A wave of new decentralized AI platforms and networks is emerging, aiming to challenge the dominance of major technology companies in artificial intelligence infrastructure and compute. Gaurav Sharma, CEO of io.net, advocates for a 'decentralized AI cloud' powered by global contributors, offering developers instant access to over 30,000 GPUs at costs reportedly 70% lower than traditional providers. Sharma notes that the combined AI and cloud market is projected to reach $2.5 trillion within seven years. Sharma also highlighted Apple's recent difficulties in securing enough GPUs to scale its AI efforts. AIOZ Network has launched AIOZ AI, a decentralized marketplace for AI models and datasets built entirely on DePIN (Decentralized Physical Infrastructure Networks). The platform enables developers and enterprises to upload, purchase, and monetize AI assets, with compute and storage tasks distributed across a global network of contributors. AIOZ DePIN reportedly includes over 200,000 contributors. The launch is accompanied by the AIOZ AI Challenge, a competition series aimed at expanding the platform's model and dataset library. Other initiatives include SingularityNET's integration of Filecoin via LighthouseWeb3 for decentralized data storage, Inference Grid's permissionless network for AI inference powered by Spark and the Lightning Network, and PublicAI's focus on edge data collection and training through DePIN_Union. Inference Grid offers realtime Bitcoin payments for compute, and supports access to both open source and proprietary large language models. Projects such as Pai3Ai and Spheron are also highlighted for their efforts to create community-driven, decentralized AI networks. Despite rapid growth and interest, some analysts note that current verifiable revenues for decentralized compute networks remain low relative to their valuations. Real adoption, they argue, will depend on achieving enterprise-grade reliability and consistent demand from large-scale users.