Distributed cloud GPU network project Aethir and Theta EdgeCloud, the hybrid cloud computing platform powered by Theta Edge Network, are preparing to launch the largest hybrid GPU marketplace for developers and enterprises. The collaboration aims to provide unprecedented global GPU compute capabilities, with the potential to deliver 20-30 times greater power than any comparable network in the industry today, according to people familiar with the matter.
Theta Network, originally focused on decentralized video delivery, has been gradually integrating AI and video processing capabilities into its platform. The pivot towards AI was notably highlighted with the announcement of the EdgeCloud platform, which is designed to provide users with “unlimited access to GPU processing power for any AI or video task,” essentially offering a decentralized cloud computing infrastructure.
The introduction of Theta EdgeCloud marks a significant step in Theta Network’s evolution, combining the benefits of edge and cloud computing to support immersive extended reality experiences, video-based AI, and other cutting-edge technologies.
Theta Network’s pivot from focusing solely on video to incorporating AI and video processing capabilities aims to enhance its platform’s functionality and offer more sophisticated and on-trend services, thereby contributing to the development of next-generation applications.
Decentralized computing power for AI
Industries utilizing AI require significant amounts of GPU-based compute to innovate and scale, but supply chain issues, shortages, and inefficiencies in traditional cloud computing models have hindered progress. Aethir and Theta’s marketplace seeks to address this challenge by enabling instant access to enterprise-grade GPU compute, empowering organizations to train, fine-tune, and deploy open-source or custom AI models with ease.
Aethir brings its network of over 40,000 enterprise-grade GPUs, including more than 8,000 NVIDIA H100s, to the partnership. This represents a total of 170,000 TFLOPS of compute power. Meanwhile, Theta’s Edge Network comprises one of the world’s largest clusters of distributed GPU computing power, with a combined total of approximately 80 PetaFLOPS, equivalent to 250 Nvidia A100s.
By shifting from centralized to decentralized network infrastructure, Aethir and Theta aim to reduce latency and provide best-in-class GPU resources to end-users, regardless of their physical location. The decentralized model offers superior GPU efficiency by pooling power from multiple GPUs, maximizing usage, and reducing costs for clients.
Aethir has already reportedly achieved a high level of community engagement and decentralization through the sale of over 64,000 Aethir Checker nodes to more than 11,000 individual buyers. These nodes ensure the stability and security of Aethir’s distributed cloud infrastructure, with node operators set to receive rewards in the form of $ATH tokens following the platform’s Mainnet launch in Q2 2024.
Theta plans to launch its hybrid cloud-decentralized GPU marketplace fully integrated into the EdgeCloud platform later this year, providing developers and enterprises with a user-friendly option to leverage AI models. Theta Labs CEO Mitchell Liu emphasized the potential of the EdgeCloud launch and subsequent releases to “unleash a new hybrid cloud-decentralized paradigm in AI computing,” leveraging the company’s recently approved patent on “Edge Computing Platform supported by Smart Contract enabled blockchain network.”
Aethir CTO Kyle Okamoto highlighted the importance of combining Theta EdgeCloud’s global network with Aethir’s processing power, stating that the partnership holds “the keys to global GPU compute, arguably the most valuable and most disruptive asset in history.” The collaboration aims to meet the increasing computational demands of emerging generative AI models, such as text-to-video, text-to-3D, and sketch-to-3D, which will purportedly require 10-100 times more power than current chatbots.
The partnership between Aethir and Theta EdgeCloud marks a significant step forward in the development of decentralized GPU compute infrastructure, with the potential to revolutionize the AI and gaming industries by providing unparalleled access to enterprise-grade resources on a global scale.
Further, decentralized infrastructure may become increasingly important for the security of those who do not wish to see centralized entities hold all the cards with respect to AGI (artificial general intelligence) when it finally arrives.
Read the full article here
