diff --git a/doc/source/whatsnew/v1.5.0.rst b/doc/source/whatsnew/v1.5.0.rst
index 71394a858aefe..d5b092e21a596 100644
--- a/doc/source/whatsnew/v1.5.0.rst
+++ b/doc/source/whatsnew/v1.5.0.rst
@@ -240,6 +240,7 @@ Other Deprecations
 - Deprecated :meth:`Series.is_monotonic` and :meth:`Index.is_monotonic` in favor of :meth:`Series.is_monotonic_increasing` and :meth:`Index.is_monotonic_increasing` (:issue:`45422`, :issue:`21335`)
 - Deprecated behavior of :meth:`DatetimeIndex.astype`, :meth:`TimedeltaIndex.astype`, :meth:`PeriodIndex.astype` when converting to an integer dtype other than ``int64``. In a future version, these will convert to exactly the specified dtype (instead of always ``int64``) and will raise if the conversion overflows (:issue:`45034`)
 - Deprecated the ``__array_wrap__`` method of DataFrame and Series, rely on standard numpy ufuncs instead (:issue:`45451`)
+- Deprecated treating float-dtype data as wall-times when passed with a timezone to :class:`Series` or :class:`DatetimeIndex` (:issue:`45573`)
 - Deprecated the behavior of :meth:`Series.fillna` and :meth:`DataFrame.fillna` with ``timedelta64[ns]`` dtype and incompatible fill value; in a future version this will cast to a common dtype (usually object) instead of raising, matching the behavior of other dtypes (:issue:`45746`)
 - Deprecated the ``warn`` parameter in :func:`infer_freq` (:issue:`45947`)
 
diff --git a/pandas/core/arrays/datetimes.py b/pandas/core/arrays/datetimes.py
index f652e0b870ae7..e21f2e9d7b46e 100644
--- a/pandas/core/arrays/datetimes.py
+++ b/pandas/core/arrays/datetimes.py
@@ -2066,7 +2066,7 @@ def _sequence_to_dt64ns(
         inferred_freq = data.freq
 
     # By this point we are assured to have either a numpy array or Index
-    data, copy = maybe_convert_dtype(data, copy)
+    data, copy = maybe_convert_dtype(data, copy, tz=tz)
     data_dtype = getattr(data, "dtype", None)
 
     if (
@@ -2246,7 +2246,7 @@ def objects_to_datetime64ns(
         raise TypeError(result)
 
 
-def maybe_convert_dtype(data, copy: bool):
+def maybe_convert_dtype(data, copy: bool, tz: tzinfo | None = None):
     """
     Convert data based on dtype conventions, issuing deprecation warnings
     or errors where appropriate.
@@ -2255,6 +2255,7 @@ def maybe_convert_dtype(data, copy: bool):
     ----------
     data : np.ndarray or pd.Index
     copy : bool
+    tz : tzinfo or None, default None
 
     Returns
     -------
@@ -2274,8 +2275,23 @@ def maybe_convert_dtype(data, copy: bool):
         #  as wall-times instead of UTC timestamps.
         data = data.astype(DT64NS_DTYPE)
         copy = False
-        # TODO: deprecate this behavior to instead treat symmetrically
-        #  with integer dtypes.  See discussion in GH#23675
+        if (
+            tz is not None
+            and len(data) > 0
+            and not timezones.is_utc(timezones.maybe_get_tz(tz))
+        ):
+            # GH#23675, GH#45573 deprecate to treat symmetrically with integer dtypes
+            warnings.warn(
+                "The behavior of DatetimeArray._from_sequence with a timezone-aware "
+                "dtype and floating-dtype data is deprecated. In a future version, "
+                "this data will be interpreted as nanosecond UTC timestamps "
+                "instead of wall-times, matching the behavior with integer dtypes. "
+                "To retain the old behavior, explicitly cast to 'datetime64[ns]' "
+                "before passing the data to pandas. To get the future behavior, "
+                "first cast to 'int64'.",
+                FutureWarning,
+                stacklevel=find_stack_level(),
+            )
 
