Author: Grayscale
Compiled by: Felix, PANews
Summary
Bittensor is at the forefront of two of the most groundbreaking and transformative trends in software: blockchain and artificial intelligence (AI). While Bitcoin helped create the crypto industry as the first peer-to-peer currency system and digital store of value, and Ethereum helped expand the ecosystem through decentralized applications, Bittensor represents a new and unique use case that plans to leverage the properties of permissionless public blockchains and economic incentives to develop advanced AI software through an open, decentralized community (not a centralized company).
Today, AI development is highly centralized, with a great deal of power concentrated in the hands of a few large tech companies. As AI grows into a more powerful and important tool, there is a risk that AI will be controlled by a few entities, which is contrary to human values and wider society. In contrast, Bittensor is a platform that economically incentivizes open collaboration in AI development through its native token TAO. By using a public blockchain, Bittensor may help democratize ownership, increase transparency in AI systems, and align decisions about AI development with the interests of society. Bittensor aims to create the "Internet of AI," envisioning a future with many interconnected AI ecosystems or subnetworks forming a global decentralized AI platform. By connecting to the Bittensor network, the platform will help anyone, anywhere easily build, deploy, and access AI applications.
Figure 1: As of August 16, TAO accounts for 12% of Grayscale’s AI domain
Tokens
TAO is the native token of the Bittensor network. Owning TAO tokens represents ownership of a portion of the ecosystem (Figure 2). TAO's supply plan is exactly the same as Bitcoin, with a maximum supply of 21 million coins, which is halved approximately every four years. Bittensor's first halving event is expected to be in August 2025.
Bittensor aims to apply Bitcoin-style incentives to AI development, using TAO tokens as incentives for network participants to perform their intended functions. These participants include network validators and subnet owners, subnet validators, and subnet miners. In addition to incentive rewards, TAO is currently used primarily as a deposit for subnet owners to register their subnets. In the future, as the nascent Bittensor network matures, potential other use cases for TAO include (i) as gas fees for network transactions, (ii) as decision-making power for allocating TAO-issued subnets, and (iii) general network governance decisions. In the long term, Bittensor may monetize the network by charging application end users for using its subnets, which may generate value for TAO tokens.
Figure 2: Basic information of TAO tokens
Network and Technology
On Bittensor, developers compete to develop the best AI models in exchange for TAO rewards. The system supports a range of AI-related services, including chatbots, video generation, deep fake detection, storage, and computing. To democratize AI development, Bittensor allows AI researchers and independent open source developers to monetize their innovations and potentially contribute to a more equitable distribution of AI benefits.
Bittensor employs various subnets that are specialized to perform different machine learning tasks. For example, one subnet is specialized for AI image generation, another for AI music generation, and another for detecting AI-generated deepfakes. Each subnet involves three main types of participants: subnet owners, subnet miners, and subnet validators. On a given subnet, miners compete to be the “best” output, while validators evaluate which miners perform the “best” (see below). While elements of this process vary from subnet to subnet, the general idea is outlined below:
How it works
End users prompt the network through consumer-facing applications. This is similar to asking ChatGPT a question.
Subnet miners run AI models on the relevant subnet and compete to generate the best output for a given prompt. For example, in the chatbot subnet, miners compete to provide the best answer to a user question.
The validator ranks the miners’ responses based on output quality and returns the highest ranked responses to the end user who asked the question.
Validators determine miner performance through a new process called Yuma consensus. This consensus mechanism aggregates the ranking of each validator and weights them according to the amount of TAO they stake to generate a collective ranking list of miner performance.
The wider Bittensor blockchain operates under a "proof of authority" consensus mechanism, where certain nodes are granted the authority to order on-chain transactions and help maintain the integrity of the network. Bittensor's block storage updates state changes and token balances to reflect new releases by network validators as well as subnet owners, miners, and validators.
Use Cases
Bittensor has a wide range of potential use cases, with each subnet representing a different example. These include:
Image Generation Subnet: Tailored for AI models specifically designed to create high-quality generated images.
Chatbot Subnet: Optimized for AI models specialized in natural language processing and allows consumers to access responsive virtual assistants.
Deepfake Detection Subnet: Leverages advanced generative and discriminative AI models in Bittensor networks and is designed to detect AI-generated images.
Several of Bittensor’s direct competitors in the decentralized AI solutions space are addressing AI development in general. For example, the Allora Network focuses on AI development in the financial services space, providing an automated trading strategy platform for decentralized exchanges and prediction markets. Other early projects that attempt to address decentralized AI at the infrastructure level include Sentient and Sahara AI.
In addition to these direct competitors, some protocols compete with specific Bittensor subnets. For example, Akash competes to some extent with the computing subnet, Filecoin competes with the data storage subnet, and Gensyn competes with the pre-training and fine-tuning subnet. However, some well-known AI companies (such as Wombo and MyShell) and crypto teams (such as Masa, Kaito, and Foundry) have established their own subnets.
Factors to consider
Market opportunity with growth potential: The market size of centralized AI is estimated at $215 billion in 2024, with a projected compound annual growth rate of 35.7%. Grayscale believes that Bittensor represents a new and unique use case in cryptocurrency. Decentralized AI is valued at only $19 billion, reflecting its nascent stage. In an era when a few technology companies seem to control AI, Bittensor represents an early investment in this intersection.
Permissionless development and use of powerful technology: As AI continues to evolve into a more powerful and important tool, there may be increasing regulations or restrictions on who can build or access these applications. Bittensor offers an alternative to permissionless access to resources to develop and use AI.
Economic incentives to promote fair AI development: Bittensor can help independent AI developers gain greater access to AI resources such as compute, storage, and data compared to centralized alternatives. It can also help AI researchers and open source AI developers monetize their contributions and potentially fund their operations. If successful, Bittensor's open and distributed ecosystem can help balance the closed-source models developed by tech giants and help ensure that the economic benefits of AI are shared more broadly.
Growing Popularity and Recognition: Bittensor has gained early traction, with over 40 subnets dedicated to specific AI tasks, and has received recognition from prominent tech and AI leaders. Companies are raising venture capital funding to build subnets and applications on Bittensor, which shows that investors and developers are increasingly interested in the ecosystem and that Bittensor has the potential to expand network effects.
Investment Risks
Adoption and Network Growth: Bittensor’s longevity depends on attracting a large number of developers and AI projects to build on the platform. If Bittensor fails to achieve mass adoption, the network may struggle to reach its full potential. Additionally, given the network’s nascent age, most network resources are focused on infrastructure-level and subnet activity. Over time, Bittensor will need to expand the number and quality of application end users to help increase token value accumulation and its relevance to everyday consumers.
Decentralization and network resilience: Bittensor's operations rely on the smooth operation of a broad network of participants. Any disruptions, such as technical glitches, errors, or network attacks, could affect its performance and reputation. Bittensor also needs to increase overall network decentralization and distribute the voting power released by TAO more widely across the network.
Implementation of incentive design: To realize its full potential, Grayscale believes that Bittensor needs to ensure that subnet owners design the right incentive mechanisms for their subnets and ensure that the network will release appropriate allocations to appropriate subnets over time.
Competing networks: Bittensor faces competition from AI-related crypto assets that attempt to address AI development through token incentives, such as Allora, Sentient, Sahara AI, and other assets covering a variety of AI-related use cases, such as Filecoin and Gensyn. This list is also likely to grow over time as this intersection matures.
Related reading: Bittensor: How does AI subnet reshape collective intelligence networks?
