TL;DR

1. LLM: The development and progress of large language models (LLMs) and AI represent monumental technological advancements in human history, ushering humanity into the AI era where “computing power” becomes the scarcest resource in this new world.
2. The trend in computing power development is edge computing, a method that effectively reduces physical latency and serves as the cornerstone for industries with low-latency demands such as the metaverse. Decentralized distributed cloud computing possesses advantages such as flexibility, low cost, and resistance to censorship, with vast prospects for development.
3. Aethir is a decentralized real-time rendering platform built on the Arbitrum network, providing enterprise-level computing power services to industries like gaming and artificial intelligence by aggregating high-computing-power GPUs like H100.
4. Aethir has collaborated with top cloud computing projects in the industry such as io.net, Theta, as well as several leading game studios and telecommunications companies. It is projected that Aethir’s Annual Recurring Revenue (ARR) for the first quarter of 2024 will exceed $20 million.
5. Aethir Edge significantly lowers the barrier for ordinary users to sell surplus computing power and greatly expands the geographical coverage of the Aethir network.
6. By selling checker node NFTs, Aethir has raised $80 million, demonstrating the project’s promising prospects and the attractiveness of its economic model to a wide user base.
7. The hourly usage cost of A100 on the Aethir network is significantly lower than that of other competitors, giving it a clear competitive advantage.

Introduction: The Integration of AI and Blockchain

The course of human social development often shifts with the realization of a few immensely significant scientific inventions and advancements. Each breakthrough in technology directly ushers in a more efficient and prosperous new era.

The Industrial Revolution, the Electrical Revolution, and the Information Revolution stand as monumental technological advancements in human history. They fundamentally altered the face of human society, bringing about unprecedented changes in productivity and lifestyle. Now, we cannot return to the era of oil lamps for lighting or horse-drawn carriages for delivering messages. With the birth of LLMs, humanity enters another great new era.

LLMs are progressively liberating human intelligence, allowing individuals to allocate their limited energy and intelligence toward more creative thinking and practice, leading to a more efficient world.

We view GPT as another transformative technological breakthrough, not only because of its significant advancements in natural language understanding and generation but also because humans have discerned the pattern of LLM capability growth through GPT’s evolution. That is, by continually expanding model parameters and training data, LLM models can experience exponential improvements in capability. With sufficient computing power, this process currently shows no visible bottlenecks.

source:https://arxiv.org/pdf/2202.05924.pdf,https://developers.io.net/docs/how-we-started

The utility of large language models (LLMs) is not limited to understanding human language and conversation; rather, this is just the beginning. Once machines possess language understanding capabilities, it’s like opening a Pandora’s box, unleashing infinite possibilities. People can harness AI’s capabilities to develop various disruptive functionalities.

Currently, LLM models are making strides across various interdisciplinary fields of technology. From humanities like video production and artistic creation to hard sciences like drug development and biotechnology, monumental changes are on the horizon.

In this era, computing power is considered a scarce resource. Large tech giants wield abundant resources, while emerging developers face barriers to entry due to insufficient computing resources. In the AI era, computing power equals strength, and those who control it have the power to change the world. GPUs, as the cornerstone of deep learning and scientific computing, play a crucial role.

In the rapidly evolving field of artificial intelligence (AI), we must recognize the dual aspects of development: model training and inference. Inference involves the functionality and outputs of AI models, while training encompasses the complex process of building intelligent models, including machine learning algorithms, datasets, and computing power.

Take GPT-4, for example. To achieve high-quality inference, developers need comprehensive foundational datasets and immense computing power to train effective AI models. These resources are primarily concentrated in the hands of industry giants like NVIDIA, Google, Microsoft, and AWS.

The high cost of computation and barriers to entry prevent more developers from entering the field, perpetuating the dominance of major players. They possess large-scale foundational datasets and abundant computing power, with the ability to continually scale up and reduce costs, making industry barriers even more formidable.

However, we must consider whether there are solutions to reduce computing costs and industry entry barriers through blockchain technology. The answer is affirmative. Decentralized distributed cloud computing precisely offers such a solution in this era.

