diff --git a/pandas/core/arrays/datetimes.py b/pandas/core/arrays/datetimes.py
index e28a1a2326d17..1b0a045dca180 100644
--- a/pandas/core/arrays/datetimes.py
+++ b/pandas/core/arrays/datetimes.py
@@ -1904,6 +1904,23 @@ def std(
 # Constructor Helpers
 
 
+def sequence_to_datetimes(
+    data, allow_object: bool = False
+) -> Union[np.ndarray, DatetimeArray]:
+    """
+    Parse/convert the passed data to either DatetimeArray or np.ndarray[object].
+    """
+    result, tz, freq = sequence_to_dt64ns(
+        data, allow_object=allow_object, allow_mixed=True
+    )
+    if result.dtype == object:
+        return result
+
+    dtype = tz_to_dtype(tz)
+    dta = DatetimeArray._simple_new(result, freq=freq, dtype=dtype)
+    return dta
+
+
 def sequence_to_dt64ns(
     data,
     dtype=None,
@@ -1912,6 +1929,9 @@ def sequence_to_dt64ns(
     dayfirst=False,
     yearfirst=False,
     ambiguous="raise",
+    *,
+    allow_object: bool = False,
+    allow_mixed: bool = False,
 ):
     """
     Parameters
@@ -1924,6 +1944,11 @@ def sequence_to_dt64ns(
     yearfirst : bool, default False
     ambiguous : str, bool, or arraylike, default 'raise'
         See pandas._libs.tslibs.tzconversion.tz_localize_to_utc.
+    allow_object : bool, default False
+        Whether to return an object-dtype ndarray instead of raising if the
+        data contains more than one timezone.
+    allow_mixed : bool, default False
+        Interpret integers as timestamps when datetime objects are also present.
 
     Returns
     -------
@@ -1987,7 +2012,11 @@ def sequence_to_dt64ns(
             # data comes back here as either i8 to denote UTC timestamps
             #  or M8[ns] to denote wall times
             data, inferred_tz = objects_to_datetime64ns(
-                data, dayfirst=dayfirst, yearfirst=yearfirst
+                data,
+                dayfirst=dayfirst,
+                yearfirst=yearfirst,
+                allow_object=allow_object,
+                allow_mixed=allow_mixed,
             )
             if tz and inferred_tz:
                 #  two timezones: convert to intended from base UTC repr
@@ -1995,6 +2024,9 @@ def sequence_to_dt64ns(
                 data = data.view(DT64NS_DTYPE)
             elif inferred_tz:
                 tz = inferred_tz
+            elif allow_object and data.dtype == object:
+                # We encountered mixed-timezones.
+                return data, None, None
 
         data_dtype = data.dtype
 
@@ -2053,6 +2085,7 @@ def objects_to_datetime64ns(
     errors="raise",
     require_iso8601=False,
     allow_object=False,
+    allow_mixed: bool = False,
 ):
     """
     Convert data to array of timestamps.
@@ -2069,6 +2102,8 @@ def objects_to_datetime64ns(
     allow_object : bool
         Whether to return an object-dtype ndarray instead of raising if the
         data contains more than one timezone.
+    allow_mixed : bool, default False
+        Interpret integers as timestamps when datetime objects are also present.
 
     Returns
     -------
@@ -2097,6 +2132,7 @@ def objects_to_datetime64ns(
             dayfirst=dayfirst,
             yearfirst=yearfirst,
             require_iso8601=require_iso8601,
+            allow_mixed=allow_mixed,
         )
         result = result.reshape(data.shape, order=order)
     except ValueError as err:
@@ -2133,7 +2169,7 @@ def objects_to_datetime64ns(
         raise TypeError(result)
 
 
-def maybe_convert_dtype(data, copy):
+def maybe_convert_dtype(data, copy: bool):
     """
     Convert data based on dtype conventions, issuing deprecation warnings
     or errors where appropriate.
diff --git a/pandas/core/dtypes/cast.py b/pandas/core/dtypes/cast.py
index d456f9c56e309..f5ec6616659ba 100644
--- a/pandas/core/dtypes/cast.py
+++ b/pandas/core/dtypes/cast.py
@@ -29,10 +29,7 @@
 
 import numpy as np
 
-from pandas._libs import (
-    lib,
-    tslib,
-)
+from pandas._libs import lib
 from pandas._libs.tslibs import (
     NaT,
     OutOfBoundsDatetime,
@@ -1605,8 +1602,8 @@ def maybe_cast_to_datetime(
     try to cast the array/value to a datetimelike dtype, converting float
     nan to iNaT
     """
+    from pandas.core.arrays.datetimes import sequence_to_datetimes
     from pandas.core.arrays.timedeltas import sequence_to_td64ns
-    from pandas.core.tools.datetimes import to_datetime
 
     if not is_list_like(value):
         raise TypeError("value must be listlike")
@@ -1664,19 +1661,19 @@ def maybe_cast_to_datetime(
 
                     try:
                         if is_datetime64:
-                            dti = to_datetime(value, errors="raise")
+                            dta = sequence_to_datetimes(value, allow_object=False)
                             # GH 25843: Remove tz information since the dtype
                             # didn't specify one
-                            if dti.tz is not None:
-                                dti = dti.tz_localize(None)
-                            value = dti._values
+                            if dta.tz is not None:
+                                dta = dta.tz_localize(None)
+                            value = dta
                         elif is_datetime64tz:
                             # The string check can be removed once issue #13712
                             # is solved. String data that is passed with a
                             # datetime64tz is assumed to be naive which should
                             # be localized to the timezone.
                             is_dt_string = is_string_dtype(value.dtype)
-                            dta = to_datetime(value, errors="raise").array
+                            dta = sequence_to_datetimes(value, allow_object=False)
                             if dta.tz is not None:
                                 value = dta.astype(dtype, copy=False)
                             elif is_dt_string:
@@ -1691,24 +1688,10 @@ def maybe_cast_to_datetime(
                             value, _ = sequence_to_td64ns(value)
                     except OutOfBoundsDatetime:
                         raise
-                    except ValueError as err:
+                    except ValueError:
                         # TODO(GH#40048): only catch dateutil's ParserError
                         #  once we can reliably import it in all supported versions
-                        if "mixed datetimes and integers in passed array" in str(err):
-                            # We need to catch this in array_to_datetime, otherwise
-                            #  we end up going through numpy which will lose nanoseconds
-                            #  from Timestamps
-                            try:
-                                i8vals, tz = tslib.array_to_datetime(
-                                    value, allow_mixed=True
-                                )
-                            except ValueError:
-                                pass
-                            else:
-                                from pandas.core.arrays import DatetimeArray
-
-                                dta = DatetimeArray(i8vals).tz_localize(tz)
-                                value = dta
+                        pass
 
         # coerce datetimelike to object
         elif is_datetime64_dtype(