diff --git a/lib/galaxy/celery/tasks.py b/lib/galaxy/celery/tasks.py
index e1ac7456ad41..3b2e4c6272a7 100644
--- a/lib/galaxy/celery/tasks.py
+++ b/lib/galaxy/celery/tasks.py
@@ -185,14 +185,19 @@ def set_metadata(
     dataset_id: int,
     model_class: str = "HistoryDatasetAssociation",
     overwrite: bool = True,
+    ensure_can_set_metadata: bool = True,
     task_user_id: Optional[int] = None,
 ):
+    """
+    ensure_can_set_metadata can be bypassed for new outputs.
+    """
     manager = _get_dataset_manager(hda_manager, ldda_manager, model_class)
     dataset_instance = manager.by_id(dataset_id)
-    can_set_metadata = manager.ensure_can_set_metadata(dataset_instance, raiseException=False)
-    if not can_set_metadata:
-        log.info(f"Setting metadata is not allowed for {model_class} {dataset_instance.id}")
-        return
+    if ensure_can_set_metadata:
+        can_set_metadata = manager.ensure_can_set_metadata(dataset_instance, raiseException=False)
+        if not can_set_metadata:
+            log.info(f"Setting metadata is not allowed for {model_class} {dataset_instance.id}")
+            return
     try:
         if overwrite:
             hda_manager.overwrite_metadata(dataset_instance)
diff --git a/lib/galaxy/jobs/__init__.py b/lib/galaxy/jobs/__init__.py
index d63ba30916df..c930229df5e0 100644
--- a/lib/galaxy/jobs/__init__.py
+++ b/lib/galaxy/jobs/__init__.py
@@ -2020,8 +2020,9 @@ def fail(message=job.info, exception=None):
         # Certain tools require tasks to be completed after job execution
         # ( this used to be performed in the "exec_after_process" hook, but hooks are deprecated ).
         param_dict = self.get_param_dict(job)
+        task_wrapper = None
         try:
-            self.tool.exec_after_process(
+            task_wrapper = self.tool.exec_after_process(
                 self.app, inp_data, out_data, param_dict, job=job, final_job_state=final_job_state
             )
         except Exception as e:
@@ -2063,6 +2064,10 @@ def fail(message=job.info, exception=None):
             self.sa_session.commit()
         if job.state == job.states.ERROR:
             self._report_error()
+        elif task_wrapper:
+            # Only task is setting metadata (if necessary) on expression tool output.
+            # The dataset state is SETTING_METADATA, which delays dependent jobs until the task completes.
+            task_wrapper.delay()
         cleanup_job = self.cleanup_job
         delete_files = cleanup_job == "always" or (job.state == job.states.OK and cleanup_job == "onsuccess")
         self.cleanup(delete_files=delete_files)
diff --git a/lib/galaxy/tools/__init__.py b/lib/galaxy/tools/__init__.py
index 065033fe3029..570008c7c64b 100644
--- a/lib/galaxy/tools/__init__.py
+++ b/lib/galaxy/tools/__init__.py
@@ -2876,7 +2876,27 @@ def exec_after_process(self, app, inp_data, out_data, param_dict, job=None, **kw
                         break
                 if copy_object is None:
                     raise exceptions.MessageException("Failed to find dataset output.")
-                out_data[key].copy_from(copy_object, include_metadata=True)
+                output = out_data[key]
+                # if change_datatype PJA is associated with expression tool output the new output already has
+                # the desired datatype, so we use it. If the extension is "data" there's no change_dataset PJA and
+                # we want to use the existing extension.
+                new_ext = output.extension if output.extension != "data" else copy_object.extension
+                require_metadata_regeneration = copy_object.extension != new_ext
+                output.copy_from(copy_object, include_metadata=not require_metadata_regeneration)
+                output.extension = new_ext
+                if require_metadata_regeneration:
+                    if app.config.enable_celery_tasks:
+                        from galaxy.celery.tasks import set_metadata
+
+                        output._state = model.Dataset.states.SETTING_METADATA
+                        return set_metadata.si(
+                            dataset_id=output.id, task_user_id=output.history.user_id, ensure_can_set_metadata=False
+                        )
+                    else:
+                        # TODO: move exec_after_process into metadata script so this doesn't run on the headnode ?
+                        output.init_meta()
+                        output.set_meta()
+                        output.set_metadata_success_state()
 
     def parse_environment_variables(self, tool_source):
         """Setup environment variable for inputs file."""