The Feasibility and Necessity of Decentralized Cloud Computing

Supply Issues: Low GPU Utilization

Despite the current expensive and scarce computing power, GPUs are not fully utilized. This is mainly because there hasn’t been a ready-made way to integrate these dispersed computing powers and make them operate commercially. Here are the typical GPU utilization figures for different workloads:

Most consumer devices with GPUs fall into the first three categories, namely idle (just booted into the Windows operating system):

GPU utilization: 0-2%;

General productivity tasks (writing, simple browsing): 0-15%;

Video playback: 15 - 35%.

The above data indicates extremely low utilization of computational resources, and in the world of Web2, there are no effective measures to collect and integrate these resources. However, the Crypto and blockchain economy may be the perfect remedy for this challenge. The crypto economy constructs an extremely efficient global market, and due to the unique token economy and characteristics of decentralized systems, the pricing, circulation, and matching of market supply and demand for resources are highly efficient.

Edge Computing

The development of AI shapes the future of humanity, and the progress of computing power determines the development of AI. Since the invention of the first computer in the 1940s, the computing paradigm has undergone several transformations. From bulky mainframe computers to lightweight laptops, from centralized server purchases to computing power leasing, the barriers to accessing computing power are gradually decreasing. Before the advent of cloud computing, enterprises had to purchase their own servers and continuously upgrade them with technological innovations. However, the emergence of cloud computing completely changed this model.

The basic concept of cloud computing is that consumers lease servers, access them remotely, and pay based on usage. Now, traditional enterprises are being disrupted by cloud computing. In the field of cloud computing, virtualization technology is at its core. Virtualized servers can divide a powerful server into smaller ones for rent and can dynamically allocate various resources.

This model has fundamentally changed the business landscape of the computing power industry. Previously, people needed to purchase computing facilities to meet their computing needs. Now, they only need to pay rent on websites to enjoy high-quality computing services. The future direction of cloud computing is edge computing. Traditional centralized systems are often too far from users, resulting in some degree of latency. Although latency can be optimized, it can never be completely overcome due to the limitation of the speed of light.

However, emerging industries such as the metaverse, autonomous driving, and remote healthcare have extremely low latency requirements. Therefore, cloud computing servers need to be moved closer to users, and more and more small data centers are being deployed around users. This is edge computing.

Advantages of Decentralized Edge Computing

Compared to centralized cloud computing providers, the advantages of decentralized cloud computing mainly lie in:

• Accessibility and Flexibility: Accessing computing power chips from centralized cloud service providers such as AWS, GCP, or Azure often takes several weeks, and high-performance GPU models like A100, H100, etc., are frequently out of stock. Additionally, consumers usually need to sign long-term, inflexible contracts with these large companies to access computing power, leading to time loss and inflexibility in operations. In contrast, distributed computing platforms can provide computing power on-demand and offer flexible hardware choices, enhancing accessibility.
• Lower Cost: Utilizing idle chips and token subsidies from network protocols, distributed computing networks may offer more affordable computing power.
• Resistance to Censorship: Web3 systems like io.net and Aethir do not position themselves as permissionless systems. They address compliance issues such as GDPR and HIPAA during GPU onboarding, data loading, data sharing, and result sharing stages.

With the further development of AI and the persistent supply-demand imbalance of GPUs, more developers will be driven towards decentralized cloud computing platforms. Simultaneously, during bullish market periods, due to the rise in cryptocurrency prices, GPU suppliers will earn higher profits, stimulating more GPU providers to enter this market, creating a positive feedback loop.

Challenges of Decentralized Edge Computing

Technical challenge

1. Parallelization challenges

Distributed computing platforms usually aggregate long-tail chip supplies, which means that a single chip supplier is almost unable to independently complete complex AI model training or inference tasks in a short period of time. If the cloud computing platform wants to be competitive, it must use parallelization to disassemble and distribute tasks to shorten the total completion time and improve the computing power of the platform.

