diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
index f5fe18010e9..a2426be1131 100644
--- a/cpp/CMakeLists.txt
+++ b/cpp/CMakeLists.txt
@@ -43,6 +43,7 @@ option(ENABLE_MULTI_DEVICE
        "Enable building with multi device support (requires NCCL, MPI,...)" ON)
 option(ENABLE_UCX "Enable building with UCX (Uniform Communication X) support"
        ON)
+option(NVRTC_DYNAMIC_LINKING "Link against the dynamic NVRTC libraries" OFF)
 option(USING_OSS_CUTLASS_LOW_LATENCY_GEMM
        "Using open sourced Cutlass low latency gemm kernel" ON)
 option(USING_OSS_CUTLASS_FP4_GEMM "Using open sourced Cutlass fp4 gemm kernel"
@@ -147,10 +148,17 @@ set(CURAND_LIB CUDA::curand)
 set(CUDA_DRV_LIB CUDA::cuda_driver)
 set(CUDA_NVML_LIB CUDA::nvml)
 set(CUDA_RT_LIB CUDA::cudart_static)
-set(NVRTC_LIB CUDA::nvrtc_static)
-set(NVRTC_BUILTINS_LIB CUDA::nvrtc_builtins_static)
+set(NVPTX_LIB CUDA::nvptxcompiler_static)
 set(CMAKE_CUDA_RUNTIME_LIBRARY Static)
 
+if(NVRTC_DYNAMIC_LINKING)
+  set(NVRTC_LIB CUDA::nvrtc)
+  set(NVRTC_BUILTINS_LIB CUDA::nvrtc_builtins)
+else()
+  set(NVRTC_LIB CUDA::nvrtc_static)
+  set(NVRTC_BUILTINS_LIB CUDA::nvrtc_builtins_static)
+endif()
+
 resolve_dirs(CUDAToolkit_INCLUDE_DIRS "${CUDAToolkit_INCLUDE_DIRS}")
 
 message(STATUS "CUDA library status:")
diff --git a/cpp/tensorrt_llm/kernels/decoderMaskedMultiheadAttention/decoderXQAImplJIT/nvrtcWrapper/CMakeLists.txt b/cpp/tensorrt_llm/kernels/decoderMaskedMultiheadAttention/decoderXQAImplJIT/nvrtcWrapper/CMakeLists.txt
index a86831256df..79b0c2ed086 100644
--- a/cpp/tensorrt_llm/kernels/decoderMaskedMultiheadAttention/decoderXQAImplJIT/nvrtcWrapper/CMakeLists.txt
+++ b/cpp/tensorrt_llm/kernels/decoderMaskedMultiheadAttention/decoderXQAImplJIT/nvrtcWrapper/CMakeLists.txt
@@ -17,7 +17,7 @@ add_dependencies(nvrtc_wrapper_src xqa_sources_h)
 target_include_directories(nvrtc_wrapper_src
                            PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/xqa_build)
 target_link_libraries(
-  nvrtc_wrapper_src PUBLIC ${NVRTC_LIB} ${NVRTC_BUILTINS_LIB} ${CUDA_DRV_LIB}
-                           ${CUDA_RT_LIB})
+  nvrtc_wrapper_src PUBLIC ${NVPTX_LIB} ${NVRTC_LIB} ${NVRTC_BUILTINS_LIB}
+                           ${CUDA_DRV_LIB} ${CUDA_RT_LIB})
 set_property(TARGET nvrtc_wrapper_src PROPERTY POSITION_INDEPENDENT_CODE ON)
 set_property(TARGET nvrtc_wrapper_src PROPERTY CUDA_RESOLVE_DEVICE_SYMBOLS ON)
diff --git a/scripts/build_wheel.py b/scripts/build_wheel.py
index 19f35e4f668..5f50e49fe3c 100755
--- a/scripts/build_wheel.py
+++ b/scripts/build_wheel.py
@@ -304,7 +304,8 @@ def main(*,
          nvtx: bool = False,
          skip_stubs: bool = False,
          generate_fmha: bool = False,
-         no_venv: bool = False):
+         no_venv: bool = False,
+         nvrtc_dynamic_linking: bool = False):
 
     if clean:
         clean_wheel = True
@@ -404,6 +405,9 @@ def main(*,
     if fast_build:
         cmake_def_args.append(f"-DFAST_BUILD=ON")
 
+    if nvrtc_dynamic_linking:
+        cmake_def_args.append(f"-DNVRTC_DYNAMIC_LINKING=ON")
+
     targets = ["tensorrt_llm", "nvinfer_plugin_tensorrt_llm"]
 
     if cpp_only:
@@ -787,6 +791,11 @@ def add_arguments(parser: ArgumentParser):
         help=
         "Use the current Python interpreter without creating a virtual environment."
     )
+    parser.add_argument(
+        "--nvrtc_dynamic_linking",
+        action="store_true",
+        help="Link against the dynamic NVRTC libraries and not the static ones."
+    )
 
 
 if __name__ == "__main__":