Background of the AI Revolution

Background of AI Boom

With the rapid advancement of artificial intelligence (AI) technology, we are entering a data-driven new era. Breakthroughs in fields such as deep learning and natural language processing have made AI applications ubiquitous. The launch of ChatGPT in 2022 ignited the AI industry, followed by a series of AI tools for video generation, automated office tasks, and the adoption of “AI+” applications. The market value of the AI industry has surged accordingly, with projections reaching $185 billion by 2030.

Figure 1: Changes in AI market value

Traditional Internet Companies Monopolize AI
Currently, the AI industry is dominated by companies like NVIDIA, Microsoft, Google, and OpenAI. While technological advancements have brought about rapid progress, they have also led to challenges such as data centralization and unequal distribution of computing resources. However, the decentralized nature of Web3 offers new possibilities to address these issues, potentially reshaping the current landscape of AI development.

Current Developments in Web3+AI

As the AI industry continues to surge, a wave of high-quality Web3+AI projects has emerged. Fetch.ai leverages blockchain technology to create decentralized economies, supporting autonomous agents and smart contracts for optimizing AI model training and applications. Numerai uses blockchain technology and a community of data scientists to predict market trends, rewarding model developers through an incentive mechanism. Velas has built a high-performance smart contract platform that integrates AI and blockchain, offering faster transaction speeds and higher security.

AI projects inherently consist of three key elements: data, algorithms, and computing power. While the Web3+data and Web3+computing power sectors are thriving, the Web3+algorithm direction has been fragmented, resulting in isolated, single-direction applications. Bittensor addresses this gap by creating a competitive AI algorithm platform with built-in selection and incentive mechanisms, ensuring only the best AI projects prevail.

Bittensor’s Development Timeline

Innovative Breakthroughs
Bittensor is a decentralized incentivized machine learning network and digital goods marketplace.

Decentralization: Bittensor operates on a distributed network of thousands of computers controlled by different companies and organizations, addressing issues like data centralization.

Fair Incentive Mechanism: In the Bittensor network, the $TAO tokens are distributed in proportion to the contribution of each subnet. Similarly, rewards provided by the subnet to its miners and validators are also proportional to their node contributions.

Machine Learning Resources: The decentralized network can provide machine learning computing resources to any individual in need.

Diverse Digital Goods Marketplace: Initially, Bittensor’s digital goods marketplace was designed specifically for the trading of machine learning models and related data. However, due to the expansion of the Bittensor network and the Yuma consensus mechanism’s data-agnostic principle, it has evolved into a marketplace where any form of data can be traded.

1. Development Process

Unlike many high-valuation VC projects in the current market, Bittensor is a more equitable, interesting, and meaningful project created by tech enthusiasts. Its development history lacks the typical “grand vision to lure investments” phase seen in other projects.

Concept Formation and Project Launch (2021): Bittensor was founded by a group of technology enthusiasts and experts committed to advancing decentralized AI networks. They used the Substrate framework to build the Bittensor blockchain, ensuring its flexibility and scalability.

Early Development and Technical Validation (2022): The team released the Alpha version of the network, validating the feasibility of decentralized AI. They also introduced the Yuma consensus, which emphasizes the principle of data-agnosticism to protect user privacy.

Network Expansion and Community Building (2023): The team launched the Beta version and introduced the token economy model (TAO) to incentivize network maintenance.

Technological Innovation and Cross-Chain Compatibility (2024): The team utilized Distributed Hash Table (DHT) integration technology to enhance data storage and retrieval efficiency. The project also began focusing on promoting and expanding its subnets and digital goods marketplace.

Figure 2: Bittensor Network Promotional Picture

In the development process of Bittensor, not many traditional VCs have intervened, avoiding the risk of centralized control. The project incentivizes nodes and miners through tokens, which also ensures the vitality of the Bittensor network. In essence, Bittensor is an AI computing power and service project driven by GPU miners.

Tokenomics

The Bittensor network token is TAO. To express its admiration for Bitcoin, TAO is similar to BTC in many aspects. Its total supply is 21 million coins, which is halved every four years. TAO tokens are distributed through fair launch when the Bittensor network launches. There is no pre-mining, so no tokens are reserved for the founding team and VC. Currently, a Bittensor network block is generated approximately every 12 seconds. Each block earns users 1 $TAO token. Approximately 7200 TAO are generated every day. These rewards are now distributed to each subnet based on contribution and then distributed to owners, validators and miners within the subnet.

Figure 3: Bittensor community promotion picture

TAO tokens can be used to purchase and obtain computing resources, data and AI models on the Bittensor network, and are also a certificate for participating in community governance.

Current Development Status

The total number of accounts on the Bittensor network has now surpassed 100,000, with over 80,000 being non-zero balance accounts.

Figure 4: Changes in Bittensor Account Numbers

Over the past year, the price of TAO has surged by several multiples, reaching a market capitalization of $2.278 billion, with the current token price at $321.