However, there are a series of problems faced during parallelization, including how tasks are decomposed (especially for complex deep learning tasks), data dependencies, and additional communication costs between devices.

2. Risk of new technology substitution

With significant capital investment in ASIC (Application-Specific Integrated Circuit) research and new inventions like Tensor Processing Units (TPUs), there could be an impact on the GPU clusters of decentralized computing platforms.

If these ASICs can offer good performance and a balanced cost, the GPU market currently monopolized by large AI organizations may revert back to the market. This would lead to an increase in GPU supply, thereby affecting the ecosystem of decentralized cloud computing platforms.

3. Regulatory risks

Due to decentralized cloud computing systems operating across multiple jurisdictions and potentially being subject to different legal regulations, there may be unique legal and regulatory challenges. Compliance requirements, such as data protection and privacy laws, may also be complex and challenging.

Currently, the users of cloud computing platforms are mainly professional developers and institutions who prefer to use a platform for the long term and are unlikely to switch arbitrarily. Whether using a decentralized or centralized platform, price is just one consideration; these users place greater emphasis on service stability. Therefore, decentralized platforms that have strong integration capabilities and stable computing power are more likely to gain favor from these customers, leading to long-term partnerships and stable cash flow income.

Below, I will introduce Aethir, a new distributed computing project focused on game rendering and AI in this cycle, and estimate its potential valuation based on similar AI projects and distributed computing projects in the current market.

Distributed Cloud Computing Platform: Aethir

Project Overview

Aethir Cloud is a decentralized real-time rendering platform built on the Arbitrum network. It aggregates and intelligently reallocates new and idle GPUs from enterprises, data centers, cryptocurrency mining operations, and consumers to help gaming and artificial intelligence companies deliver their products directly to consumers.

One of the key innovations of the project is the resource pool, which gathers dispersed computing power contributors under a unified interface to serve global customers. A notable feature of the resource pool is that GPU providers can freely connect or disconnect from the network, allowing enterprises or data centers with idle equipment to participate in the network during downtime, thereby increasing supplier flexibility and device utilization.

The Aethir ecosystem operates on three core infrastructures:

The operation of the Aethir ecosystem relies on three core infrastructure components:

• Container: The key function of containers is to provide real-time remote rendering services, offering a “zero latency” experience. Containers serve as the actual locations of cloud computing, acting as virtual endpoints to execute and render applications. This shifts workloads from local devices to containers.
• Checker: Checker nodes ensure the integrity and service quality of the Aethir network by inspecting containers and their service processes. After completing tasks, Checker Nodes sign the results with their private keys and pass them to the referee. The referee receives 2N+1 results for each container, and nodes providing results similar to the majority will receive token rewards.
• Indexer: Serving as the core of the Aethir network, indexers match consumers with suitable containers, ensuring fast startup of cloud applications and services. The goal is to provide “second-level” services - the transition from a consumer’s request to actual delivery (e.g., from a player’s request to the game screen) should occur in the shortest possible time. This requires concise signals and efficient scheduling. To maintain decentralization and enhance user experience, indexers are randomly selected to reduce potential fraud risks and signal latency.

Partners and Business Data

Aethir boasts a powerful team and substantial resources, as evidenced by its achievements:

• Aethir has entered into a partnership agreement with io.netT network, integrating both ecosystems to provide customers with a seamless GPU computing experience, including services like clustering and serverless inference. Additionally, Aethir and io.netT communities are conducting a $50 million airdrop exchange, wherein each community will airdrop tokens worth $50 million to the other, fostering the development of the entire ecosystem.
• Aethir is collaborating with Theta EdgeCloud, the first decentralized AI cloud edge computing platform, to launch the world’s largest hybrid GPU marketplace.
• Aethir has signed contracts with WellLink, the largest cloud gaming company with over 64 million monthly active users, focusing on mixed reality metaverse and game development. Contracts have also been signed with the world’s largest game studio, boasting 150 million monthly active users.
• Contracts have been secured with the world’s largest telecommunications company, generating annual recurring revenue of $5 million (potentially rising to $13.9 million in the near future).
• A partnership has been established with Meta48, becoming the exclusive GPU cloud computing service provider for its mixed reality products. Meta48 is a tech company focused on Web3.0 metaverse development, with its core product being a metaverse world combining elements of augmented reality (AR), virtual reality (VR), and mixed reality (MR).
• An additional 10 contracts in the gaming sector are currently in the pipeline.

