diff --git a/tensorflow/core/kernels/cross_op.cc b/tensorflow/core/kernels/cross_op.cc
index b29524f1f9e..ab81e9ac300 100644
--- a/tensorflow/core/kernels/cross_op.cc
+++ b/tensorflow/core/kernels/cross_op.cc
@@ -87,7 +87,7 @@ class CrossOp : public OpKernel {
 TF_CALL_REAL_NUMBER_TYPES(REGISTER_CPU_KERNEL);
 #undef REGISTER_CPU_KERNEL
 
-#if GOOGLE_CUDA
+#if GOOGLE_CUDA || TENSORFLOW_USE_ROCM
 // Forward declarations of the function specializations for GPU (to prevent
 // building the GPU versions here, they will be built compiling _gpu.cu.cc).
 namespace functor {
diff --git a/tensorflow/core/kernels/cross_op_gpu.cu.cc b/tensorflow/core/kernels/cross_op_gpu.cu.cc
index 4a37f6cfbbc..e9fd4523a53 100644
--- a/tensorflow/core/kernels/cross_op_gpu.cu.cc
+++ b/tensorflow/core/kernels/cross_op_gpu.cu.cc
@@ -13,7 +13,7 @@ See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
 
-#if GOOGLE_CUDA
+#if GOOGLE_CUDA || TENSORFLOW_USE_ROCM
 
 #define EIGEN_USE_GPU
 
@@ -31,4 +31,4 @@ TF_CALL_REAL_NUMBER_TYPES(INSTANTIATE_GPU_KERNEL);
 
 }  // namespace tensorflow
 
-#endif  // GOOGLE_CUDA
+#endif  // GOOGLE_CUDA || TENSORFLOW_USE_ROCM