Develop and deploy AI models

Oraichain’s GPU infrastructure will provide several options for AI developers to develop and deploy AI models on the Distilled AI Layer and GPU infra. They can run their code directly within Jupyter Notebooks or deploy their AI models using pre-configured Helm charts with team-managed Kubernetes clusters. Additionally, users can create on-demand Ray clusters for distributed computing on decentralized GPU grids, allowing them to scale their workloads as needed. For coding and development, they can utilize Visual Studio Code (VSCode) or Jupyter Notebook to run their scripts and manage their projects efficiently. This flexibility ensures users can leverage their preferred tools and workflows to effectively develop and deploy their AI models.

Credit System

The credit GPU usage system allows users to earn credits through various methods, such as staking ORAI, purchasing credits with ORAI, or actively participating in the development of the Oraichain ecosystem. These credits are stored securely in OWallet and can be redeemed to access GPU resources for training and inferencing AI models.

Global GPU infrastructure & monetization

We’ve experienced a steady increase in GPU demand on our infrastructure and we expect to see this accelerate more rapidly as our Distilled AI layer is released. Alongside the GPU resources managed by the team and partners, users will also have access to a decentralized GPU network spanning global regions. GPU owners looking to monetize their idle GPU resources can easily register their compute resources into the network. End users who need to utilize GPUs can create on-demand clusters in a permissionless manner to run or deploy their AI models. This approach fosters a collaborative ecosystem where resources are dynamically allocated and efficiently utilized, ensuring widespread availability and scalability of GPU power for AI tasks.

GPU status (as of September 2024)

• Total GPU cards: 38
• Total VRAM: 1896 GB
• Usage: 60~80%