diff --git a/superset/connectors/sqla/models.py b/superset/connectors/sqla/models.py
index b49e9ba4d33..c48a222e382 100644
--- a/superset/connectors/sqla/models.py
+++ b/superset/connectors/sqla/models.py
@@ -1652,8 +1652,20 @@ class SqlaTable(
             if has_timegrain or force_type_check:
                 try:
                     # probe adhoc column type
+                    # Most databases populate cursor.description from query-plan
+                    # metadata, so WHERE FALSE (zero rows, no table scan) is
+                    # preferred — it avoids hitting row-read limits enforced by
+                    # engines like ClickHouse (max_rows_to_read).
+                    # A small number of drivers (Druid, Pinot) instead build
+                    # cursor.description by inspecting the first returned row;
+                    # for those we fall back to LIMIT 1.
                     tbl, _ = self.get_from_clause(template_processor)
-                    qry = sa.select([sqla_column]).limit(1).select_from(tbl)
+                    if self.db_engine_spec.type_probe_needs_row:
+                        qry = sa.select([sqla_column]).limit(1).select_from(tbl)
+                    else:
+                        qry = (
+                            sa.select([sqla_column]).where(sa.false()).select_from(tbl)
+                        )
                     sql = self.database.compile_sqla_query(
                         qry,
                         catalog=self.catalog,
diff --git a/superset/db_engine_specs/base.py b/superset/db_engine_specs/base.py
index 965eec46b12..82d26c2bd2b 100644
--- a/superset/db_engine_specs/base.py
+++ b/superset/db_engine_specs/base.py
@@ -526,6 +526,14 @@ class BaseEngineSpec:  # pylint: disable=too-many-public-methods
     force_column_alias_quotes = False
     arraysize = 0
     max_column_name_length: int | None = None
+
+    # Some databases (e.g. Druid, Pinot) build cursor.description by inspecting
+    # the values in the first returned row rather than from query-plan metadata.
+    # For those engines WHERE FALSE returns no rows and therefore leaves
+    # cursor.description as None, which breaks the adhoc column type probe.
+    # Set this to True on any engine spec where at least one row must be
+    # fetched for cursor.description to be populated.
+    type_probe_needs_row: bool = False
     try_remove_schema_from_table_name = True  # pylint: disable=invalid-name
     run_multiple_statements_as_one = False
     custom_errors: dict[
diff --git a/superset/db_engine_specs/druid.py b/superset/db_engine_specs/druid.py
index 453231c3910..f80c1e1d353 100644
--- a/superset/db_engine_specs/druid.py
+++ b/superset/db_engine_specs/druid.py
@@ -46,6 +46,10 @@ class DruidEngineSpec(BaseEngineSpec):
     allows_joins = is_feature_enabled("DRUID_JOINS")
     allows_subqueries = True
 
+    # pydruid builds cursor.description from the first returned row, so a
+    # WHERE FALSE query (zero rows) leaves description as None.
+    type_probe_needs_row = True
+
     metadata = {
         "description": (
             "Apache Druid is a high performance real-time analytics database."
diff --git a/superset/db_engine_specs/pinot.py b/superset/db_engine_specs/pinot.py
index 9e0b8a888da..7942df79efe 100644
--- a/superset/db_engine_specs/pinot.py
+++ b/superset/db_engine_specs/pinot.py
@@ -31,6 +31,10 @@ class PinotEngineSpec(BaseEngineSpec):
     allows_alias_in_select = False
     allows_alias_in_orderby = False
 
+    # pinotdb only sets cursor.description when the response contains
+    # columnDataTypes, which Pinot omits for zero-row results.
