AI and Blockchain: Key Takeaways from 2024 and Industry Forecasts for 2025

According to CoinMarketCap, the total market capitalization of tokens in the AI category reached $62 billion by mid-December, peaking at $72 billion in early December. At the start of 2024, this figure was $22 billion, meaning the AI sector grew by 180% over the year.

Let’s take a closer look at the rankings of the largest AI tokens. The top two spots are held not by AI protocols but by L1 blockchains: NEAR ($8.4 billion) and ICP (Internet Computer, $6.4 billion). What are they doing here?

NEAR started as a general-purpose non-EVM blockchain focusing on games, DeFi, NFTs, and similar applications. However, in 2024, the network rebranded itself as a blockchain optimized for AI projects. NEAR even announced plans to create the world's largest open-source AI model with 1.4 trillion parameters.

The second project on the list is the L1 network Internet Computer, which was previously positioned as a global decentralized network for building any type of application. It, too, has shifted its focus toward AI.

In third place is Render Network (RNDR), a decentralized GPU network for tasks like video rendering, game visual effects, and more. On Render's service marketplace, there are AI solutions from third-party studios, including «smart» rendering, image generation, and similar tools.

Other notable entries in the top ten include:

• Injective Protocol – a blockchain for DeFi that announced the integration of AI tools.
• Filecoin – a decentralized file storage network.
• The Graph – a blockchain data indexing protocol.

As a result, six of the ten largest tokens on the list—accounting for $29.5 billion, or 47% of the sector’s total capitalization — cannot truly be considered pure AI tokens.

However, one interesting conclusion can be drawn: rebranding with an AI focus proved to be an effective and popular strategy for attracting attention to existing projects in 2024.

At the top of the rankings for token price growth over the year, Virtuals Protocol leads by a massive margin. This new AI platform for creating AI agents on the Base blockchain was launched in March 2024. In just nine months, the project managed to break into the top ten largest AI protocols by market capitalization.

Among long-standing projects in the top ten, the best performance came from NEAR, with nearly 200% growth.

1. VIRTUAL 4572%
2. NEAR 196%
3. FET 152%
4. THETA 144%
5. ICP 111%
6. RENDER 102%
7. GRT 60%
8. TAO 51%
9. FIL 36%

Among the tokens in the second tier, it’s worth highlighting the success of another new project, Grass Markets, which allows users to «rent out» their home internet. Since its listing on October 28, the $GRASS cryptocurrency has grown by more than 400% (from $0.64 to $3.30).

When discussing AI projects in Web3, it’s important to understand that modern AI models cannot operate directly on existing blockchains. Writing data to blockchains is too expensive, and transaction processing takes too long. True «on-chain AI» is unlikely to ever become a reality. For instance, the ICP project recently launched the first on-chain AI model running entirely on smart contracts, but…

So, what real advantages can decentralization offer the AI industry compared to the traditional centralized approach of giants like Google and Microsoft? Are AI and blockchain truly made for each other, or is this just a trendy narrative? Let’s analyze two popular approaches from 2024: the decentralization of computations and data for AI models, and applied AI for gaming and trading.

Decentralized Computing

Arguably the most common use case for Web3 projects in the AI space is the decentralization of computing and data storage. Training AI models requires enormous computational resources, memory, and energy. For example, GPT-4 required thousands of state-of-the-art processors and 7.2 gigawatt-hours of electricity over six months — equivalent to the energy consumption of some small countries.

Such resources are only available to the largest corporations, and the AI products they create are centralized, with all the associated consequences, such as extensive user data collection.

An alternative could be decentralized networks, where independent participants provide their computing resources to AI model developers in exchange for rewards. Blockchain is used here to transparently and decentrally record and distribute these rewards in tokens.

Another form of «decentralized AI» is networks of AI nodes or agents that collaboratively work on complex tasks.

Examples of decentralized computing and resource protocols include:

• Bittensor (TAO)
• Render Network (RENDER)
• Fetch.AI (FET)
• Theta Network (THETA)
• Akash Network (AKT)
• Aioz Network (AIOZ)
• Livepeer (LPT)
• Io.net (IO) and others

Web3 Gaming

Another intersection of Web3 and AI is gaming. In this case, it’s more about using artificial intelligence to optimize gameplay rather than decentralizing AI tools themselves.

