Vector databases can consume 10x to 20x more storage due to index expansion.

When designing RAG architectures or AI solutions, do you factor in storage requirements as part of your decision-making?