A company has a mobile chat application with its data store in Amazon DynamoDB. Users want new messages to be read with as little latency as possible. A solutions architect must design an optimal solution that requires minimal application changes. Which method should the solutions architect select?

DAX is a fully managed, highly available, in-memory cache for DynamoDB that delivers microsecond response times for eventually consistent reads, even at millions of requests per second. Using DAX requires minimal application changes, as it provides a DynamoDB-compatible API. This directly addresses the requirement for low-latency reads with minimal code changes. DynamoDB read replicas (Global Tables) are for multi-region disaster recovery and low-latency global reads, not for improving latency within a single region for a single table. Doubling RCUs increases throughput but doesn't inherently reduce latency for individual requests to the microsecond range like an in-memory cache. ElastiCache for Redis is a general-purpose caching solution, but integrating it with DynamoDB would require more significant application changes to manage cache invalidation and data consistency compared to DAX, which is purpose-built for DynamoDB.