Deployment profiles
Depending on your workload's performance needs, your budget and how mission-critical is your app, you may choose from a variety of deployment profiles that will determine your final costs and infrastructure performance. Deployment profiles ultimately specify a required compute node affinity inside the decentralized infrastructure.
A deployment profile is bound to a Project. This means that you can organize all your resources and ensure that all components of your deployment will have the same infrastructure placement (e.g: all in the same datacenter or all on your personal nodes).
When you create or edit a project you will have the option to choose a different profile than the default one (HYBRID):

Profiles and their main characteristics
HPC
- High Performance Computing
- Powerful bare-metal servers, suitable for your mission-critical enterprise apps
- Prices may be higher than on regular deployment profiles
All resources in the current project will be deployed on the same datacenter.
GPU
- Powerful Graphics Processing Units
- Dedicated for AI/ML apps. Mandatory profile for training, fine-tuning or publishing AI models
- Prices are based on GPU compute rather than CPU
CLOUD
- Public cloud VMs
- Reliable Virtual Machines created on well-known public clouds
- Prices are usually considerably lower than the ones charged by competitors
HYBRID
- A mix between public cloud VMs and personal computers
- The best option if you want to balance high performance with small costs
- Prices are definitely considerably lower than the ones charged by competitors
EDGE
- Personal computers
- Less reliable compute units. Being the cheapest option, this profile is perfect for non mission-critical apps
- The cheapest compute power out there
MYNODE
- Your own StateMesh registered nodes
- If you are also a node operator on StateMesh you can opt for your own compute for reliability and cost reduction
- Everything you pay for this app goes directly to your wallet
If you move a resource into a different project which also has a different deployment profile configured, then you must re-deploy it.