     elif is_timedelta64_dtype(data.dtype) or is_bool_dtype(data.dtype):
         # GH#29794 enforcing deprecation introduced in GH#23539
diff --git a/pandas/core/tools/datetimes.py b/pandas/core/tools/datetimes.py
index 3f0a5deb97548..33ed64c7ae364 100644
--- a/pandas/core/tools/datetimes.py
+++ b/pandas/core/tools/datetimes.py
@@ -27,6 +27,7 @@
     iNaT,
     nat_strings,
     parsing,
+    timezones,
 )
 from pandas._libs.tslibs.parsing import (  # noqa:F401
     DateParseError,
@@ -364,7 +365,7 @@ def _convert_listlike_datetimes(
     # NB: this must come after unit transformation
     orig_arg = arg
     try:
-        arg, _ = maybe_convert_dtype(arg, copy=False)
+        arg, _ = maybe_convert_dtype(arg, copy=False, tz=timezones.maybe_get_tz(tz))
     except TypeError:
         if errors == "coerce":
             npvalues = np.array(["NaT"], dtype="datetime64[ns]").repeat(len(arg))
diff --git a/pandas/tests/dtypes/test_missing.py b/pandas/tests/dtypes/test_missing.py
index 1917fc615118a..a986e8d659202 100644
--- a/pandas/tests/dtypes/test_missing.py
+++ b/pandas/tests/dtypes/test_missing.py
@@ -415,23 +415,28 @@ def test_array_equivalent(dtype_equal):
         TimedeltaIndex([1, np.nan]),
         dtype_equal=dtype_equal,
     )
+
+    msg = "will be interpreted as nanosecond UTC timestamps instead of wall-times"
+    with tm.assert_produces_warning(FutureWarning, match=msg):
+        dti1 = DatetimeIndex([0, np.nan], tz="US/Eastern")
+        dti2 = DatetimeIndex([0, np.nan], tz="CET")
+        dti3 = DatetimeIndex([1, np.nan], tz="US/Eastern")
+
     assert array_equivalent(
-        DatetimeIndex([0, np.nan], tz="US/Eastern"),
-        DatetimeIndex([0, np.nan], tz="US/Eastern"),
+        dti1,
+        dti1,
         dtype_equal=dtype_equal,
     )
     assert not array_equivalent(
-        DatetimeIndex([0, np.nan], tz="US/Eastern"),
-        DatetimeIndex([1, np.nan], tz="US/Eastern"),
+        dti1,
+        dti3,
         dtype_equal=dtype_equal,
     )
     # The rest are not dtype_equal
+    assert not array_equivalent(DatetimeIndex([0, np.nan]), dti1)
     assert not array_equivalent(
-        DatetimeIndex([0, np.nan]), DatetimeIndex([0, np.nan], tz="US/Eastern")
-    )
-    assert not array_equivalent(
-        DatetimeIndex([0, np.nan], tz="CET"),
-        DatetimeIndex([0, np.nan], tz="US/Eastern"),
+        dti2,
+        dti1,
     )
 
     assert not array_equivalent(DatetimeIndex([0, np.nan]), TimedeltaIndex([0, np.nan]))
diff --git a/pandas/tests/internals/test_internals.py b/pandas/tests/internals/test_internals.py
index f286dc4a09cb2..eda902d34bff5 100644
--- a/pandas/tests/internals/test_internals.py
+++ b/pandas/tests/internals/test_internals.py
@@ -140,7 +140,7 @@ def create_block(typestr, placement, item_shape=None, num_offset=0, maker=new_bl
         assert m is not None, f"incompatible typestr -> {typestr}"
         tz = m.groups()[0]
         assert num_items == 1, "must have only 1 num items for a tz-aware"
-        values = DatetimeIndex(np.arange(N) * 1e9, tz=tz)._data
+        values = DatetimeIndex(np.arange(N) * 10**9, tz=tz)._data
         values = ensure_block_shape(values, ndim=len(shape))
     elif typestr in ("timedelta", "td", "m8[ns]"):
         values = (mat * 1).astype("m8[ns]")
diff --git a/pandas/tests/series/methods/test_astype.py b/pandas/tests/series/methods/test_astype.py
index ff9bb3edeedb1..630f4cc3194f4 100644
--- a/pandas/tests/series/methods/test_astype.py
+++ b/pandas/tests/series/methods/test_astype.py
@@ -381,7 +381,12 @@ def test_astype_nan_to_bool(self):
     )
     def test_astype_ea_to_datetimetzdtype(self, dtype):
         # GH37553
-        result = Series([4, 0, 9], dtype=dtype).astype(DatetimeTZDtype(tz="US/Pacific"))
+        ser = Series([4, 0, 9], dtype=dtype)
+        warn = FutureWarning if ser.dtype.kind == "f" else None
+        msg = "with a timezone-aware dtype and floating-dtype data"
+        with tm.assert_produces_warning(warn, match=msg):
+            result = ser.astype(DatetimeTZDtype(tz="US/Pacific"))
+
         expected = Series(
             {
                 0: Timestamp("1969-12-31 16:00:00.000000004-08:00", tz="US/Pacific"),