diff --git a/spark/src/main/scala/org/apache/spark/sql/comet/execution/shuffle/CometNativeShuffleWriter.scala b/spark/src/main/scala/org/apache/spark/sql/comet/execution/shuffle/CometNativeShuffleWriter.scala
index 3fc222bd19..d704d3fd88 100644
--- a/spark/src/main/scala/org/apache/spark/sql/comet/execution/shuffle/CometNativeShuffleWriter.scala
+++ b/spark/src/main/scala/org/apache/spark/sql/comet/execution/shuffle/CometNativeShuffleWriter.scala
@@ -191,7 +191,7 @@ class CometNativeShuffleWriter[K, V](
       shuffleWriterBuilder.setCompressionLevel(
         CometConf.COMET_EXEC_SHUFFLE_COMPRESSION_ZSTD_LEVEL.get)
       shuffleWriterBuilder.setWriteBufferSize(
-        CometConf.COMET_SHUFFLE_WRITE_BUFFER_SIZE.get().max(Int.MaxValue).toInt)
+        CometConf.COMET_SHUFFLE_WRITE_BUFFER_SIZE.get().min(Int.MaxValue).toInt)
 
       outputPartitioning match {
         case p if isSinglePartitioning(p) =>