Until mid-2024, $OCEAN (Ocean Protocol) and $AGIX (Singularity.NET) were regulars in the rankings of top AI tokens. However, try searching for them on major crypto exchanges now—you won’t find them. That’s because, as of July 1, these two protocols merged with Fetch.ai ($FET).

The new project is called Artificial Superintelligence Alliance, and it will eventually launch a new token, $ASI. For now, however, $FET serves as the official token of the entire alliance.

The stated goal of the «Artificial Superintelligence Alliance» is to accelerate the development of decentralized AI and create an alternative to AI systems controlled by large corporations.

For reference, here’s what the three founding companies of the alliance specialize in:

• Fetch.ai: A platform for creating AI agents (semi-autonomous AI applications) designed to collaborate with one another to solve shared tasks.
• Ocean Protocol: Data tokenization. Owners of datasets valuable for AI model development can publish and monetize their data on the platform.
• Singularity.NET: A suite of tools for creating and monetizing AI solutions.

The AI Sector Awaits Another Rally – the Last in the Cycle

The «golden hour» for AI tokens in 2024 was during March and April, while their performance during the December rally was average. Instead, the spotlight shifted to DeFi projects and L1 networks. For example, let’s take a look at the annual price chart of $FET.

The first half of 2025 is expected to bring the third and final rally of the current market cycle, during which AI projects will likely reach new price highs. While 10x gains are unlikely, the prices of major tokens like $FET, $RENDER, and $TAO could realistically double or triple.

We can't offer financial advice, but it's worth noting that rallies typically progress in waves, with investor funds flowing from one narrative to another in order of capitalization: for example, starting with L1s, followed by DeFi, then AI, gaming, and so on. Crypto investors need to be keenly aware of trend shifts during rallies and not miss the optimal exit point.

Rising Popularity of End-User Applications

As mentioned, the top price gainers in the AI sector's top 20 are Virtuals Protocol ($VIRTUAL) and Grass ($GRASS). They managed to push aside serious and ambitious protocols that were launched several years ago. What sets them apart is that they don't aim to overhaul the entire AI industry paradigm. Instead, they deliver accessible — and, most importantly, functional—tools for end users.

• Virtuals Protocol: The protocol uses generative AI to create autonomous characters that can talk, move in 3D spaces, learn, make decisions, and even manage their own crypto wallets. Regular users can purchase a share of ownership in an AI character and earn a portion of the income generated by its use in games, art, advertising, and entertainment. A single character can have joint ownership shared among dozens of users.

• Grass: Users share a portion of their home internet bandwidth with companies (including those in the neural network sector), accumulate points, and earn $GRASS tokens, while the platform guarantees the security of their personal data. The project already has 2.5 million active users, and although its services are utilized by a wide range of organizations, its PR materials emphasize AI—hence the inclusion of $GRASS in AI token rankings.

It can be assumed that in the future, there will be more applied projects with clear profit mechanisms for end users, especially on blockchains with low fees. There are many potential use cases: generating and selling AI agents, integrating them with NFTs, creating memecoins and content using AI agents, assisting in trading, and more.

AI Projects Will Conquer TON and Other Alternative L1s

Virtuals Protocol operates on the Base blockchain, while Grass is on Solana. In 2025, we will likely see an increase in the number of user-friendly dApps with AI technology elements on chains already popular for everyday tasks and entertainment (gaming, buying memecoins, NFTs, DEX trading in small amounts, etc.), not just for investment purposes.
In this context, the TON ecosystem is worth watching. It’s currently experiencing rapid growth due to the spread of TMA (Telegram Mini Apps) integrated into Telegram. For example, a significant portion of memecoin trading already takes place through the Telegram bot Banana Gun Bot. It’s logical to assume that AI apps—or at least TMAs claiming a connection to AI to promote their products—will soon emerge on TON.
For now, AI hasn’t fully arrived on TON, and the projects that take the lead in this space will gain a crucial market advantage.