In the first quarter of 2024, Aethir generated sales revenue of $80 million from the sale of nodes, setting a record surpassing both Depin and AI projects’ sales. Aethir has validated the market demand from enterprise-level GPUaaS customers in AI model training, virtual computing, and gaming domains. Three existing contracts are expected to generate annual recurring revenue (ARR) exceeding $20 million in the first quarter of 2024. Since its launch, revenue from service fees has reached $1.8 million.

In contrast, the entire DePin industry generated only $24 million in ARR last year. These sales revenues not only significantly exceed those of other leading agreements in the industry but also demonstrate Aethir’s exceptional position in the sector, providing substantial valuation opportunities. Furthermore, its robust resources and strong capabilities contribute to expanding its scale, enhancing its technological prowess, and establishing formidable competitiveness, thus creating a sustainable and self-reinforcing growth cycle.

Aethir Edge

Aethir Edge is the hardware device launched by the Aethir network to power the next generation of GPU cloud computing. The device utilizes cutting-edge technology, including the Qualcomm® Snapdragon 865 chipset and 12GB LPDDR5 memory, to handle data-intensive workloads at the edge.

With 256GB UFS 3.1 storage capacity, Aethir Edge provides high-speed data access, along with a 1000M GE LAN network port and WIFI6 2T2R + BT5.2 connectivity. This endows Aethir Edge with enterprise-level computing capability.

It can aggregate various idle GPU resources through the Aethir network over the Internet and directly deliver them to end-users. This industry model of aggregating resources significantly enhances the supply and accessibility of computing power, holding significant implications for the artificial intelligence and gaming industries.

Aethir Edge enables every user to participate as a computing power provider in the Aethir network and earn token rewards without the need for specialized graphics cards. Users can connect it to the network via wireless or wired connections. Once connected, Aethir Edge can be managed through a dedicated mobile application, allowing users to optimize device performance and configure it for specific tasks, such as GPU computing.

With this device, users can leverage their device’s bandwidth, IP address, or GPU computing resources to the fullest extent. Unlike the high barriers to entry of specialized mining rigs, Aethir Edge lowers the threshold for individuals to provide computing resources, allowing everyone to purchase one and profit from their underutilized computing power.

Furthermore, the portability and flexibility of Aethir Edge enable it to be deployed across vast geographical regions, expanding the coverage of Aethir’s GPU cloud compared to traditional clouds. This significantly reduces network latency, allowing users in remote areas to enjoy the convenience of cloud computing.

Team Background and Funding Situation

Team background

Aethir’s core team boasts a diverse background. The founders of Aethir are seasoned executives with a track record of success in artificial intelligence, cloud computing, Web3, and the gaming industry. The team comprises former Web3 professionals, traditional financial investment practitioners, and technical personnel from traditional tech companies, indicating strong networking capabilities and resource integration abilities in both the Web2 and Web3 domains.

• Mark Rydon: Co-founder and CEO, previously held key positions at NOTA Platform, Flux Capital, Gaas LTD, Kulture Athletics, Inc., and Bechtel Corporation.
• Daniel Wang: Co-founder and COO, formerly served as a venture partner at IVC, Chief Information Officer at YGG SEA, and International Publishing Director and Operations Director at Riot Games and Riot Games China, respectively.
• Kyle Okamoto: Chief Technology Officer, previously served as CEO and General Manager of Ericsson’s IoT, Automotive, and Security business, CEO of Edge Gravity, and Chief Network Officer at Verizon Media.

Funding situation

• Seed Round (Token Round, early 2022): $60 million valuation, undisclosed funding amount.
• Pre-Series A Round (Token Round, early 2023): Raised $9 million, valuation of $150 million.

