Stempoint is a distributed artificial intelligence (AI) infrastructure platform designed to provide a unified environment for AI development by integrating access to AI models with a decentralized network of GPU computing resources. The project aims to serve as a comprehensive solution for developers and enterprises by combining a multi-model API with on-demand computational power. ^{[1] [2]}

Overview

Stempoint is being developed to address challenges in the AI industry, such as fragmented access to various foundational AI models and the high cost or scarcity of high-performance GPU computing power. The platform's stated goal is to create a single, globally oriented interface for AI model training, fine-tuning, and inference. It seeks to achieve this by marshaling underutilized GPU resources from a global network of individual and institutional providers, making them accessible on demand. The project's official X (formerly Twitter) account was established in January 2025. ^[3]

The platform's core design combines two primary components: an AI Agent Aggregation Layer and a hybrid compute infrastructure. The aggregation layer functions as a unified gateway, simplifying how developers interact with a wide range of leading AI models. The compute infrastructure utilizes a decentralized physical infrastructure network (DePIN) alongside a centralized elastic cloud to supply the necessary processing power for intensive AI tasks. This hybrid approach is intended to offer flexibility, scalability, and cost optimization for users. ^[4]

The entire ecosystem is built around a native utility token, SPT, which underpins the platform's economy. The token is designed to facilitate a closed-loop value system where developers pay for compute resources, and providers are rewarded for contributing their hardware. This model is intended to create a self-sustaining and participatory ecosystem where token holders can also engage in governance to influence the platform's future development. ^[5]

Architecture

The Stempoint system architecture is structured as a multi-layered framework designed to manage the flow of information and computational tasks between developers and the underlying hardware resources. Each layer performs a distinct function to create an integrated environment for AI development and deployment.

The layers of the architecture include:

• Developer Dashboard and API Gateway: This serves as the primary user interface and entry point for developers. It provides a console for managing projects, monitoring resource usage, and accessing the platform's unified API.
• API Mesh & Router: Positioned behind the gateway, this layer is responsible for managing incoming requests. It handles dynamic scheduling of tasks, routing them to the appropriate AI model or compute resource. It also oversees the token-based billing system and logs API calls for analytics and accounting.
• AI Agent Aggregation Layer: This layer packages and standardizes access to multiple third-party AI foundational models. It includes a translation middleware component to handle multi-language requests and is designed to support context persistence, which is necessary for managing multi-turn dialogues in advanced AI agent applications.
• Compute Layer (Hash Forest & DePIN Node Internet): This is the core infrastructure layer that provides computational power. It consists of two interconnected parts: the Hash Forest, which is a centralized GPU scheduling center, and the DePIN Node Internet, the decentralized network of third-party compute nodes. The API Mesh & Router dynamically allocates tasks to one or both of these components based on factors like cost, performance requirements, and resource availability.
• Distributed Compute & Sharding Layer: This layer manages the execution of large-scale computational tasks. It is designed to support distributed multi-threaded execution and sharded training, which involves breaking down large AI models or datasets into smaller parts that can be processed in parallel across multiple nodes in the network.
• Token Layer: The foundational economic layer that underpins the entire system. It utilizes the native SPT token to manage all transactions, including payments for compute services, distribution of mining rewards to node providers, and staking for governance participation.

This layered structure is intended to create a full-stack, integrated platform that connects AI model access with the necessary computing power in a secure and efficient manner. ^[2]

Products

Stempoint's platform is composed of several core products that deliver its unified AI infrastructure services. These products are divided between the AI model access layer and the computational resource layer.

AI Agent Aggregation Layer

This service layer functions as a unified gateway that provides developers with a single Application Programming Interface (API) to access a variety of major AI foundational models. The platform plans to integrate models from providers such as OpenAI, Anthropic (Claude), and Google (Gemini). This is designed to eliminate the need for developers to manage multiple integrations, API keys, and billing systems. The layer incorporates an intelligent routing system that aims to dynamically select the optimal model service based on real-time metrics like latency and availability. A key feature of this layer is its support for context memory and multi-turn dialogue persistence, enabling the development of more complex and stateful AI agents. Additionally, it allows developers to package their own custom algorithms into callable APIs, complete with a built-in pay-per-use billing mechanism. ^{[1] [2]}

Compute Network

Stempoint's compute infrastructure is a hybrid system that combines centralized and decentralized resources to provide on-demand GPU power for AI training and inference.

Hash Forest

Hash Forest is described as the platform's centralized, elastic GPU cloud service. It is designed to provide reliable, on-demand access to a range of high-performance GPUs, including NVIDIA's A100, H100, and RTX 4090 models. The service offers flexible billing options, with resources available at hourly or minute-level increments to optimize costs for users. Hash Forest includes an intelligent scheduling engine to efficiently allocate resources and supports development tools like Notebook for interactive debugging. It also aims to provide functionality for the rapid deployment of inference microservices, allowing developers to quickly scale their applications. ^[2]

