AI Tokens

Exploring the AI token market through the lens of emergent decentralised AI solutions of today.

Mar 01, 2024 Decentralised AI Tokens

Research Disclaimer Research and Insights disclaimer for Research reports

Key Takeaways

AI Sector Market Overview

The year 2024 is welcomed with fresh developments and new milestones in the artificial intelligence (AI) industry, which has contributed to the resurging optimism in the space. A few key events are fuelling the booming AI sector: OpenAI unveiled its novel text-to-video AI model Sora, an AI-powered content generator that can produce high-quality and dynamic video content with just simple text prompts, and OpenAI founder Sam Altman’s Worldcoin token (WLD) recorded double-digit gains during the reporting period; Mistral AI entered into a partnership with Microsoft, following the launch of its proprietary large language model (LLM) Mistral Large, which is considered to closely compete with GPT-4 and other LLMs. Additionally, the latest NVIDIA earnings report beat Wall Street’s earnings forecast and has surpassed Amazon and Alphabet in market capitalisation. Decentralised AI Tokens

Following these AI developments, the AI token space has also been rallying. The market capitalisation of AI-related tokens surpassed US$10 billion and surged by 74% year-to-date. Leading the pack is decentralised AI marketplace SingularityNET (AGIX), which has seen a substantial gain of +193.52% month-on-month as of 25 February. Making up the majority (over US$3.6 billion) of market cap is Bittensor (TAO), an open-source protocol that utilises blockchain technology to create a decentralised machine learning network. Decentralised AI Tokens

In this current cycle, the intersection of blockchain and AI mainly falls into the categories of decentralised compute, zero-knowledge machine learning (zkML), and decentralised AI Agent. A non-exhaustive list of projects, along with other emerging sub-categories, is mapped out below.

3 Special Crypto Ai Landscape

Decentralised Compute

AI models require significant computational resources like graphical processing units (GPUs) to be trained and run inferences. However, the compute space has been characterised by scarce resources and limited competition, with key players like Amazon, Google, and Microsoft holding a firm grip on the market. Scarce compute resources have resulted in rising variable costs associated with training and inferencing AI models — and are estimated to continue increasing over the years. As a result, segments like decentralised compute marketplaces have emerged, connecting providers with those who are looking to lease high-performance compute resources. These marketplaces unlock a significant amount of new GPU supply and computational efforts, enabling anyone in the world to become a resource provider. Decentralised Compute Tokens
  • Render Network (RNDR) has one of the largest distributed GPU networks globally, with over 100,000 node operators on its waitlist. Launched in 2020, the network facilitates advanced AI and 3D rendering capabilities, thanks to its decentralised GPU offering invaluable compute resources. 

    It harnesses unused GPU cycles by connecting creators (customers) in need of computation power with Node Operators (GPU rendering service providers) with available GPU power. In return, providers earn Render’s native utility token, RNDR Token. An ERC-20 token, RNDR is pivotal to the network’s operations, required by both creators and Node Operators. 
  • Powered by the Proof of Intelligence consensus mechanism, Bittensor (TAO) is a peer-to-peer decentralised protocol facilitating machine learning collaboration and incentivising the production of machine intelligence. As part of its Revolution network upgrade, it introduced subnets, which are specialised networks dedicated to a specific machine learning use case or resource provision.

    Bittensor’s native token is TAO, which is used for governance, staking, and as payment for accessing AI services and applications within the Bittensor ecosystem. Users who contribute quality data or computational resources to the network, as well as subnet participants, are rewarded with TAO tokens.
  • Akash Network (AKT) is an open-source decentralised cloud computing platform, providing decentralised processing and storage to users seeking alternatives for Google and Amazon Web Services (AWS). It features Akash ML, a supercloud for AI designed to provide decentralised and open access to the most sought-after commodity for machine learning: GPUs. In December 2023, Akash unveiled Akash Chat, its own version of a ChatGPT alternative that runs on an open-source version of the ChatGPT model (Mistral-7B open-source large language model). 

    AKT is Akash Network’s multi-utility token used for governance, security, incentives, and value exchange within the network. 
  • Golem Network (GLM) is a decentralised compute marketplace, establishing a peer-to-peer network where application owners and users can rent resources of other providers’ machines. Part of its vision is to create a global supercomputer that will offer purchasable computing power through a globally distributed network of computers. To demonstrate Golem’s capabilities, researchers were able to rent out 20,000 CPU cores to simulate 11 billion-plus chemical reactions. 