The second round of investment was led by Hashkey Capital, with other participating investors including Mirana Ventures, Animoca Brands, Maelstrom Capital, Sanctor Capital, Merit Circle, Big Brain Holdings, Builder Capital, Momentum 6, Tess Ventures, CitizenX Crypto Ventures, and Lapin Digital.

Aethir Economic Model Review

Core members of the Aethir network

In the Aethir network, there are five main user groups: miners, developers, users, token holders, and Aethir DAO.

Among them, miners are divided into three categories:

Containers: They provide remote rendering services.

Checkers: They evaluate the performance and service quality of containers.

Indexers: They match suitable containers for users.

Users utilize Aethir Edge to offer idle GPU computing power.

Tokens and Token Flow and Distribution

The ecosystem token of the system is $ATH, with a total token supply of 42 billion.

50% of the total token supply is allocated to miners, namely, the providers of computing power and checker nodes.

Out of this allocation, 35% is allocated to the providers of computing power:

Edge/Enterprise/IDC: This includes providers of idle GPU computing power and professional institutions offering high-performance GPU computing power. Among these, 23% of the tokens are allocated to Edge, while the remaining 12% are allocated to Enterprise and IDC.

15% of the tokens are allocated to checker nodes, with 10% of them being distributed over a four-year period and the remaining 5% being used for performance incentives for checker nodes. Private sale and team member tokens are subject to a lock-up period of one year or more.

Three types of uses for $ATH:

1. Transactional Purpose: $ATH is used for purchasing computing power and compensating miners for their services. As the Aethir ecosystem continues to evolve, with the introduction of merged mining and integrated markets into the Aethir network, $ATH will continue to serve as a medium of exchange in various applications.

2. Governance Purpose: Holders of ATH tokens have the right to propose governance suggestions on the platform within the DAO and participate in discussions and voting.

3. Staking: Node operators are required to stake $ATH tokens to ensure the security and reliability of their nodes. If a node misbehaves, tokens may be partially or fully slashed. Additionally, containers can stake tokens to earn more rewards, with the amount of staked tokens correlating to better container performance and higher token rewards.

Moreover, rewards are distributed in phases based on the workload and performance of containers. For invalid computations from these nodes, the system will deduct a portion of the rewards and handle them in three ways: 10% will be burned, 25% will be reallocated to eligible nodes for rewards, and the remaining 75% will be allocated to the DAO for future redistribution.

Analyzing The Economic Ecology Of The Aethir Network From The Perspective Of Capital Inflow/Outflow And Project Profitability

Purchase of Aethir Checker licenses

Miners need to purchase Aethir Checker licenses to become Checker nodes. Node sales adopt a pyramid-tiered pricing structure, with a starting price of 0.1259 wETH and a total quantity of 100,000. Node sales are conducted through whitelist and public sale, with higher-level nodes priced lower and corresponding quantities smaller. As lower-priced nodes sell out, buyers can only purchase nodes at higher prices. Currently, node sales have exceeded $80 million.

To encourage node buyers to promote node sales, the Aethir project has implemented a commission system. Buyers using a referral code for node sales will receive a 10% or 5% purchase rebate, and the provider of the referral code will receive a 10% or 5% kickback. Based on purchasing results, referral code providers receive substantial kickbacks. This system design will incentivize node buyers to promote the project to more people, achieving both community expansion and increased sales revenue.

source:https://www.aethir.com/checkersaleinfo

Upon successful purchase, buyers will receive an NFT representing a perpetual license for the Checker Node. NFT owners can operate the node themselves or host it to earn mining rewards. The value of the node NFT can be estimated using a model similar to perpetual bonds. The theoretical total value of NFTs in the current market equals 15% of the total market capitalization of ATH, simplifying the process of capital discounting. Which is:

If the token price is too high and the NFT price is too low, meaning the left side is less than the right side of the equation, it will increase the demand for NFTs. Conversely, if the token price is too low, leading to the left side being greater than the right side, it will increase the supply of NFTs until the price returns to an equilibrium level. Additionally, NFTs are non-transferable within the first year of node sale, ensuring the continued stability of Checker nodes in the network. However, the price of NFTs will to some extent affect the price of $ATH tokens.

