Try alternative approach

src/gt4py/storage/cartesian/utils.py

@@ -55,12 +55,35 @@
             device_type=core_defs.DeviceType.CUDA,
             array_utils=allocators.cupy_array_utils,
         )
-    else:
+    elif CUPY_DEVICE == core_defs.DeviceType.ROCM:
         _GPUBufferAllocator = allocators.NDArrayBufferAllocator(
             device_type=core_defs.DeviceType.ROCM,
             array_utils=allocators.cupy_array_utils,
         )
 
+        class ROCmNDArray(cp.ndarray):
+            def __new__(cls, input_array: "cp.ndarray") -> ROCmNDArray:
+                return (
+                    input_array
+                    if isinstance(input_array, ROCmNDArray)
+                    else cp.asarray(input_array).view(cls)
+                )
+
+            @property
+            def __cuda_array_interface__(self) -> dict:
+                return {
+                    "shape": self.shape,
+                    "typestr": self.dtype.descr[0][1],
+                    "descr": self.dtype.descr,
+                    "stream": 1,
+                    "version": 3,
+                    "strides": self.strides,
+                    "data": (self.data.ptr, False),
+                }
+
+    else:
+        raise ValueError("Cupy is available but no suitable device was found.")
+
 
 def _idx_from_order(order):
     return list(np.argsort(order))
@@ -274,4 +297,10 @@ def allocate_gpu(
         byte_alignment=alignment_bytes,
         aligned_index=aligned_index,
     )
-    return buffer.buffer, cast("cp.ndarray", buffer.ndarray)
+
+    buffer_ndarray = cast("cp.ndarray", buffer.ndarray)
+
+    if cp is not None and cp.cuda.get_hipcc_path() is not None:
+        buffer_ndarray = ROCmNDArray(buffer_ndarray)
+
+    return buffer.buffer, buffer_ndarray