Google spiders takes a snapshot of each webpage they index. The cached version is what Google uses to judge if a page is a good match for your query.