPILER
-  # So we un-fix cmake after them:
-  postPatch = ''
-    echo > cmake/ThrustCompilerHacks.cmake
-  '';
-
-  buildInputs = lib.optionals tbbSupport [ tbb ];
-
-  nativeBuildInputs = [
-    cmake
-    pkg-config
-  ] ++ lib.optionals cudaSupport [
-    # Goes in native build inputs because thrust looks for headers
-    # in a path relative to nvcc...
-    cudaJoined
-  ];
-
-  cmakeFlags = [
-    "-DTHRUST_INCLUDE_CUB_CMAKE=${if cudaSupport then "ON" else "OFF"}"
-    "-DTHRUST_DEVICE_SYSTEM=${deviceSystem}"
-    "-DTHRUST_HOST_SYSTEM=${hostSystem}"
-    "-DTHRUST_AUTO_DETECT_COMPUTE_ARCHS=OFF"
-    "-DTHRUST_DISABLE_ARCH_BY_DEFAULT=ON"
-  ] ++ lib.optionals cudaFlags.enableForwardCompat [
-    "-DTHRUST_ENABLE_COMPUTE_FUTURE=ON"
-  ] ++ map (sm: "THRUST_ENABLE_COMPUTE_${sm}") cudaCapabilities;
-
-  passthru = {
-    inherit cudaSupport cudaPackages cudaJoined;
-  };
-
-  meta = with lib; {
-    description = "A high-level C++ parallel algorithms library that builds on top of CUDA, TBB, OpenMP, etc";
-    homepage = "https://github.com/NVIDIA/thrust";
-    license = licenses.asl20;
-    platforms = platforms.unix;
-    maintainers = with maintainers; [ SomeoneSerge ];
-  };
-}
diff --git a/pkgs/development/libraries/science/math/faiss/default.nix b/pkgs/development/libraries/science/math/faiss/default.nix
index 21e6cbf858cd..25ac539e05f2 100644
--- a/pkgs/development/libraries/science/math/faiss/default.nix
+++ b/pkgs/development/libraries/science/math/faiss/default.nix
@@ -6,8 +6,6 @@
 , cmake
 , cudaPackages ? { }
 , cudaSupport ? config.cudaSupport
-, nvidia-thrust
-, useThrustSourceBuild ? true
 , pythonSupport ? true
 , pythonPackages
 , llvmPackages
@@ -27,8 +25,6 @@
 , runCommand
 }@inputs:
 
-assert cudaSupport -> nvidia-thrust.cudaSupport;
-
 let
   pname = "faiss";
   version = "1.7.4";
@@ -44,9 +40,6 @@ let
       cuda_cudart # cuda_runtime.h
       libcublas
       libcurand
-    ] ++ lib.optionals useThrustSourceBuild [
-      nvidia-thrust
-    ] ++ lib.optionals (!useThrustSourceBuild) [
       cuda_cccl
     ] ++ lib.optionals (cudaPackages ? cuda_profiler_api) [
       cuda_profiler_api # cuda_profiler_api.h
diff --git a/pkgs/top-level/aliases.nix b/pkgs/top-level/aliases.nix
index e655c34c7e82..daab3488385d 100644
--- a/pkgs/top-level/aliases.nix
+++ b/pkgs/top-level/aliases.nix
@@ -644,6 +644,7 @@ mapAliases ({
   noto-fonts-cjk = noto-fonts-cjk-sans; # Added 2021-12-16
   noto-fonts-emoji = noto-fonts-color-emoji; # Added 2023-09-09
   noto-fonts-extra = noto-fonts; # Added 2023-04-08
+  nvidia-thrust = throw "nvidia-thrust has been removed because the project was deprecated; use cudaPackages.cuda_cccl";
 
   ### O ###
 
diff --git a/pkgs/top-level/all-packages.nix b/pkgs/top-level/all-packages.nix
index 75e47757a8eb..1e9069ef0ead 100644
--- a/pkgs/top-level/all-packages.nix
+++ b/pkgs/top-level/all-packages.nix
@@ -11276,16 +11276,6 @@ with pkgs;
 
   nvfetcher = haskell.lib.compose.justStaticExecutables haskellPackages.nvfetcher;
 
-  nvidia-thrust = callPackage ../development/libraries/nvidia-thrust { };
-
-  nvidia-thrust-intel = callPackage ../development/libraries/nvidia-thrust {
-    hostSystem = "TBB";
-    deviceSystem = if config.cudaSupport then "CUDA" else "TBB";
-  };
-
-  nvidia-thrust-cuda = callPackage ../development/libraries/nvidia-thrust {
-    deviceSystem = "CUDA";
-  };
 
   miller = callPackage ../tools/text/miller { };
 
@@ -39988,7 +39978,6 @@ with pkgs;
 
   faissWithCuda = faiss.override {
     cudaSupport = true;
-    nvidia-thrust = nvidia-thrust-cuda;
   };
 
   fityk = callPackage ../applications/science/misc/fityk { };