Why does folding dataframes cause a NullPointerException?

Question

The code below is close to what I'm trying to do. I have an RDD of dataframes and I want to append those dataframes to make up 1 dataframe in the end.

import org.apache.spark.sql.SparkSession

//scalastyle:off
object test{
  def main(args: Array[String]): Unit = {
    val spark = SparkSession.builder()
      .appName("Ingester")
      .master("local[2]")
      .getOrCreate()

    val rdd = spark.sparkContext.parallelize(List(spark.emptyDataFrame, spark.emptyDataFrame))

    rdd
      .fold(spark.emptyDataFrame)(_.union(_))

  }
}

For some reason I get an NPE when trying to do this.

Caused by: java.lang.NullPointerException
    at org.apache.spark.sql.Dataset$.ofRows(Dataset.scala:73)
    at org.apache.spark.sql.Dataset.withSetOperator(Dataset.scala:3313)
    at org.apache.spark.sql.Dataset.union(Dataset.scala:1834)
    at test$$anonfun$main$1.apply(test.scala:14)
    at test$$anonfun$main$1.apply(test.scala:14)
    at scala.collection.TraversableOnce$$anonfun$foldLeft$1.apply(TraversableOnce.scala:155)
    at scala.collection.TraversableOnce$$anonfun$foldLeft$1.apply(TraversableOnce.scala:155)
    at scala.collection.Iterator$class.foreach(Iterator.scala:742)
    at org.apache.spark.InterruptibleIterator.foreach(InterruptibleIterator.scala:28)
    at scala.collection.TraversableOnce$class.foldLeft(TraversableOnce.scala:155)
    at org.apache.spark.InterruptibleIterator.foldLeft(InterruptibleIterator.scala:28)
    at scala.collection.TraversableOnce$class.fold(TraversableOnce.scala:210)
    at org.apache.spark.InterruptibleIterator.fold(InterruptibleIterator.scala:28)
    at org.apache.spark.rdd.RDD$$anonfun$fold$1$$anonfun$19.apply(RDD.scala:1096)
    at org.apache.spark.rdd.RDD$$anonfun$fold$1$$anonfun$19.apply(RDD.scala:1096)
    at org.apache.spark.SparkContext$$anonfun$33.apply(SparkContext.scala:2130)
    at org.apache.spark.SparkContext$$anonfun$33.apply(SparkContext.scala:2130)
    at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
    at org.apache.spark.scheduler.Task.run(Task.scala:109)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:345)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
    at java.lang.Thread.run(Thread.java:748)

Does anybody know why this happens?