DePIN Node Internet

The DePIN Node Internet is a decentralized computing network that sources GPU power from a global pool of contributors. Individuals, data centers, and other organizations can connect their idle GPU hardware to the network by deploying a containerized client. The platform's scheduler then allocates inference or training tasks to these nodes based on a reputation system that considers metrics such as hardware stability, response speed, and historical performance. To ensure security, the network architecture supports container isolation to keep user tasks separate and data encryption during transit. For tasks requiring verifiable computation, the platform has an option to incorporate zero-knowledge (ZK) proof verification to ensure that results are trustworthy without revealing the underlying data. ^{[4] [2]}

Features

The Stempoint platform is being developed with several features intended to create a comprehensive and user-friendly AI infrastructure. It aims to provide unified access to a diverse range of AI models and compute resources through a single API, which is designed to minimize integration complexity and development time. The hybrid compute model, which combines an elastic cloud with a decentralized network, is intended to deliver elastic compute capabilities, allowing resources to scale dynamically with demand while helping to optimize costs for users. The use of a decentralized GPU network is a core feature, designed to tap into a global pool of underutilized computing power, thereby increasing resource availability and resilience.

The platform architecture includes multi-tenant environment isolation to ensure data security and the independent execution of tasks for different users. It is built for scalability through a permissionless node participation model, which allows the DePIN network to expand as more providers connect their hardware. The system also aims for high compatibility by supporting mainstream model frameworks and enabling cross-platform deployment. A clear revenue model based on pay-per-use for compute consumption and API calls is central to its design, alongside a governance framework intended to foster a sustainable and community-driven ecosystem. The platform also plans to incorporate multi-language recognition and translation routing capabilities to support a global user base. ^{[1] [2]}

Ecosystem

The Stempoint ecosystem is designed as a closed-loop economy that connects several key participants: AI developers, enterprises, computing power providers, and Web3.0 ecosystem participants. AI developers can utilize the platform's unified API to access multiple large language models, reducing technical overhead and simplifying the development of complex AI applications. They can consume elastic compute resources from either the Hash Forest or the DePIN network for model training, fine-tuning, and inference, paying for these services with the SPT token.

Enterprises and larger organizations can use the platform to procure predictable GPU resources for large-scale training tasks or to integrate multi-model AI services into their products, thereby reducing vendor management complexity. On the supply side, computing power node providers—ranging from individuals with consumer-grade GPUs to data centers with enterprise-level hardware—can monetize their idle resources by connecting to the DePIN Node Internet. These providers earn SPT tokens as rewards for successfully completing computational tasks assigned to them by the network. Finally, Web3.0 ecosystem participants can engage with the platform by holding or staking the SPT token, which grants them the ability to participate in governance and vote on proposals that shape the platform's future. This model is intended to create a self-sustaining cycle where the demand for AI applications drives the need for compute power, which in turn incentivizes providers to supply it. ^{[5] [2]}

Use Cases

The Stempoint platform is designed to support a variety of use cases within the artificial intelligence and decentralized computing sectors.

• AI Model Training: High-performance, distributed training of large-scale AI models using the combined power of the Hash Forest and the DePIN network.
• AI Model Inference: Running scalable and cost-effective inference tasks for deployed AI applications that require real-time or batch processing.
• AI Agent Development: Building complex AI agents that require persistent context memory and access to multiple foundational models through a single, unified interface.
• Algorithm Monetization: Wrapping custom or proprietary algorithms into callable, pay-per-use APIs that can be offered to other developers on the platform.
• GPU Hardware Monetization: Enabling individuals and data centers to earn passive income by contributing their underutilized GPU hardware to the DePIN Node Internet.
• Research and Development: Providing researchers and developers with on-demand access to powerful computing resources for experimenting with and fine-tuning custom AI models.

These use cases are supported by the platform's integrated access to models and flexible compute resources. ^{[1] [2]}

Tokenomics

The Stempoint ecosystem is powered by its native utility and governance token, SPT. The token is integral to the platform's operations, creating the economic incentives that connect compute resource providers with AI application developers.

Token Utilities

The SPT token is designed with several primary functions within the platform:

• Compute Payments: Developers and enterprises use SPT to pay for GPU resources consumed during AI model training and inference on both the Hash Forest and the DePIN network, as well as for API calls to the AI Agent Aggregation Layer.
• Node Rewards: Compute providers who contribute their GPU hardware to the DePIN Node Internet are rewarded with SPT tokens for successfully completing tasks. This serves as the primary incentive for expanding the network's capacity.
• Governance: SPT token holders can stake their tokens to participate in the platform's governance. This allows them to vote on proposals related to the platform's development, policies, and future incentive structures.

Token Allocation

• Information regarding the specific token allocation percentages, total supply, and distribution schedule for the SPT token is not available in the publicly released documentation.

Governance

The governance model for Stempoint is intended to allow stakeholders to influence the platform's strategic direction. By staking SPT tokens, holders gain voting rights on key proposals. This mechanism is designed to guide decisions related to computing power incentives, model integration priorities, and other platform parameters. The goal is to create a decentralized and participatory ecosystem where the community has a voice in the project's long-term evolution. ^{[4] [2]}

Partnerships

• Based on available information, Stempoint has not announced any formal partnerships. The platform's documentation notes planned integration with models from providers like OpenAI, Claude, and Gemini, but this does not constitute a formal partnership with these entities. ^[2]