Decentralized AI is not about decentralized compute; it's actually about decentralized & proprietary data. Once AI systems exhaust existing and potential public data sources, the frontier systems will need to scale by leveraging private data (think trade secrets, proprietary information, personal data, etc.). As proprietary & domain-specific data is fragmented across independent organizations, AI systems need to scale in a decentralized manner, mirroring the fragmentation of proprietary data. Decentralized AI ~= Interoperable AI systems empowered with proprietary data