    Golem is amplifying its efforts to expand its ecosystem and to develop a decentralised physical infrastructure for open-source developers and AI companies, mainly through its GPU Beta Testing Program.

    The GLM token plays a vital role in the network, as it is the main payment method and the store of value on the network and is used in the reward system to incentivise providers.

AI Agent

The integration of AI agents with blockchain technology enables the automation and optimisation of processes in a decentralised and efficient manner. AI agents are trained autonomous bots that utilise AI models to execute requests on behalf of a user. Within the space of cryptocurrency and blockchain, they play a crucial role in various applications, such as fraud detection, risk management, market analysis, smart contract optimisation, identity verification, bug identification in code, and many more. Decentralised AI Tokens
  • (FET) is a blockchain platform that uses AI to help automate everyday tasks and optimise solutions to everyday problems through intelligent data sharing, machine learning, and AI. It offers a smart wallet serving as a machine learning platform and an AI agent-based trading tool, which is integrated with OpenAI’s GPT API. It also introduced uAgents Framework, an open-source framework designed for the development of decentralised AI agents. 

    FET is the network’s utility token, used for paying services in the Fetch ecosystem and network transaction fees, securing the network via staking, and incentivising data distributors. 
  • SingularityNET (AGIX) is a decentralised marketplace designed for ‘self-organising’ intelligence networks. It envisions a future where a network of AI agents can seamlessly interact with each other, enabling developers to publish and monetise their AI algorithms easily with its public API and decentralised marketplace. 

    AGIX is the network’s native token, used for governance, staking, incentives, and payment within the SingularityNET network.
  • Autonolas (OLAS) is a unified network for off-chain services, automation, relayers, and co-owned AI. Operating through the Open Autonomy framework, it aims to provide a foundation for off-chain autonomous applications, coordinated by the OLAS token and powered by autonomous agents.

    The OLAS token serves as the native token and cornerstone of the Autonolas ecosystem. It acts as a coordination mechanism within the Autonolas protocol, facilitating decentralised governance and enabling on-chain registration of autonomous services and components.


ZkML is a field that combines machine learning, advanced cryptography, and decentralisation. In an era where AI-generated content increasingly resembles content created by humans, the potential use of zero-knowledge cryptography has emerged as a way to determine the origin of specific content. By applying a particular model to a given input and creating a zero-knowledge circuit representation, outputs of large language models like GPT4 or text-to-image models like DALL-E 2 can be verified. The advantage of zero-knowledge proofs is that they allow for parts of the input or the model to be hidden, if necessary.

For instance, in sensitive data scenarios, a machine learning model can be applied to data without revealing the input to any third party. This approach, known as zkML — a field that combines machine learning, advanced cryptography, and decentralisation — focuses on creating zero-knowledge proofs specifically for the inference step of the machine learning model, rather than the model training process.

At the forefront of this is zkML-focused startup Modulus Labs, which develops zkML solutions that enable trustful integration of AI outputs into blockchain systems without revealing sensitive data or model details. It is currently building on several use cases, including Rocky Bot app (on-chain verifiable ML trading bot), Lyra Finance (enhancing its AMM with intelligent features), and Astraly (transparent AI-based reputation system). By building a ZK coprocessor for Ethereum, Axiom enables reliable machine learning computation for on-chain data, improving user access to blockchain data and providing more sophisticated views of on-chain data in the process.


We can expect more integration between AI and crypto, as new AI technologies emerge and the industry itself continues to scale. We are already starting to see the potential of this integration within the cryptocurrency space, from enhanced user experience, improved developer tools and environment, innovations in smart contract functionalities, and automation, amongst others. Looking ahead, as emergent solutions build the current narrative of the market, new challenges can influence its direction in the long run.

In terms of verticals, supply constraints will continue to drive the decentralised compute market forward, in parallel with the advancements of AI technologies. The permissionless and trustless foundation that cryptocurrency provides will allow for more experimentation in the AI agent and agent provider space. Finally, for zkML, it is still in its early stages, and developments are focused on building the infrastructure and tooling to integrate the solution.

Authors Research and Insights Team

Get the latest market, DeFi, and NFT updates delivered to your inbox:

Subscribe to newsletters  

Be the first to hear about new insights:

Follow us on X (formerly known as Twitter)


ai tokens

Artificial Intelligence


Share with Friends

Ready to start your crypto journey?

Get your step-by-step guide to setting up an account with

By clicking the Get Started button you acknowledge having read the Privacy Notice of where we explain how we use and protect your personal data.
Mobile phone screen displaying total balance with App

Common Keywords: 

Ethereum / Dogecoin / Dapp / Tokens