We should look for ways to minimize the cost of the metal node used for CI. That might include using spot instance types or running in a region where prices are lower. We should coordinate with the test platform team so we have an accurate understanding of what different VM types cost Red Hat.