Lots of these services: - Fluidstack: https://fluidstack.io https://fluidstack.io - Vast: https://vast.ai https://vast.ai - QBlocks: https://qblocks.cloud ht

Lots of these services: - Fluidstack: https://fluidstack.io https://fluidstack.io - Vast: https://vast.ai https://vast.ai - QBlocks: https://qblocks.cloud https://qblocks.cloud - RunPod: https://runpod.io https://runpod.io - Sonm: https://sonm.com https://sonm.com (blockchain) - Golem: https://golem.network https://golem.network (blockchain) - Rentaflop: https://rentaflop.com https://rentaflop.com (rendering specific, blockchain) - RNDR: https://rendertoken.com https://rendertoken.com (rendering specific, blockchain) If you want HPC specific cloud providers: - Crusoe Cloud: https://crusoecloud.com https://crusoecloud.com - Coreweave: https://coreweave.com https://coreweave.com - Lambda Labs: https://lambdalabs.com https://lambdalabs.com - Paperspace: https://paperspace.com https://paperspace.com As others have pointed out, the decentralized clouds can't offer high performance interconnects (e.g. InfiniBand) that a lot of folks are using for LLM training. There are definitely initiatives underway to reduce dependence on these interconnects and build performant distributed training (again, some threads below mention this), but I think it's mostly academic at this point. Disclosure: I run product at Crusoe Cloud, which aims to provide ML training at half the cost of a hyperscaler, while also being carbon reducing (https://crusoecloud.com/climate-impact/ https://crusoecloud.com/climate-impact/).