diff --git a/pandas/core/construction.py b/pandas/core/construction.py
index 78a7f1890b5de..ddb6473280a05 100644
--- a/pandas/core/construction.py
+++ b/pandas/core/construction.py
@@ -328,7 +328,7 @@ def array(
         elif inferred_dtype == "interval":
             try:
                 return IntervalArray(data, copy=copy)
-            except ValueError:
+            except (ValueError, TypeError):
                 # We may have a mixture of `closed` here.
                 # We choose to return an ndarray, rather than raising.
                 pass
diff --git a/pandas/core/indexes/base.py b/pandas/core/indexes/base.py
index 12b343ab5d895..7c504edfe08d0 100644
--- a/pandas/core/indexes/base.py
+++ b/pandas/core/indexes/base.py
@@ -6327,13 +6327,6 @@ def _maybe_cast_data_without_dtype(subarr):
     converted : np.ndarray or ExtensionArray
     dtype : np.dtype or ExtensionDtype
     """
-    # Runtime import needed bc IntervalArray imports Index
-    from pandas.core.arrays import (
-        DatetimeArray,
-        IntervalArray,
-        PeriodArray,
-        TimedeltaArray,
-    )
 
     assert subarr.dtype == object, subarr.dtype
     inferred = lib.infer_dtype(subarr, skipna=False)
@@ -6356,38 +6349,10 @@ def _maybe_cast_data_without_dtype(subarr):
         data = np.asarray(subarr).astype(np.float64, copy=False)
         return data
 
-    elif inferred == "interval":
-        try:
-            data = IntervalArray._from_sequence(subarr, copy=False)
-            return data
-        except (ValueError, TypeError):
-            # GH27172: mixed closed Intervals --> object dtype
-            pass
-    elif inferred == "boolean":
-        # don't support boolean explicitly ATM
-        pass
-    elif inferred != "string":
-        if inferred.startswith("datetime"):
-            try:
-                data = DatetimeArray._from_sequence(subarr, copy=False)
-                return data
-            except (ValueError, OutOfBoundsDatetime):
-                # GH 27011
-                # If we have mixed timezones, just send it
-                # down the base constructor
-                pass
-
-        elif inferred.startswith("timedelta"):
-            tda = TimedeltaArray._from_sequence(subarr, copy=False)
-            return tda
-        elif inferred == "period":
-            try:
-                data = PeriodArray._from_sequence(subarr)
-                return data
-            except IncompatibleFrequency:
-                pass
-
-    return subarr
+    else:
+        alt = sanitize_array(subarr, None)
+        alt = ensure_wrapped_if_datetimelike(alt)
+        return alt
 
 
 def _try_convert_to_int_array(
diff --git a/pandas/tests/arrays/test_array.py b/pandas/tests/arrays/test_array.py
index 5d2b7c43f6765..3e04fb3ceb40b 100644
--- a/pandas/tests/arrays/test_array.py
+++ b/pandas/tests/arrays/test_array.py
@@ -153,6 +153,16 @@
             "category",
             pd.Categorical([pd.Period("2000", "D"), pd.Period("2001", "D")]),
         ),
+        # mixed-dtype intervals -> cast to object, dont raise TypeError
+        (
+            [pd.Interval(0, 1), pd.Interval(pd.Timestamp(0), pd.Timestamp(1))],
+            None,
+            PandasArray(
+                np.array(
+                    [pd.Interval(0, 1), pd.Interval(pd.Timestamp(0), pd.Timestamp(1))]
+                )
+            ),
+        ),
     ],
 )
 def test_array(data, dtype, expected):
diff --git a/pandas/tests/indexes/test_index_new.py b/pandas/tests/indexes/test_index_new.py
index 5937f43102190..38d8b36a0b137 100644
--- a/pandas/tests/indexes/test_index_new.py
+++ b/pandas/tests/indexes/test_index_new.py
@@ -1,6 +1,10 @@
 """
 Tests for the Index constructor conducting inference.
 """
+from datetime import (
+    date,
+    datetime,
+)
 from decimal import Decimal
 
 import numpy as np
@@ -138,6 +142,24 @@ def test_constructor_mixed_nat_objs_infers_object(self, swap_objs):
         tm.assert_index_equal(Index(data), expected)
         tm.assert_index_equal(Index(np.array(data, dtype=object)), expected)
 
+    def test_constructor_datetime_inference(self):
+        # Index.__new__ inference to datetime matches Series
+
+        # mixed str/datetime
+        data = ["2012-12-31", datetime(2013, 1, 1), datetime(2013, 1, 2)]
+        result = Index(data)
+        expected = DatetimeIndex([Timestamp(x) for x in data])
+        tm.assert_index_equal(result, expected)
+
+        # date objects cast to Timestamp
+        data = [date(2020, 1, 1), Timestamp("2020-01-02 00:00:00")]
+        result = Index(data)
+        expected = DatetimeIndex([Timestamp(x) for x in data])
+        tm.assert_index_equal(result, expected)
+
+        result = Index(data[::-1])
+        tm.assert_index_equal(result, expected[::-1])
+
 
 class TestDtypeEnforced:
     # check we don't silently ignore the dtype keyword