Date: 11/18/2025
Okay, so this video by Simon L is all about moving away from the cloud for SaaS infrastructure and embracing a self-hosted setup. He dives into his reasons, which are primarily cost optimization, control, and avoiding vendor lock-in. He outlines his own self-hosted setup and also clarifies what he still uses AWS for. He’s not advocating a complete cloud abandonment, but rather a strategic shift.
For someone like me, deep into AI-driven development, this is gold. We’re constantly looking at ways to optimize infrastructure, and the cloud, while powerful, can be a black hole for resources. The insights into cost savings and greater control are particularly relevant. I’m especially interested in the points made regarding data ownership and the ability to fine-tune the environment for specific AI/ML workloads, something that can get expensive and restrictive in a purely cloud-based setting.
Think about it: with the rise of on-prem LLMs and tools like Ollama, the idea of running some AI components locally is becoming more feasible. We could potentially build a hybrid setup – using the cloud for scalability and globally distributed services, but keeping the AI “brain” closer to home for data privacy and performance. It’s absolutely worth experimenting with because it challenges the default “everything in the cloud” mentality. Plus, the thought of optimizing infrastructure costs while also gaining greater control is something every developer should be exploring. I’m keen to see if moving some workloads in-house gives more granular control over GPU usage and can speed up development cycles.