They help customers get dense AI systems online fast by bundling servers, cooling, racks, and deployment, so they control a real AI bottleneck. The risk is that larger OEMs, ODMs, or
hyperscalers make that integration look ordinary, leaving them with big revenue but thin margins.