diff --git a/pandas/core/arrays/datetimelike.py b/pandas/core/arrays/datetimelike.py
index b2bff0b0142e2..098feb9f1290f 100644
--- a/pandas/core/arrays/datetimelike.py
+++ b/pandas/core/arrays/datetimelike.py
@@ -396,6 +396,34 @@ def floor(self, freq, ambiguous="raise", nonexistent="raise"):
     def ceil(self, freq, ambiguous="raise", nonexistent="raise"):
         return self._round(freq, RoundTo.PLUS_INFTY, ambiguous, nonexistent)
 
+    def _with_freq(self, freq):
+        """
+        Helper to set our freq in-place, returning self to allow method chaining.
+
+        Parameters
+        ----------
+        freq : DateOffset, None, or "infer"
+
+        Returns
+        -------
+        self
+        """
+        # GH#29843
+        if freq is None:
+            # Always valid
+            pass
+        elif len(self) == 0 and isinstance(freq, DateOffset):
+            # Always valid.  In the TimedeltaArray case, we assume this
+            #  is a Tick offset.
+            pass
+        else:
+            # As an internal method, we can ensure this assertion always holds
+            assert freq == "infer"
+            freq = frequencies.to_offset(self.inferred_freq)
+
+        self._freq = freq
+        return self
+
 
 class DatetimeLikeArrayMixin(ExtensionOpsMixin, AttributesMixin, ExtensionArray):
     """
@@ -1157,7 +1185,7 @@ def _add_timedeltalike_scalar(self, other):
         if new_freq is not None:
             # fastpath that doesnt require inference
             return type(self)(new_values, dtype=self.dtype, freq=new_freq)
-        return type(self)._from_sequence(new_values, dtype=self.dtype, freq="infer")
+        return type(self)(new_values, dtype=self.dtype)._with_freq("infer")
 
     def _add_timedelta_arraylike(self, other):
         """
@@ -1187,7 +1215,7 @@ def _add_timedelta_arraylike(self, other):
             mask = (self._isnan) | (other._isnan)
             new_values[mask] = iNaT
 
-        return type(self)._from_sequence(new_values, dtype=self.dtype, freq="infer")
+        return type(self)(new_values, dtype=self.dtype)._with_freq("infer")
 
     def _add_nat(self):
         """
diff --git a/pandas/core/arrays/datetimes.py b/pandas/core/arrays/datetimes.py
index 2d74582b049f7..e2a13df069ae2 100644
--- a/pandas/core/arrays/datetimes.py
+++ b/pandas/core/arrays/datetimes.py
@@ -697,7 +697,7 @@ def _add_offset(self, offset):
                 # GH#30336 _from_sequence won't be able to infer self.tz
                 return type(self)._from_sequence(result).tz_localize(self.tz)
 
-        return type(self)._from_sequence(result, freq="infer")
+        return type(self)._from_sequence(result)._with_freq("infer")
 
     def _sub_datetimelike_scalar(self, other):
         # subtract a datetime from myself, yielding a ndarray[timedelta64[ns]]
@@ -1031,7 +1031,7 @@ def normalize(self):
             new_values[not_null] = new_values[not_null] - adjustment
         else:
             new_values = conversion.normalize_i8_timestamps(self.asi8, self.tz)
-        return type(self)._from_sequence(new_values, freq="infer").tz_localize(self.tz)
+        return type(self)(new_values)._with_freq("infer").tz_localize(self.tz)
 
     def to_period(self, freq=None):
         """
diff --git a/pandas/core/arrays/period.py b/pandas/core/arrays/period.py
index 6ee439de414f1..c24b0b5fa64b8 100644
--- a/pandas/core/arrays/period.py
+++ b/pandas/core/arrays/period.py
@@ -185,6 +185,7 @@ def _from_sequence(
             validate_dtype_freq(scalars.dtype, freq)
             if copy:
                 scalars = scalars.copy()
+            assert isinstance(scalars, PeriodArray)  # for mypy
             return scalars
 
         periods = np.asarray(scalars, dtype=object)
@@ -452,7 +453,7 @@ def to_timestamp(self, freq=None, how="start"):
         new_data = self.asfreq(freq, how=how)
 
         new_data = libperiod.periodarr_to_dt64arr(new_data.asi8, base)
-        return DatetimeArray._from_sequence(new_data, freq="infer")
+        return DatetimeArray(new_data)._with_freq("infer")
 
     # --------------------------------------------------------------------
 
diff --git a/pandas/core/indexes/datetimelike.py b/pandas/core/indexes/datetimelike.py
index 2f641a3d4c111..25333b3a08dce 100644
--- a/pandas/core/indexes/datetimelike.py
+++ b/pandas/core/indexes/datetimelike.py
@@ -43,7 +43,7 @@
 from pandas.core.ops import get_op_result_name
 from pandas.core.tools.timedeltas import to_timedelta
 
-from pandas.tseries.frequencies import DateOffset, to_offset
+from pandas.tseries.frequencies import DateOffset
 
 _index_doc_kwargs = dict(ibase._index_doc_kwargs)
 
@@ -623,19 +623,7 @@ def _set_freq(self, freq):
         freq : DateOffset, None, or "infer"
         """
         # GH#29843
-        if freq is None:
-            # Always valid
-            pass
-        elif len(self) == 0 and isinstance(freq, DateOffset):
-            # Always valid.  In the TimedeltaIndex case, we assume this
-            #  is a Tick offset.
-            pass
-        else:
-            # As an internal method, we can ensure this assertion always holds
-            assert freq == "infer"
-            freq = to_offset(self.inferred_freq)
-
-        self._data._freq = freq
+        self._data._with_freq(freq)
 
     def _shallow_copy(self, values=None, name: Label = lib.no_default):
         name = self.name if name is lib.no_default else name