+    type_probe_needs_row = True
+
     metadata = {
         "description": "Apache Pinot is a real-time distributed OLAP datastore.",
         "logo": "apache-pinot.svg",
diff --git a/tests/unit_tests/models/helpers_test.py b/tests/unit_tests/models/helpers_test.py
index c0354844737..6a423d5f828 100644
--- a/tests/unit_tests/models/helpers_test.py
+++ b/tests/unit_tests/models/helpers_test.py
@@ -1856,3 +1856,112 @@ def test_extras_having_is_parenthesized(
         assert "(COUNT(*) > 0 OR 1 = 1)" in sql, (
             f"extras.having should be wrapped in parentheses. Generated SQL: {sql}"
         )
+
+
+def _run_probe(
+    database: Database,
+    type_probe_needs_row: bool = False,
+) -> str:
+    """
+    Run adhoc_column_to_sqla with a mocked get_columns_description and
+    return the SQL that was passed to it.
+
+    ``type_probe_needs_row`` is patched onto the database's real engine spec
+    class so every other method keeps working normally.
+    """
+    from unittest.mock import patch
+
+    from superset.connectors.sqla.models import SqlaTable, TableColumn
+
+    table = SqlaTable(
+        database=database,
+        schema=None,
+        table_name="t",
+        columns=[TableColumn(column_name="a", type="INTEGER")],
+    )
+
+    adhoc_col: AdhocColumn = {
+        "sqlExpression": "round(a / 50) * 50 / 1000",
+        "label": "Duration",
+        "columnType": "BASE_AXIS",
+        "timeGrain": "P1D",
+    }
+    captured: list[str] = []
+
+    def fake_get_columns_description(
+        db: object,
+        catalog: object,
+        schema: object,
+        sql: str,
+    ) -> list[dict[str, object]]:
+        captured.append(sql)
+        return [
+            {
+                "column_name": "Duration",
+                "name": "Duration",
+                "type": "FLOAT",
+                "is_dttm": False,
+            }
+        ]
+
+    spec_cls = database.db_engine_spec
+    with (
+        patch(
+            "superset.connectors.sqla.models.get_columns_description",
+            side_effect=fake_get_columns_description,
+        ),
+        patch.object(spec_cls, "type_probe_needs_row", type_probe_needs_row),
+    ):
+        result = table.adhoc_column_to_sqla(adhoc_col)
+
+    assert result is not None
+    assert len(captured) == 1, "Expected exactly one type-probe query"
+    return captured[0]
+
+
+def test_adhoc_column_type_probe_uses_where_false(database: Database) -> None:
+    """
+    Most engines populate cursor.description from query-plan metadata, so the
+    probe uses WHERE FALSE — zero rows read, no table scan.
+
+    SQLAlchemy renders sa.false() differently per dialect:
+      - Most databases:  WHERE false
+      - SQLite:          WHERE 0 = 1
+    """
+    probe_sql = _run_probe(database, type_probe_needs_row=False).lower()
+
+    always_false_patterns = ("where false", "where 0 = 1")
+    assert any(p in probe_sql for p in always_false_patterns), (
+        f"Probe query should use a WHERE false condition to avoid scanning the "
+        f"table, but got: {probe_sql}"
+    )
+    assert "limit" not in probe_sql, (
+        f"WHERE false probe must not contain LIMIT, but got: {probe_sql}"
+    )
+
+
+def test_adhoc_column_type_probe_uses_limit_1_for_row_dependent_engines(
+    database: Database,
+) -> None:
+    """
+    Druid and Pinot build cursor.description by inspecting the first returned
+    row. WHERE FALSE yields no rows, so description stays None. These engines
+    must fall back to LIMIT 1.
+    """
+    from superset.db_engine_specs.druid import DruidEngineSpec
+    from superset.db_engine_specs.pinot import PinotEngineSpec
+
+    for spec_cls in (DruidEngineSpec, PinotEngineSpec):
+        assert spec_cls.type_probe_needs_row is True, (
+            f"{spec_cls.__name__} must declare type_probe_needs_row = True"
+        )
+
+    probe_sql = _run_probe(database, type_probe_needs_row=True).lower()
+
+    assert "limit 1" in probe_sql, (
+        f"Row-dependent engines: probe should use LIMIT 1, got: {probe_sql}"
+    )
+    always_false_patterns = ("where false", "where 0 = 1")
+    assert not any(p in probe_sql for p in always_false_patterns), (
+        f"Row-dependent engines: probe must NOT use WHERE false, got: {probe_sql}"
+    )