A stateless application must run on VMs and automatically scale with traffic. Which deployment should you use?

For a stateless application running on VMs that needs to scale with traffic, a Compute Engine managed instance group (MIG) with autoscaling is the most direct and appropriate solution. MIGs allow you to deploy and manage a group of identical VMs, and autoscaling automatically adjusts the number of instances based on demand. While Google Kubernetes Engine (GKE) can run stateless applications and scale, the question specifically mentions "VMs" as the deployment target, and GKE introduces an orchestration layer that isn't strictly necessary if the application simply needs VM-level autoscaling. Cloud Functions and Cloud Run are serverless platforms, not VM-based deployments, making them unsuitable for the requirement of running on VMs.