diff --git a/doc/source/whatsnew/v1.1.0.rst b/doc/source/whatsnew/v1.1.0.rst
index 85fe33b7c83e8..fe9715313f0db 100644
--- a/doc/source/whatsnew/v1.1.0.rst
+++ b/doc/source/whatsnew/v1.1.0.rst
@@ -367,6 +367,7 @@ Reshaping
 - :meth:`Series.append` will now raise a ``TypeError`` when passed a DataFrame or a sequence containing Dataframe (:issue:`31413`)
 - :meth:`DataFrame.replace` and :meth:`Series.replace` will raise a ``TypeError`` if ``to_replace`` is not an expected type. Previously the ``replace`` would fail silently (:issue:`18634`)
 - Bug in :meth:`DataFrame.apply` where callback was called with :class:`Series` parameter even though ``raw=True`` requested. (:issue:`32423`)
+- Bug in :meth:`DataFrame.pivot_table` losing timezone information when creating a :class:`MultiIndex` level from a column with timezone-aware dtype (:issue:`32558`)
 
 
 Sparse
diff --git a/pandas/core/indexes/multi.py b/pandas/core/indexes/multi.py
index 122097f4478d7..5bffc4ec552af 100644
--- a/pandas/core/indexes/multi.py
+++ b/pandas/core/indexes/multi.py
@@ -565,6 +565,7 @@ def from_product(cls, iterables, sortorder=None, names=lib.no_default):
         if names is lib.no_default:
             names = [getattr(it, "name", None) for it in iterables]
 
+        # codes are all ndarrays, so cartesian_product is lossless
         codes = cartesian_product(codes)
         return MultiIndex(levels, codes, sortorder=sortorder, names=names)
 
diff --git a/pandas/core/reshape/util.py b/pandas/core/reshape/util.py
index d8652c9b4fac9..7abb14303f8cc 100644
--- a/pandas/core/reshape/util.py
+++ b/pandas/core/reshape/util.py
@@ -2,8 +2,6 @@
 
 from pandas.core.dtypes.common import is_list_like
 
-import pandas.core.common as com
-
 
 def cartesian_product(X):
     """
@@ -51,9 +49,20 @@ def cartesian_product(X):
         # if any factor is empty, the cartesian product is empty
         b = np.zeros_like(cumprodX)
 
-    return [
-        np.tile(
-            np.repeat(np.asarray(com.values_from_object(x)), b[i]), np.product(a[i])
-        )
-        for i, x in enumerate(X)
-    ]
+    return [_tile_compat(np.repeat(x, b[i]), np.product(a[i])) for i, x in enumerate(X)]
+
+
+def _tile_compat(arr, num: int):
+    """
+    Index compat for np.tile.
+
+    Notes
+    -----
+    Does not support multi-dimensional `num`.
+    """
+    if isinstance(arr, np.ndarray):
+        return np.tile(arr, num)
+
+    # Otherwise we have an Index
+    taker = np.tile(np.arange(len(arr)), num)
+    return arr.take(taker)
diff --git a/pandas/tests/reshape/test_pivot.py b/pandas/tests/reshape/test_pivot.py
index 75c3c565e9d58..cdb1a73abc431 100644
--- a/pandas/tests/reshape/test_pivot.py
+++ b/pandas/tests/reshape/test_pivot.py
@@ -1026,6 +1026,14 @@ def test_pivot_table_multiindex_only(self, cols):
 
         tm.assert_frame_equal(result, expected)
 
+    def test_pivot_table_retains_tz(self):
+        dti = date_range("2016-01-01", periods=3, tz="Europe/Amsterdam")
+        df = DataFrame({"A": np.random.randn(3), "B": np.random.randn(3), "C": dti})
+        result = df.pivot_table(index=["B", "C"], dropna=False)
+
+        # check tz retention
+        assert result.index.levels[1].equals(dti)
+
     def test_pivot_integer_columns(self):
         # caused by upstream bug in unstack
 
diff --git a/pandas/tests/reshape/test_util.py b/pandas/tests/reshape/test_util.py
index cd518dda4edbf..9d074b5ade425 100644
--- a/pandas/tests/reshape/test_util.py
+++ b/pandas/tests/reshape/test_util.py
@@ -25,6 +25,22 @@ def test_datetimeindex(self):
         tm.assert_index_equal(result1, expected1)
         tm.assert_index_equal(result2, expected2)
 
+    def test_tzaware_retained(self):
+        x = date_range("2000-01-01", periods=2, tz="US/Pacific")
+        y = np.array([3, 4])
+        result1, result2 = cartesian_product([x, y])
+
+        expected = x.repeat(2)
+        tm.assert_index_equal(result1, expected)
+
+    def test_tzaware_retained_categorical(self):
+        x = date_range("2000-01-01", periods=2, tz="US/Pacific").astype("category")
+        y = np.array([3, 4])
+        result1, result2 = cartesian_product([x, y])
+
+        expected = x.repeat(2)
+        tm.assert_index_equal(result1, expected)
+
     def test_empty(self):
         # product of empty factors
         X = [[], [0, 1], []]