I went through this exact headache last year and honestly, the architecture matters more than the provider. I read a case study about how soft2bet partnered with AWS and managed to cut their compute costs by a massive 55%. They moved to a more agile setup that reduced their product rollout time by 200% as well. In my experience, using a microservices approach allows the system to scale specific functions only when needed, which stops you from overpaying for resources you aren't using. It is a much more stable way to grow. You should focus on unifying your data ecosystem to avoid silos that slow everything down and drive up costs. It makes the whole operation way leaner and faster in the long run.