diff --git a/vllm/model_executor/model_loader/default_loader.py b/vllm/model_executor/model_loader/default_loader.py
index deee2324960dd..5d24e72919693 100644
--- a/vllm/model_executor/model_loader/default_loader.py
+++ b/vllm/model_executor/model_loader/default_loader.py
@@ -241,9 +241,6 @@ class DefaultModelLoader(BaseModelLoader):
                     self.load_config.pt_load_map_location,
                 )
 
-        if current_platform.is_tpu():
-            pass
-
         if self.counter_before_loading_weights == 0.0:
             self.counter_before_loading_weights = time.perf_counter()
         # Apply the prefix.
diff --git a/vllm/v1/worker/tpu_worker.py b/vllm/v1/worker/tpu_worker.py
index 085b119e12600..4c73d6c92d391 100644
--- a/vllm/v1/worker/tpu_worker.py
+++ b/vllm/v1/worker/tpu_worker.py
@@ -11,6 +11,7 @@ logger = init_logger(__name__)
 
 _R = TypeVar("_R")
 
+# TODO(weiyulin) Remove this file after adding an official way to use hardware plugin
 if USE_TPU_INFERENCE:
     from tpu_inference.worker.tpu_worker import TPUWorker as TpuInferenceWorker