Elasticsearch (via Lucene) compresses all the data it writes to disk, but snapshots are not compressed any further than regular indices. I've heard anecdotal reports that some additional savings are possible in some circumstances.
Be aware that compressed data can be much harder to recover if it suffers even a tiny amount of corruption, which is a big risk in long term archival. See e.g. this article:
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.