Decentralization Cloud Computing: A New Chapter in the AI Computing Power Revolution

With the long-term rapid development of technology, the market values of companies like OpenAI and NVIDIA have shown several times growth in the past two years. Crypto x AI has become the core narrative of this cycle, with strong market sentiment and sustained capital investment proving the formation of a strong consensus. In the context of AI, decentralization, as a powerful tool for AI development, indeed possesses great attraction and imaginative space. Although there is still a significant gap in practical business implementation compared to centralized models, utilizing the advantages of web3 to expand the four core aspects of AI (data, model, training, inference), and continuously optimizing to unleash greater potential has become a common goal for web3 participants.

Currently, Decentralization can provide technical support in the four aspects mentioned above. Data is the raw material for AI technology, while the others are processing methods. Whether it is data labeling or storage, Decentralization plays a significant role and has great value.

If data is the raw material, then Computing Power is the tool for processing the raw material, used to maximize output efficiency. This article will briefly analyze the ecological framework and economic model of "Computing Power" in Crypto x AI x DePIN.

1. DePIN and Decentralized Computing Power Ecological Framework

Pain Point: High-quality Computing Power is an essential resource for AI development, and this scarce resource has been monopolized by traditional giants, making it difficult for startups and individual users to obtain reasonably priced Computing Power, with high prices being unacceptable to most buyers.

Decentralization solutions: Currently, DePIN projects mostly adopt a P2P economic model to provide high-quality resources to resource demanders, allowing each user to act as a physical facility resource provider while earning token rewards.

With the surge in demand for decentralized AI Computing Power, a balanced and comprehensive framework for decentralized AI Computing Power supply ecosystem has been formed to better meet customer needs. Certain platforms among leading projects play different important roles in the ecosystem, and the technological barriers and the landscape for the future development of decentralized Computing Power of these projects are quite impressive.

The decentralized AI computing power ecosystem mainly consists of three parts, which serve as resource agents, resource providers, and distributors.

Resource Agent

A certain decentralized computing network acts as a computing power agent, providing high-quality AI computing power to customers at low prices. On the supply side, it has globally distributed GPUs, and the client base currently consists mainly of seed to Series B startup companies focused on AI inference.

The DePIN project based on a certain public blockchain recently completed a $30 million Series A financing, participated by several well-known investment institutions.

As a top AI Computing Power resource agency, this project aims to aggregate 1 million GPUs to form a large DePIN Computing Power network, providing customers with lower-priced computing power. Users can manually contribute idle GPU/CPU computing power to the platform to receive token incentives. The core goal is to provide high-quality AI computing power through Decentralization to help AI startups reduce costs.

The computing services provided by this project use a cluster construction module to keep all GPUs interconnected, enabling large-scale coordinated work during training and inference. When coordinating GPUs, Computing Power can concentrate on accessing larger databases and computing more complex models, allowing AI startups to obtain what they need while completing hardware deployment at one-tenth of the centralization price. More notably, this project focuses on aggregating machine learning Computing Power, which can help other DePIN projects format GPU supply for machine learning, providing the most fundamental and direct resource support for the underlying technology.

Currently, the number of GPU clusters assembled by this project is the first in the industry. There are over 200,000 GPUs available online, including nearly 50,000 GeForce RTX 4090 and over 30,000 GeForce RTX 3090 Ti.

Resource Provider

As the most promising AI Computing Power provider, a certain project acts as an AI Computing Power service node and can provide ample chips for deep machine learning. Its team stands out in the traditional AI Computing Power resource sector. The team has been a primary agent for AI giant companies and, relying on technological resource barriers, can directly access hundreds of data centers on the resource supply side, with access to A/H100, RTX4090, and A6000 machines available.

The project provides large-scale machine learning computing power for web3 computing power giants on the client side. Compared to a certain client who needed to spend $140,000 per month to obtain cloud services, after migrating to this project, the monthly cloud service usage fee has dropped to around $40,000, reducing expenses by 70% while improving efficiency by 30%.

The main purpose of this project is to provide customers with the fastest, highest quality, and most reliable Computing Power through unique Computing Power supply channels. High-quality Computing Power can save users costs while offering a full range of service options.

The quality of AI Computing Power provided by this project has been recognized by multiple AI Computing Power agents, and it has now partnered with several Computing Power giants, aiming to contribute to machine learning through Decentralization.

Resource Channel Provider

A certain project acts as a DePIN resource channel provider, offering services through demand matching. It adopts a platform-style open protocol, providing underlying aggregated resources before delivering services. The goal is to become a service aggregator for DePIN, which can be understood as the 1inch or the aggregated version of Uber for DePIN.

Service provision method: Obtain SDK by controlling layer to provide various networks and strategies, resource conditions, performance, stability, etc., and then deliver the SDK to users through routing algorithms.

Pain points: Each DePIN network has limitations on resources and services, and the globalization of resource allocation leads to poor service quality due to excessive regional concentration.

Solution: Routing Algorithm - Obtain data, basic network information, and machine information, aggregate them to generate strategies, and provide services that match customer requirements. The goal is to improve the quality and service of the DePIN application layer, and to find the optimal price Computing Power network under resource constraints.