Due to the pyramid-shaped pricing model of node NFTs, it determines that players holding low-cost NFTs at the beginning will obtain lower-cost chips. In other words, their breakeven period is shorter, and their incentive to hold tokens is weaker compared to players with longer breakeven periods, leading to heavier selling pressure.

To address this issue, Aethir has designed the $vATH program to alleviate selling pressure. Firstly, Checker nodes earn income through mining rewards, with 10% of the total token supply distributed to Checker nodes over a period of 4 years, and an additional 5% of tokens used to provide bonuses to eligible nodes. Secondly, these rewards are distributed in the form of $vATH. When Checkers claim their token rewards, these tokens will have a lock-up period of 30 or 120 days depending on various conditions. After the lock-up period ends, $vATH will convert to $ATH and can be sold.

If $vATH holders wish to withdraw tokens early and bypass the lock-up period, 75% of the converted tokens will be fined, and they will lose eligibility for receiving bonuses. After 120 days, the default lock-up period will be extended to 180 days, meaning the time for Checkers to claim token rewards will be extended, effectively alleviating selling pressure.

Purchase of Aethir Edge

Aethir Edge is a hardware component of the Aethir Depin stack. Through Aethir Edge, ordinary users can provide their underutilized computing power, bandwidth, and other resources to the Aethir network and receive token rewards, which account for 23% of the total $ATH supply. Currently, the official sales price of Aethir Edge has not been disclosed.

It can be speculated that the token rewards obtained through Edge mining may have a lock-up period similar to Checker nodes. Since transferring Edge is more difficult and users incur higher exit costs, the lock-up period for their token rewards may be shorter, thereby incentivizing more users to participate.

In conclusion, selling Checker nodes and Edge mining machines is one of the important means for the project to profit. In the context of Checker nodes being sold out, selling more mining machines has become an important goal of the current project operation. Increasing mining expected returns is an important means to achieve this goal. At the time of $ATH token issuance, only about 5% of the circulating chips were available in the market. The project team is motivated to actively boost the token price, thereby increasing the project’s popularity and annualized returns, and attracting more users to participate in the construction of the Aethir ecosystem.

Track Analysis

Let’s summarize the valuations and funding amounts of other projects in the same field:

• io.net: io.net is a decentralized computing network aiming to combine 1 million GPUs to form the world’s largest GPU cluster and DePIN infrastructure. It operates in the fields of cloud computing, AI, and DePIN concepts, with a Series A funding of $30 million and a valuation of $1 billion. Investors include Hack VC, OKX Ventures, and other renowned institutions.
• Grass: Grass is the flagship product of Wynd Network, allowing users to profit from selling unused network resources through internet connections. It operates in the fields of cloud computing and AI, with a total funding of $4.5 million over two rounds.
• Render Network: RNDR is a rendering solution provider based on decentralized GPU, connecting users executing rendering jobs with users owning idle GPUs through a rendering network. It operates in the fields of cloud computing and AI, with a funding of $30 million and a circulating market value of $3.5 billion.
• Akash: Akash Network is a distributed peer-to-peer marketplace for cloud computing, primarily utilizing idle CPU power for rent. It operates in the cloud computing field with a circulating market value of $1.25 billion.
• Gensyn: Gensyn is a distributed computing network for training AI models, with a total funding of $49.5 million led by a16z. It operates in the fields of cloud computing and AI.
• GAIMIN: GAIMIN has created a decentralized data processing network, monetizing the computing power of surplus gaming PCs. It operates in the fields of cloud computing and gaming, with an FDV of $900 million.
• Hive: Hive is a distributed data storage and computing platform, allowing users to rent out their computing resources for localized storage and processing. It operates in the fields of cloud computing and infrastructure, with a total funding of $20 million.
• Meson Network: Meson Network serves as the data transmission foundation for decentralized storage, computing, and emerging Web3 DApp ecosystems. It integrates and monetizes idle bandwidth from long-tail users, operating in the infrastructure and cloud computing fields with a valuation of $1 billion.
• Flux: Flux is a decentralized Web3 cloud infrastructure consisting of scalable and globally distributed computing nodes operated by users. It operates in the fields of infrastructure and cloud computing, with a circulating market value of $340 million.
• Aleph.im: Aleph.im is a decentralized cloud computing platform providing on-demand serverless computing, database, and storage infrastructure for decentralized application and protocol development. It operates in the fields of cloud computing and infrastructure, with a funding of $10 million and an FDV of $180 million.
• FormAI: FormAI is creating a decentralized and distributed computing network primarily for training and inference of AI models, operating in the fields of cloud computing, Depin, and AI. Funding and valuation information is unknown.

