AI Made in Cosmos
summary
Cosmos panel lays out concrete work: Akash runs permissionless GPU rental and claims it already hosts distributed training efforts (examples cited include a 10 billion parameter model and ambitions toward 100+ billion), Aran Network builds an execution layer and interchain agent platform for on‑chain sovereignty, Cyber develops a bootloader and CyberTensor to plug into distributed training, and multiple projects name proofs for inference and ZK verification as the biggest unsolved bottlenecks; bandwidth, power (50 MW→250 MW→1.2 GW), data labeling and fractionalized model ownership remain the practical hurdles.
AI-written summary · about
chapters (8)
- 0:00AI in Cosmos ecosystem
- 2:51Cosmos for decentralized AI
- 8:31Open source AI necessity
- 15:56AI orchestration platforms
- 19:48Distributed AI training
- 22:45AI agents state and evolution
- 25:36Decentralized AI tech stack future
- 35:00Closing remarks & Q&A
related talks
49mThe Role AI & Crypto Play in Consumer Facing Applications
16mAI Without Limits: How ZeroG is Building the Future of Decentralized AI
16mBuild an Onchain AI dapp with 0G's Inferencing
34mOpen Source AI Agents on User-Owned Infra: K3s, MCP, & GPU Sharing in Practice
21mThe Cosmos Development Platform
02mStarcluster
22mKeynote
22mAI @Scale