They control the databases, business workflows, and cloud contracts that many large companies already run on, so more AI work can pull more spending through their stack. The risk is that power,
GPU, and funding bottlenecks slow the buildout, or that compute becomes too commodity-like to turn demand into great returns.