Ankr is introducing its upcoming decentralized AI computing network called Neura.

The blockchain infrastructure development company Ankr announced the launch of the testnet for the blockchain network Neura, specifically designed for GPU resource allocation for artificial intelligence and machine learning. The ANKR token will be used as the native cryptocurrency of Neura, with plans to launch the Neura mainnet later this year.

What Problems Does Neura Aim to Solve?

Neura's goal is to enable developers to build decentralized products that combine AI, cloud computing, and blockchain technology. Specifically, Neura will focus on assisting startups in need of machine learning computing resources by providing funding, accessing GPU resources, and storing data.

Scarcity of Computing Resources

In recent years, GPUs used for model computations have been in short supply due to changes in industry trends, creating a seller's market for cloud computing providers. This has led to increased costs for development teams training AI models, with limited customization options. For more detailed information on the development dilemma of computing resources, please refer to the following article.

Lack of Revenue Models for Models

In addition to technical barriers related to infrastructure scalability, AI and cloud computing projects also need to consider how to build sustainable revenue-generating business models.

Traditional Web2 platforms and processes often do not fully support strategies for monetizing training data, which is crucial for the survival and development of AI projects. If models are open-sourced, they cannot generate revenue, but if they are not, they face stricter regulatory scrutiny and challenges in obtaining training data. This has led to an economic dilemma in the development of machine learning models.

Recommended Read: ETHTaipei | How Blockchain Verifies the Authenticity of AI Information
Reason for Recommendation: This article addresses the two major dilemmas in the AI industry, "opaque information verification" and "unprofitable open-source models," and proposes the ERC-7641 revenue-sharing token as a solution, which aligns with the issues discussed in this article.

Introduction to Neura's Design Concept

Neura Infrastructure

Neura is a blockchain built on the Cosmos SDK framework, which is compatible with EVM. It facilitates the registration of GPU providers, registration of users in need of computational services, and the distribution of ANKR tokens based on computational results.

Additionally, by integrating storage solutions such as Celestia, EigenLayer, IPFS, and Arweave, Neura stores AI model data and GPU provider computational results off-chain. Most AI computations occur off-chain on GPU provider servers, while data storage is managed on infrastructure like Celestia. Neura's on-chain activities focus on incentivizing models.

Neura Operating Process

In the Neura platform, the operating process is as follows:

1. When a task provider, such as a team in need of computing resources, submits a task, such as a demand for AI model computations, the task provider can choose the most suitable GPU provider and obtain their public key.
2. The task provider encrypts the task content with the GPU provider's public key and submits the encrypted task content as a transaction to off-chain storage facilities like Arweave, EigenDA.
3. The task owner creates a new transaction on the Neura network through RPC, including the ID of the previous transaction, the ID of the selected GPU provider, the duration the task should be maintained, and specific requirements for GPU, RAM, and CPU.
4. Once the transaction content is confirmed by validators on the Neura network, the corresponding smart contract on Neura is updated.
5. The selected GPU provider detects relevant transactions by monitoring the contract, validates them, retrieves the task from off-chain storage facilities, and executes it.
6. After completing the task, the GPU provider uploads the results as transaction content to off-chain storage facilities. The results may be unencrypted, depending on the task provider's privacy and confidentiality requirements.
7. The GPU provider creates another transaction on Neura to update the task status and provide a reference ID for the results, which is confirmed by validators on the Neura network.

Neura Token Economics

Adopting ANKR Tokens

Neura will adopt ANKR tokens as the native token. Once the Neura mainnet is live, the Ankr team will provide cross-chain bridges for holders to transfer ANKR tokens to the Neura blockchain.

ANKR tokens on the Neura network will serve as payment for GPU resources, participation in IMOs, and rewards for GPU providers.

Revenue Sharing Model Mechanism

In the Neura network, task owners can issue ERC-7641 tokens to share the income generated from creating AI models. These tokens allow holders to receive a portion of the generated income, ensuring a fair reward distribution system for contributors and investors.

Additionally, ERC-404 will be introduced to tokenize model ownership as NFTs shared among participants, facilitating project initial token issuance through crowdfunding by task providers.

How does the Pandora project use ERC404 to bring NFTs into Uniswap's trading pool?

The team states that these mechanisms leverage Neura's blockchain transparency to cultivate bilateral trust and establish a sustainable computational ecosystem.

AI Cloud Computing Competition is Fierce

In addition to Neura, projects like io.net on Solana, Akash Network, and even Near have entered the competitive arena, with decentralized computing infrastructure flourishing. However, competition is intensifying. The entry of resource-constrained Ankr into the fray is expected to heighten industry competition.

Alliance of AI Coins? Rumors suggest SingularityNET, Fetch.ai, and Ocean are considering a merger to focus on AI development.

Given the limited familiarity of traditional Web2 services with the crypto ecosystem's development, the challenge for decentralized GPU computing platforms like Neura will be how to compete with traditional cloud computing platform players like AWS and Google Cloud to attract teams outside the industry with computing needs. This will pose a greater challenge for Neura and other decentralized GPU computing platforms.