BUG: date comparison fails when series is all pd.NaT values (#61200)

Author: Mohit-Kundu

Diff for: doc/source/whatsnew/v3.0.0.rst

@@ -647,6 +647,7 @@ Categorical
 
 Datetimelike
 ^^^^^^^^^^^^
+- Bug in :attr:`Series.dt.date` returned datetime dtype on Series with all NaT values; now returns object dtype (:issue:`61188`)
 - Bug in :attr:`is_year_start` where a DateTimeIndex constructed via a date_range with frequency 'MS' wouldn't have the correct year or quarter start attributes (:issue:`57377`)
 - Bug in :class:`DataFrame` raising ``ValueError`` when ``dtype`` is ``timedelta64`` and ``data`` is a list containing ``None`` (:issue:`60064`)
 - Bug in :class:`Timestamp` constructor failing to raise when ``tz=None`` is explicitly specified in conjunction with timezone-aware ``tzinfo`` or data (:issue:`48688`)

Diff for: pandas/core/indexes/accessors.py

@@ -108,7 +108,9 @@ def _delegate_property_get(self, name: str):
         else:
             index = self._parent.index
         # return the result as a Series
-        return Series(result, index=index, name=self.name).__finalize__(self._parent)
+        return Series(
+            result, index=index, name=self.name, dtype=result.dtype
+        ).__finalize__(self._parent)
 
     def _delegate_property_set(self, name: str, value, *args, **kwargs) -> NoReturn:
         raise ValueError(

Diff for: pandas/tests/series/indexing/test_datetime.py

@@ -491,3 +491,23 @@ def test_compare_datetime_with_all_none():
     result = ser > ser2
     expected = Series([False, False])
     tm.assert_series_equal(result, expected)
+
+
+def test_dt_date_dtype_all_nat_is_object():
+    # Ensure .dt.date on all-NaT Series returns object dtype and not datetime64
+    # GH#61188
+    s = Series([pd.NaT, pd.NaT], dtype="datetime64[s]")
+    result = s.dt.date
+
+    expected = Series([pd.NaT, pd.NaT], dtype=object)
+
+    tm.assert_series_equal(result, expected)
+
+
+def test_dt_date_all_nat_le_date():
+    # All-NaT Series should not raise error when compared to a datetime.date
+    # GH#61188
+    s = Series([pd.NaT, pd.NaT], dtype="datetime64[s]")
+    result = s.dt.date <= datetime.now().date()
+    expected = Series([False, False])
+    tm.assert_series_equal(result, expected)