diff --git a/pandas/core/indexes/base.py b/pandas/core/indexes/base.py
index 40215ea87f978..a412b8477897d 100644
--- a/pandas/core/indexes/base.py
+++ b/pandas/core/indexes/base.py
@@ -41,8 +41,8 @@
 
 from pandas.core.dtypes.cast import (
     find_common_type,
+    infer_dtype_from,
     maybe_cast_to_integer_array,
-    maybe_promote,
     validate_numeric_casting,
 )
 from pandas.core.dtypes.common import (
@@ -87,7 +87,7 @@
     ABCTimedeltaIndex,
 )
 from pandas.core.dtypes.inference import is_dict_like
-from pandas.core.dtypes.missing import array_equivalent, isna
+from pandas.core.dtypes.missing import array_equivalent, is_valid_nat_for_dtype, isna
 
 from pandas.core import missing, ops
 from pandas.core.accessor import CachedAccessor
@@ -5735,16 +5735,14 @@ def insert(self, loc: int, item):
         # Note: this method is overridden by all ExtensionIndex subclasses,
         #  so self is never backed by an EA.
         item = lib.item_from_zerodim(item)
+        if is_valid_nat_for_dtype(item, self.dtype) and self.dtype != object:
+            item = self._na_value
 
         try:
             item = self._validate_fill_value(item)
         except TypeError:
-            if is_scalar(item):
-                dtype, item = maybe_promote(self.dtype, item)
-            else:
-                # maybe_promote would raise ValueError
-                dtype = np.dtype(object)
-
+            inferred, _ = infer_dtype_from(item)
+            dtype = find_common_type([self.dtype, inferred])
             return self.astype(dtype).insert(loc, item)
 
         arr = np.asarray(self)