A company moves workloads from public subnets to private subnets to improve security.

Instances in private subnets do not have direct routes to an internet gateway, so they cannot reach public internet endpoints unless the VPC provides a managed egress path. The standard AWS architecture for private-subnet outbound internet access is to use a NAT device (typically a NAT gateway) placed in a public subnet, with an Elastic IP address and a route from the private subnet to that NAT gateway for 0.0.0.0/0 traffic. The NAT gateway then forwards traffic to the internet through the VPC’s internet gateway while preventing unsolicited inbound connections to the private instances.

The scenario states that the VPC has a NAT gateway in each private subnet. That placement prevents proper egress because a NAT gateway itself must be in a public subnet with a route to the internet gateway to function for internet-bound traffic. Without that, private instances will fail to reach external services such as a third-party API, exactly as observed.

Option C corrects the architecture by deploying NAT gateways in the public subnets with Elastic IPs and updating the private subnet route tables so that outbound internet traffic targets the NAT gateways. This enables private workloads to initiate outbound connections while remaining unreachable from the public internet directly, maintaining the intended security posture.

Option A is incorrect because an outbound-only internet gateway is designed for IPv6 egress, not IPv4 internet access. Option B misunderstands routing: API Gateway is an application-layer proxy service and is not a target for VPC route tables to provide generic internet access. Option D is not applicable because VPC interface endpoints provide private connectivity to supported AWS services via PrivateLink, not to arbitrary external public APIs.

Therefore, deploying NAT gateways in the public subnets and routing private subnet egress through them is the correct solution.