diff --git a/libs/pageserver_api/src/config.rs b/libs/pageserver_api/src/config.rs
index e49d15ba87a0..09cfbc55fd1c 100644
--- a/libs/pageserver_api/src/config.rs
+++ b/libs/pageserver_api/src/config.rs
@@ -442,7 +442,14 @@ impl Default for ConfigToml {
             tenant_config: TenantConfigToml::default(),
             no_sync: None,
             wal_receiver_protocol: DEFAULT_WAL_RECEIVER_PROTOCOL,
-            page_service_pipelining: PageServicePipeliningConfig::Serial,
+            page_service_pipelining: if !cfg!(test) {
+                PageServicePipeliningConfig::Serial
+            } else {
+                PageServicePipeliningConfig::Pipelined(PageServicePipeliningConfigPipelined {
+                    max_batch_size: NonZeroUsize::new(32).unwrap(),
+                    execution: PageServiceProtocolPipelinedExecutionStrategy::ConcurrentFutures,
+                })
+            },
         }
     }
 }
diff --git a/test_runner/fixtures/neon_fixtures.py b/test_runner/fixtures/neon_fixtures.py
index f55f06bebc00..f829855c3dca 100644
--- a/test_runner/fixtures/neon_fixtures.py
+++ b/test_runner/fixtures/neon_fixtures.py
@@ -1094,6 +1094,13 @@ def __init__(self, config: NeonEnvBuilder):
                 # Disable pageserver disk syncs in tests: when running tests concurrently, this avoids
                 # the pageserver taking a long time to start up due to syncfs flushing other tests' data
                 "no_sync": True,
+                # Batching (https://github.com/neondatabase/neon/issues/9377):
+                # enable batching by default in tests and benchmarks.
+                "page_service_pipelining": {
+                    "mode": "pipelined",
+                    "execution": "concurrent-futures",
+                    "max_batch_size": 32,
+                },
             }
             if self.pageserver_virtual_file_io_engine is not None:
                 ps_cfg["virtual_file_io_engine"] = self.pageserver_virtual_file_io_engine