On May 7th, Bithumb added Korean won trading pairs for two AI projects, AIOZ and NEAR. While NEAR is a well-known L1 project, AIOZ Network might seem unfamiliar. Previously focused on storage and streaming media, AIOZ Network is now gradually converging towards AI as a service and shared computing power, leveraging its accumulated advantages. Recently, it released the whitepaper for its decentralized AI project, W3AI.

In the increasingly competitive AI arena, what new strategies can established projects offer to secure a position in a market where liquidity and attention are both scarce?

Due to the complexity of the whitepaper, Deep Tide TechFlow conducted thorough research on its content to help readers quickly understand the technical features and implementation of the AIOZ W3AI project.

Under the Wave, AIOZ’s Entry into the AI Market Opportunities

AIOZ is not a new project, but its transition to AI seems logical.

Previously, AIOZ Network operated as a Layer-1 network with interoperability between Ethereum and Cosmos. It utilized the AIOZ DePIN, driven by over 120,000 global nodes, to provide computational resources. This setup supports AI processing speed, rapid iteration, scalability, and network security, serving as the foundation for the project’s narrative shift.

Moreover, in the broader context, the development of AI faces challenges with centralized cloud computing solutions struggling to handle large volumes of data. This limitation leads to scalability issues and high usage costs. Additionally, concerns arise regarding data privacy and security when control lies with centralized providers rather than users.

Furthermore, accessing top-tier AI resources may be difficult, limiting the participation of small enterprises and individuals and impeding innovation. Edge computing offers a solution by providing near-end services for data sources. Applications initiate on the edge, resulting in faster network service responses. Since data processing occurs locally at nodes, eliminating the need for long-distance transmission to central servers, edge computing naturally reduces the risk of data breaches. With AIOZ DePIN’s globally distributed edge computing nodes, AIOZ gains substantial confidence in entering the AI domain at scale.

AIOZ Network currently operates node data.

W3AI: DePIN + AI as a Service Dual-layer Architecture

In its move towards the AI arena, AIOZ’s pivotal step is W3AI — a dual-layer architecture encompassing both infrastructure and applications.

The dual-layer architecture is at the core of the AIOZ W3AI project, offering an innovative solution to fundamental issues in AI computation, such as scalability, cost efficiency, and user privacy protection.

This architectural design divides the network’s operation into two main layers: the infrastructure layer (W3AI Infrastructure) and the application layer (W3AI Application). Each layer has unique functions and roles, collectively supporting the efficient operation of the entire network.

Infrastructure Layer (W3AI Infrastructure) as the Network Foundation

AIOZ DePIN: Globally distributed artificial nodes

The foundation of AIOZ W3AI lies in its vast distributed artificial edge computing nodes. These globally distributed nodes contribute computing resources, including storage, CPU, and GPU, forming a decentralized power source. The multigraph topology ensures efficient communication routes between AIOZ DePIN, minimizing communication costs and enhancing processing speed. These nodes collaborate through distributed computing methods to train and execute AI models collectively. Through this approach, the AIOZ W3AI platform effectively utilizes dispersed computing resources to reduce costs, enhance efficiency for AI applications, and bolster data privacy protection. This decentralized approach significantly reduces the risk of server bottlenecks and strengthens user privacy by eliminating single-point control.

The decentralized computing infrastructure of W3AI, is driven by the AIOZ node network. The purple areas represent the distribution of storage nodes, while the blue areas represent the distribution of computing nodes.

Data Processing and Storage

Through AIOZ W3S, data is securely stored on multiple globally dispersed nodes, enhancing data security while also improving the responsiveness of data processing.

The use of distributed file systems like AIOZ IPFS and encryption technologies protects the data stored on nodes, preventing unauthorized access and data leaks.

Flexible Application Layer (W3AI Application)

Web 3 AI platform provides AI as a service.

AI as a Service (AIaaS) refers to the model where AI technology is provided as an online service to users, allowing enterprises or individuals to enjoy the benefits of AI technology without high costs.

Imagine an e-commerce merchant wanting to understand user purchase history and analyze user consumption behavior to provide personalized shopping recommendations. AI technology can be used to collect and analyze user data, generating corresponding sales strategies. This is the application of AI as a service in e-commerce.

In terms of product form, W3AI provides a simplified AI training workflow and intuitive UI/UX, offering user interfaces and APIs that enable developers to easily access W3AI services, and develop, and deploy AI models, among other tasks. This layer’s design focuses on user experience and service accessibility. Additionally, the platform integrates various AI-as-a-service offerings, including machine learning, deep learning, and neural networks, allowing users to choose different services and tools as needed.

Model Training and Inference

The W3AI platform supports model training and inference in a decentralized environment. W3AI training (AIOZ W3AI Infrastructure) utilizes decentralized federated learning and homomorphic encryption technologies, enabling numerous edge computing nodes (DePINs) to collaborate on training AI models without sharing their own data. This improves model training performance while also ensuring data privacy. Trained models can be run on edge AIOZ DePINs, bringing AI closer to the data source. Supported by W3S technology, W3AI inference (AIOZ W3S Infrastructure) allows users to upload their own datasets for model training or use existing models on the platform for data analysis and prediction.