Figure 5: TAO Token Price Changes

Gradual Implementation of Subnet Architecture

Bittensor protocol

The Bittensor Protocol is a decentralized machine learning protocol that enables network participants to exchange machine learning capabilities and predictions. It facilitates the sharing and collaboration of machine learning models and services in a peer-to-peer manner.

Figure 6: Bittensor Protocol

The Bittensor Protocol encompasses network architecture, sub-tensors, subnet architecture, validator nodes, miner nodes within the subnet ecosystem, and more. Essentially, the Bittensor network consists of groups of nodes participating in the protocol, with each node running the Bittensor client software to interact with other networks. These nodes are managed by subnets, which operate on a survival-of-the-fittest basis. Poorly performing subnets are replaced by new ones, and underperforming validators and miners within each subnet are also pushed out. Thus, subnets are a crucial component of the Bittensor network architecture.

Subnet Logic

Subnets can be considered independently operating pieces of code that establish unique user incentives and functionalities, while maintaining the same consensus interface as the Bittensor mainnet. Subnets are categorized into local subnets, testnet subnets, and mainnet subnets. Excluding the root subnet, there are currently 45 subnets, with the number expected to grow from 32 to 64 between May and July 2024, adding four new subnets each week.

Subnet Roles and Emissions

The entire Bittensor network includes six functional roles: users, developers, miners, staker validators, subnet owners, and committees. Within a subnet, the roles consist of subnet owners, miners, and staker validators.

1. Subnet Owners: Subnet owners are responsible for providing the base miner and validator code. They can set unique additional incentive mechanisms and allocate work incentives to miners.
2. Miners: Miner nodes are encouraged to iterate their servers and mining code to stay ahead in the competition within the same subnet. Miners with the lowest emissions are replaced by new miners and must re-register their nodes. Notably, miners can operate multiple nodes across multiple subnets.
3. Validators: Validators are rewarded for assessing each subnet’s contributions and ensuring their correctness. They can also stake TAO tokens in validator nodes, earning a staking reward of 0-18% (adjustable).

Subnet Emissions refer to the mechanism within the Bittensor network that distributes TAO tokens as rewards to miners and validators. Typically, the emission rewards within a subnet are designed to allocate 18% to the subnet owner, 41% to validators, and 41% to miners. A subnet consists of 256 UID slots, with 64 UID slots allocated to validators and 192 UID slots to miners. Only the top 64 validators with the highest staking amounts can obtain validator permissions and be recognized as active validators within the subnet. A validator’s stake and performance determine their rank and rewards in the subnet. Miner performance is evaluated and scored based on requests and assessments by subnet validators. Underperforming miners are replaced by newly registered miners. Therefore, the greater the total amount of tokens staked by validators, and the higher the computing efficiency of miners, the higher the total emission of the subnet, resulting in a better ranking.

Subnet Registration and Replacement

After registration, a subnet enters a 7-day immunity period. The initial registration fee is 100 $TAO, and the fee doubles upon re-registration, eventually decreasing back to 100 $TAO over time. When all subnet slots are occupied, registering a new subnet will result in the removal of the subnet with the lowest emissions that is not in the immunity period to accommodate the new subnet. Hence, subnets must maximize the staking amount within UID slots and miner efficiency to avoid being deleted after the immunity period ends.

Figure 7: Subnet Name

Benefiting from the sub-network architecture of the Bittensor network, the decentralized AI data network Masa was implemented and became the first dual-currency reward system in the Bittensor network, attracting US$18 million in financing.

Figure 8: Promotion of Masa

Consensus and Proof Mechanisms

The Bittensor network incorporates various consensus and proof mechanisms. In traditional decentralized networks, miner nodes typically use PoW (Proof of Work) to ensure their contribution to the network, earning rewards based on their computational power and data processing quality. Validator nodes often operate under PoV (Proof of Validation) mechanisms, which ensure network security and integrity. However, within the Bittensor network, an innovative PoI (Proof of Intelligence) mechanism, combined with the Yuma Consensus, is used to achieve validation and reward distribution.

Proof of Intelligence Mechanism

Bittensor’s PoI mechanism is a unique validation and incentive system that measures participants’ contributions through the completion of intelligent computational tasks. This ensures the network’s security, data quality, and efficient use of computational resources.

Miner nodes prove their work by completing intelligent computational tasks, which may include natural language processing, data analysis, machine learning model training, etc.

Tasks are assigned by validators to miners, who then complete the tasks and return the results to the validators. The validators assess the quality of the task completion and assign scores accordingly.

Yuma Consensus

The Yuma Consensus is the core consensus mechanism within the Bittensor network. After validators score the completed tasks, the scores are input into the Yuma Consensus algorithm. In this algorithm, validators with a higher staked amount of TAO have more weight in their scores. The algorithm filters out results that deviate significantly from the majority of validators. Finally, the system allocates token rewards based on the aggregated scores.

Figure 9: Consensus Algorithm Illustration

1. Data Agnosticism Principle: This principle ensures privacy and security during data processing. Nodes can complete computations and validations without needing to understand the specific content of the data they are handling.
2. Performance-Based Rewards: Rewards are allocated based on the performance and contributions of nodes, ensuring efficient and high-quality computational resources and data processing.

