A company uses high-concurrency AWS Lambda functions to process an increasing number of messages from a queue during marketing events. The Lambda functions run CPU-intensive code. The company wants to reduce compute costs while maintaining service latency. Which solution meets these requirements?

The correct solution involves configuring provisioned concurrency and increasing memory based on AWS Compute Optimizer recommendations. Provisioned concurrency keeps Lambda functions initialized and ready to respond instantly, significantly reducing cold start latency during high-concurrency events. Since the functions are CPU-intensive, increasing memory also allocates more vCPUs, which can reduce execution time and, consequently, overall compute costs for a given workload. AWS Compute Optimizer provides data-driven recommendations for optimal memory settings. Incorrect options: Decreasing memory for CPU-intensive functions would likely increase execution time and cost, and potentially lead to timeouts, worsening latency. Reserved concurrency prevents other functions from consuming all available concurrency but doesn't eliminate cold starts, which is crucial for maintaining latency during high-concurrency events.