Firestore - Why parallel writes/reads are faster in a http pipelined request?

Question

What is the fastest way to write a lot of documents to Firestore?

Here it says that with "individual sequential write operations" you can reach 10 document