2. Analyzing Decentralization Computing Power Ecosystem

Some projects have reached strategic cooperation, jointly committed to improving network speed and stability. They unanimously believe that the success of the decentralized computing industry and the combination of web3 and AI can only be achieved through close collaboration among early industry leaders. With the growing demand for Computing Power, traditional cloud computing faces some challenges:

• Limited Availability: It usually takes weeks to gain hardware access using mainstream cloud services, and commonly used GPU models are often unavailable.
• Limited choices: Users are restricted in their choices regarding GPU hardware, location, security level, latency, and more.
• High cost: Quality GPUs are expensive, with monthly project training and inference expenses easily reaching hundreds of thousands of dollars.

The vision of decentralized computing is to provide an open, accessible, and affordable alternative to address the core issues of centralized cloud service providers. Currently, it seems that challenging the dominance of major cloud computing players requires innovators to work together and support each other in order to take a revolutionary step.

Asset Mode

Heavy Asset Model

A certain supply-side project has top GPU manufacturers as its backing, creating an absolute barrier. The machines with valuable machine learning Computing Power are only the A100, RTX4090, and H100, with a single unit priced at around $300,000. These machines have become highly scarce resources, long monopolized by traditional AI giants. In this situation, the supply-side resources that this project can connect to are extremely valuable. Since individual users sharing their personal GPU idle Computing Power do not have the quality to support large-scale AI model computing and processing, this project plays a crucial and irreplaceable role in the decentralized Computing Power ecosystem.

The project adopts a heavy asset model requiring a large amount of fixed asset investment, and this scale of capital and technology input makes it difficult for startups to replicate. If it can collaborate with more decentralized Computing Power agents and continuously expand the supply side, providing the necessary Computing Power resources for the industry, it is expected to achieve monopoly in the B2B decentralized Computing Power field and generate economies of scale.

However, the biggest risk is that after investing a large amount of capital, it cannot continuously provide resources for Computing Power agents. Therefore, the supply side's ability to achieve large-scale profitability is extremely dependent on whether Computing Power agents can have a steady stream of clients. Regardless of who the Computing Power agents are, as long as there are clients and demand, the value of the project as a supply side will grow with the increase in demand.

Light Asset Model

A certain Computing Power agency relies on globally distributed GPUs to form a vast Decentralization computing network. From a business perspective, it adopts a light asset operation model, establishing a strong brand in the AI Computing Power agency field through community operations and building a high level of consensus.

Core Business:

1. Aggregate retail GPU Computing Power and reward tokens
2. Obtain high-quality Computing Power from the supply side to sell to AI startups.

From a business perspective:

1. Buy low and sell high quality Computing Power from the supply side to C-end customers.
2. Help users earn tokens by sharing idle GPU Computing Power.
3. Provide customers with Computing Power mining and staking platforms, but an initial investment of about 4000 USD is required to achieve better returns.

Customer Perspective:

1. The price of Computing Power is about 80% cheaper than other centralized cloud computing services.
2. Staking Earnings & Sharing Earnings
3. After customers invest a certain capital, they can earn compound interest.

As a typical light asset model company, the biggest advantage is the lower risk, as the team does not need to invest a large amount of machinery costs upfront like the supply side. Due to the lower capital investment, both the company and investors can more easily achieve higher profit margins. At the same time, due to the low entry barrier of the industry, the business model can be easily copied and replicated, which requires careful consideration for long-term value investors.

Three, from 10 to 100?

If certain project collaborations can help the decentralized Computing Power ecosystem go from 1 to 10, then adding more participants may have the opportunity to reach 100.

The goal of a certain project is to become the largest DePIN service aggregator, directly competing with web2's Uber. As a channel merchant, it connects customers with the best quality resources in terms of price and quality by aggregating various resource real-time situations. It adopts a B2B2C light asset business model, where the first B side is the supply side, the second B side is resource agents, and the C side offers customers the best resource choices through information provision.

As a platform, if the distributor can develop into a platform for issuing assets, it will make the products more valuable. By using the SDK provided by the routing algorithm to create AI Agents with computable resources, it converts new financial assets while dynamically assisting customers using the application in dynamic mining through the SDK, focusing on mining Computing Power that is useful for computable resources. This model is understood as Assets on assets, which can greatly enhance the liquidity of resources and funds.

For this project, we hope to see more suppliers and agents enter the Decentralization Computing Power ecosystem, better highlighting their own advantages, expanding their business lines, and having more customers. Simply put, the reason certain information platforms can dominate the information field is that more merchants and information are uploaded to the internet, which creates a high demand for channel providers from customers.

Four, the future is promising

Decentralized cloud computing is gradually developing. Although the ecological framework and model have become clear, and leading roles are fulfilling ecological responsibilities, it is still too early to shake the position of traditional cloud computing giants. Compared to traditional centralized cloud computing, decentralization does conceptually solve many problems for customers, but the overall resources and scale of this market are still very small in comparison. In the context where the computing power resources supported by AI are far from sufficient, the market needs another fresh force or new model to alleviate the predicament. The decentralized cloud computing we see now can indeed meet certain needs of startup AI companies. Where we go from here in the future, let us all be witnesses and participants on this path of disruption, and follow the evolution of the revolution together!