Here, we will select four cloud computing projects, io.net, Render, Akash, and Gensyn, which are comparable to Aethir in terms of investment lineup and valuation strength, for detailed comparison and introduction.

io.net

io.net’s decentralized applications are built on top of Solana, focusing primarily on consumer and enterprise AI/ML. As of the completion of this article, io.net has a total of 648,043 GPUs and 94,750 CPUs. Among them, there are 44,398 H100 and A100 GPUs, 21,777 A100-SXM4-80GB GPUs, 18,409 A100 80GB PCIe GPUs, and 320 A100 PCIe 80GB K8S GPUs, with the latter having the highest utilization rate. As of the date of the author’s research (2024/4/24), the utilization rate reached 94%.

Render

Render Network primarily focuses on rendering services. Currently, the network has 23,651 GPUs and 1,006 CPUs. Among them, there are 141 H100 PCIe GPUs and 2 H100 80GB HBM3 GPUs.

Akash

Akash has a total of 21,500 CPUs and 374 GPUs, with 149 GPUs being of various models including H100 and A100.

Gensyn

Gensyn aims to aggregate idle computing resources worldwide to achieve low-cost and large-scale programmatic machine learning training. Its network is currently in Devnet mode, and there are no publicly available statistics regarding its network scale.

io.net has a leading advantage in the number of connected GPUs. In comparison, Aethir has 41,756 GPUs, with 3,000 H100 GPUs and 440 H100 GPUs online. Its enterprise-grade high-performance GPU quantity is more advantageous compared to RNDR, Akash, and Gensyn.

Additionally, Aethir can provide the most favorable rental rates for A100 devices at $0.33 per hour. This price presents a significant competitive advantage compared to both Web2 giants and Web3 competitors.

Source: Messari, io.net, Akash Network, Aethir Network

Summary

“Whoever controls the computing power, controls the world.” In the era of AI, we do not want computing power to be monopolized by a few, like rare diamonds. Therefore, people have chosen decentralized cloud computing to reshape productivity and production relations in this revolution, using blockchain technology.

Aethir, as a powerful player with a triple core narrative of AI+DEPIN+GAMEFI, has achieved many outstanding results even before the token is launched. Its strong team capabilities and resource background ensure that this project is just getting started and will eventually achieve astonishing results.

The launch of Aethir Edge breaks the bottleneck of high-latency edge computing, allowing everyone to contribute computing power and everyone to obtain computing power. Aethir Cloud can now be deployed to every corner of the world, thanks to the convenience provided by Edge.

Currently, this computing power revolution is just beginning. Aethir will stand out among many projects by bridging the gap between high-performance computing demand and Web3 AI technology. I believe that after its token is launched, its performance will become a dazzling new star in the market!

Statement:

1. This article is reproduced from Gryphsis Academy，Original title “Aethir: A powerful player in decentralized cloud computing with three tracks”, the copyright belongs to the original author [@dadayu34], if you have any objection to the reprint, please contact Gate Learn Team, the team will handle it as soon as possible according to relevant procedures.

2. Disclaimer: The views and opinions expressed in this article represent only the author’s personal views and do not constitute any investment advice.

3. Other language versions of the article are translated by the Gate Learn team and are not mentioned inGate.ioThe translated article may not be reproduced, distributed or plagiarized.