MOE Mechanism Collaboration

Bittensor integrates the MOE (Mixture of Experts) mechanism within the network, which incorporates multiple expert-level sub-models into a single model architecture. Each expert model has a relative advantage when handling specific domain issues. Therefore, when new data is introduced into the model architecture, different sub-models can collaborate, resulting in better outcomes than a single model could achieve.

Under the Yuma Consensus mechanism, validators can also score and rank expert models based on their capabilities, distributing token rewards accordingly. This incentivizes the optimization and improvement of models.

Figure 10: Problem-Solving Approach

Subnet Projects

As of the time of writing, the number of registered subnets in the Bittensor network has reached 45, with 40 of them officially named. In the past, when the number of subnets was limited, competition for subnet registration was intense, with registration prices soaring to as high as one million USD. Currently, Bittensor is gradually opening up more slots for subnet registration. Newly registered subnets may not match the stability and model effectiveness of those that have been operational for a longer period. However, due to the subnet elimination mechanism introduced by Bittensor, this process will, in the long run, favour the survival of the fittest. Subnets with poor model performance and insufficient capabilities will struggle to survive.

Figure 11: Bittensor Subnet Project Details

Excluding the root subnet, subnets 19, 18, and 1 have garnered significant attention, with emission shares of 8.72%, 6.47%, and 4.16%, respectively.

Subnet 19

Subnet 19, named Vision, was registered on December 18, 2023. Vision focuses on decentralized image generation and inference. This network provides access to the best open-source LLMs, image generation models (including those trained on Subnet 19’s datasets), and other miscellaneous models, such as embedding models.

Currently, the registration fee for a Vision subnet slot is 3.7 TAO. The total 24-hour node revenue is approximately 627.84 TAO, and nodes have reclaimed about 64.79 TAO in the past 24 hours. If newly registered nodes reach the average performance level, daily earnings could be as high as 2.472 TAO, equivalent to approximately $866.

Figure 12: Vision Subnet Registration Fee Data

Currently, the total reclaimed node value for the Vision subnet is approximately 19,200 TAO.

Figure 13: Vision Subnet Reclaimed Fees

Subnet 18

Subnet 18, named Cortex.t, was developed by Corcel. Cortex.t is dedicated to building a cutting-edge AI platform that provides users with reliable, high-quality text and image responses through an API.

Currently, the registration fee for a Cortex.t subnet slot is 3.34 TAO. The total 24-hour node revenue is approximately 457.2 TAO, and nodes have reclaimed about 106.32 TAO in the past 24 hours. If newly registered nodes reach the average performance level, daily earnings could be as high as 1.76 TAO, equivalent to approximately $553.64.

Figure 14: Cortex.t Subnet Registration Fee Data

At present, the total reclaimed node value for the Cortex.t subnet is approximately 27,134 TAO.

Figure 15: Cortex.t Subnet Reclaimed Fees

Subnet 1

Subnet 1 was developed by the Opentensor Foundation and is a decentralized subnet specialized in text generation. As the first project under the Bittensor subnet, it initially faced significant skepticism. In March of this year, Taproot Wizards founder Eric Wall labeled Bittensor’s TAO token as a “meme coin” in the AI space and criticized Subnet 1 for generating similar results across hundreds of nodes when answering text-based questions, failing to effectively improve problem-solving outcomes.

Others

In terms of model categories, Subnets 19, 18, and 1 all belong to the generative model category. Additionally, there are data processing large models, trading AI models, and others. For example, Subnet 22, Meta Search, analyzes Twitter data to provide market sentiment, and Subnet 2, Omron, optimizes staking strategies through deep neural network learning.

From a revenue-risk perspective: If a newly registered node can successfully operate for several weeks, it offers substantial revenue potential. However, if the node cannot employ high-performance GPUs and optimize local algorithms, it will be difficult to survive in competition with other nodes.

Future Development

In terms of popularity: The AI concept is as hot as the Web3 concept, if not more so, with much of the capital that might have flowed into the Web3 industry now being attracted to AI. Therefore, Web3+AI is likely to remain a market focus for a long time.

From a project architecture perspective: Bittensor is not a traditional VC-backed project; since its launch, it has increased in value several times over, supported by both technology and market demand.

From a technological innovation perspective: Bittensor has broken the past pattern of Web3+AI projects working independently. Its innovative subnet architecture can lower the barriers for AI-competent teams to migrate to decentralized networks and quickly generate revenue. Additionally, due to the competitive elimination mechanism, subnet projects must continuously optimize models and increase staking to avoid being replaced by new subnets.

From a risk perspective: As Bittensor increases the number of subnet slots, it inevitably lowers the registration threshold, raising the possibility of low-quality projects entering the mix. At the same time, as the number of subnets increases, the TAO rewards for previously registered subnets will gradually decline. If the TAO token price does not rise in line with the number of subnets, returns may fall short of expectations.

Disclaimer：

1. This article is reprinted from [PANews]. All copyrights belong to the original author [rustless Labs]. 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.