Decentralized W3AI Market and Incentive Mechanism

The application layer also provides users with decentralized markets, such as the AIOZ AI dApp Store and AI Model & Dataset Marketplace. Individual users and business organizations can freely contribute, sell AI datasets and models, build and deploy innovative AI applications, and convert their contributions into token rewards.

AIOZ W3AI’s two-layer architecture

Traversing the “Dual-layer Architecture” with “Artificial Intelligence Routing”.

In the midst of a well-structured architecture, managing the logic resources and task data flow between the operation of the dual-layer architecture is crucial. Hence, W3AI introduces artificial intelligence routing into the dual-layer architecture, dynamically optimizing each task to enhance the overall system’s efficiency.

At the infrastructure layer, artificial intelligence routing assesses computational demands and current node loads, dynamically allocating tasks to ensure each node participates in suitable tasks based on its capabilities and real-time network conditions. It also monitors node health, promptly detecting and addressing potential node failures or performance bottlenecks to prevent single-point failures from affecting overall efficiency.

At the application layer, intelligent routing enables rapid response to user requests, dynamically adjusting data flow and processing strategies in real-time. Additionally, it intelligently allocates the most suitable nodes based on user-specific geographical locations and requirements. Facing large-scale high-concurrency tasks, the AI routing architecture optimizes task scheduling intelligently, supporting the application layer in handling complex AI models and big data analysis.

The whitepaper includes numerous complex formulae to illustrate the specific implementation of routing. Interested readers can refer to the whitepaper document for further details.

Artificial intelligence routing allocates task transmission paths for AIOZ DePIN nodes. In the diagram, green represents connected nodes, while blue represents portions skipped due to low confidence.

Workflow: An Example of AI Task Implementation

With these rich infrastructure architectures, how does W3AI unfold its workflow? From data input to result output, W3AI’s workflow embodies a complete decentralized operational mode: encryption of output → task segmentation and allocation → execution of computing tasks and storage → gathering completed computations in containers → users obtaining decrypted output results.

We can refine the above process into simple steps:

Firstly, before data input and encryption, user-uploaded data undergoes homomorphic encryption to ensure data security throughout the processing journey — data input and encryption;

The encrypted data is then segmented into multiple small segments based on task requirements, with each task assigned to the most suitable node for execution — task segmentation and allocation;

Selected nodes execute specific computing tasks, such as AI model training or data analysis, while also being responsible for relevant data storage — computing and storage execution;

After task completion, results are re-encrypted and stored in transformed containers, awaiting retrieval by end-users — result collection and encryption;

Only authorized users can access the final results, which undergo homomorphic decryption before output—result decryption and output.

Workflow Architecture of W3AI

Through the above process, W3AI enhances processing efficiency while also balancing flexible and scalable characteristics with data security and privacy. It optimizes system resource utilization, reduces manual intervention, and lowers operational costs.

Token Economy Surrounding the Entire Ecosystem

$AIOZ plays a crucial role in linking the entire AIOZ W3AI ecosystem. With the emergence of AI-as-a-service and shared computing power businesses, its token has gained more usage scenarios and value capture.

Data Trading and Contribution Incentives

$AIOZ is used to reward users who provide computing power and storage resources, ensuring the stable operation of the network. In the platform’s trading market, users can use $AIOZ to purchase various AI services or buy and sell AI models and datasets. Additionally, token holders can participate in network governance by voting to decide the ecosystem’s next steps.

Ecosystem Maintenance

A portion of the transaction fees paid in $AIZO is used for AIOZ network operation and financial management, ensuring the platform’s ongoing maintenance and development. Another portion is burned directly to help regulate token supply and mitigate inflation. This carefully designed token flow cycle incentivizes innovation, rewards participation, and drives the continuous development of the AIOZ W3AI ecosystem.

Token flow within the W3AI ecosystem

Conclusion

As a decentralized project transitioning to AI, AIOZ W3AI possesses natural advantages in technological resources and operational mechanisms. In terms of technology and concepts, W3AI demonstrates significant potential to provide users with more secure, flexible, and efficient computing services and an engaging ecosystem experience. However, it’s essential to note that W3AI also faces challenges such as market maturity in recognizing and trusting decentralized AI solutions and the potential high operating costs under a system with stringent standards.

The current whitepaper resembles more of a blueprint drafted in the early stages of the project, preparing for the future but yet to be implemented and executed. Questions remain about how many people will use it and whether there are other security and technical issues, all of which await market validation.

Nevertheless, embracing a positive narrative transition remains a correct posture for Web3 projects when business relevance is high. Both new and established projects are enthusiastically staging the AI drama, and only time will tell whether the cryptographic players offstage will get their money’s worth.

Disclaimer：

1. This article is reprinted from [TechFlow]. All copyrights belong to the original author [TechFlow]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.