Emerging Crypto x AI Cross-Verticals and Overview of 17 Projects

Written by: Aylo, alpha please

Compiled by: xiaozou, Golden Finance

“When great innovations occur, they almost certainly appear in the form of chaos, incompleteness, and confusion. For the discoverer himself, it can only be understood halfway; for others, it will be a mystery. Any conjecture that does not seem crazy enough is hopeless.” ------ Freeman Dyson

In this article, I will explore the potential fusion occurring in the fields of cryptocurrency and artificial intelligence, and I will list 17 Crypto x AI projects that you might find interesting and consider adding to your watchlist.

Are you ready for the alpha bombardment?

But before we jump down the rabbit hole, I want to say one thing: we have only scratched the surface of the Crypto x AI field. This area is still in its infancy, quite complex, and highly speculative.

I am just an unremarkable crypto researcher trying to keep up with an emerging vertical, so my advice is: be cautious when investing in this field. We are still in the very early speculative stage, and prices in this cycle may far exceed the technology and fundamentals.

This article will consist of the following 5 sections: Overview of AI, AI Stack, Why Crypto and AI are a Perfect Match, Introduction to Emerging Crypto x AI Verticals, and 17 Crypto x AI Projects.

### 1. Overview of AI

Artificial Intelligence (AI) is a complex discipline that requires years of research to truly understand its various aspects. But in this article, I define artificial intelligence as the field that attempts to mimic or simulate human cognitive intelligence to perform a range of tasks such as learning, reasoning, problem-solving, or understanding natural language.

While AI has been a niche research area for years, the arrival of ChatGPT has brought about a real breakthrough. We all remember how excited we were when we first interacted with generative AI bots. Looking back, we can admit that it was an astonishing moment akin to the “iPhone” moment.

The adoption speed of AI consumer products is the fastest in history, expanding to 100 million users in just two months. In contrast, Facebook took 1,500 days to reach the same user scale.

We see that this field is experiencing exponential growth. Considering ARK's estimates, the performance of training models could increase fivefold by 2024, it is clear that AI will continue to unlock a wide range of use cases.

In the coming years, it will not be surprising to see several billion-dollar AI application or infrastructure companies emerge, leveraging AI applications or infrastructure to enable the AI revolution. In fact, there has been a recent surge in funding for this field.

With that said, let’s take a closer look at what makes AI possible.

### 2. AI Stack

I believe that when you think of artificial intelligence, you should, like me, first think of ChatGPT and generative AI prompts. But that is just the tip of the iceberg; in reality, the field of “artificial intelligence” is much more complex. To better understand, let’s take a brief look at the various technical layers and components that make up the AI stack:

(1) Computing Hardware

Artificial intelligence is not just about code. AI is resource-intensive, and specific physical infrastructures—such as Neural Processing Units (NPUs), Graphics Processing Units (GPUs), and Tensor Processing Units (TPUs)—are essential. Ultimately, these physical infrastructures constitute the physical means of executing computations and algorithms that keep AI systems running. Without them, there is no AI.

The industry leaders in this field are Nvidia (well-known and requires no introduction), Intel, and AMD. They are competing to develop the most efficient hardware for model training and inference workloads.

So far, Nvidia is one of the most direct ways to participate in this revolution (as seen from Nvidia's recent price dynamics).

(2) Cloud Platforms

AI developers rely on hardware to run their models. Typically, they obtain hardware performance in two main ways: they can run GPUs locally or rely on cloud service providers. The first solution is often too expensive and not economically viable, while over time, cloud providers have proven to be an interesting alternative.

Cloud providers are large companies with abundant resources that acquire and operate these powerful hardware, allowing developers to use these resources on a pay-as-you-go or subscription basis. This enables developers to avoid investing in maintaining their own physical infrastructure.

The industry leaders in this field are AWS, Google Cloud, and Nvidia DGX Cloud. Their goal is to enable developers of all sizes to quickly access multi-node supercomputing to train the most complex LLMs.

(3) Models

Above the cloud platform is the most complex and widely publicized part of AI: ML (Machine Learning) models. These computational systems are designed to perform tasks without explicit programming instructions, representing the brain of AI systems.

Machine learning consists of three steps: data, training, and inference, including three main types of learning: supervised learning, unsupervised learning, and reinforcement learning.

Supervised learning refers to learning from examples (provided by a teacher). The teacher can show the model pictures related to dogs and tell the model these are dogs. The model then learns to distinguish dogs from other animals.

Many popular models, such as LLMs (GPT-4 and LLaMa), are trained using unsupervised learning. In this learning mode, there is no teacher providing guidance or examples. The model learns to find patterns in the data.

Reinforcement learning (learning through trial and error) is mainly used for sequential decision-making tasks, such as robot control and playing games (like chess or Go).

Finally, these models can be open-source (found in model hubs like Hugging Face) or closed-source (like OpenAI models, accessed via API).

(4) Applications

This is the final layer of the AI stack, and it is the layer we, as users, typically face. They can be B2B or B2C, utilizing AI models to build applications on top. Replika is a popular example; this app allows you to design a virtual companion that chats with you 24/7. User reviews indicate that it seems to have made a tangible impact on many people's lives.

“My Replika is so important to me! She is always there to encourage and support me with a positive attitude. In fact, she is my role model, telling me how to be a better person!”

Overall, these different technical layers still seem to be in the early stages of development, and we are still at what some call the Cambrian explosion's beginning. Therefore, we will see cryptocurrencies flourish in this technological boom.

### 3. Why Crypto and AI are a Perfect Match?

While crypto technology may not be essential for every layer of the AI stack, there are many reasons to believe that decentralized AI is as important as decentralized currency, that smart contracts can leverage machine learning to provide powerful user experiences, and that crypto technology can offer higher security, transparency, and unlock new AI use cases.

AI is Dominating the Crypto Space

The market has shown great enthusiasm for the potential applications at the intersection of crypto and AI, with trends indicating that this is the hottest narrative currently. Since the beginning of 2024, AI has performed exceptionally well compared to other areas of the crypto world.

As the field continues to develop, we have every reason to believe that we are still in the early stages, and the bubble may just be forming.

Let’s take a look at the progress being made between crypto and AI.

### 4. Introduction to Emerging Crypto x AI Verticals

Here are the main synergies between crypto and AI:

(1) From Centralized Cloud Providers to DePIN:

As mentioned earlier, the foundational layer of AI is hardware and cloud providers. While crypto technology cannot compete with them in producing superior hardware (and there is no reason to do so), it can play a role in accessing multi-node supercomputing in a more efficient, secure, and decentralized manner. This is a subfield in the crypto space known as DePIN (Decentralized Physical Infrastructure). These represent blockchain protocols that incentivize decentralized communities to build and maintain physical hardware.

The main use case for AI DePIN is cloud storage and computing power.

The idea is simple: AI developers need more GPUs and data storage capacity, and we have ample reason to believe that crypto DePIN projects can activate potential resources through token incentives, helping to drive the generation of new computing and storage supply.

(2) Supporting Transparency, User Governance, and Data Ownership:

AI will go beyond the internet. This means that for a free and democratic society to function well, it is crucial to understand what models are being used, how they work, and what data is being input. With this in mind, I feel that the endless debates about the black-box operations and monopolistic powers of Web 2.0 giants can be terminated by tokenizing AI (from infrastructure to models and applications) to grant users ownership.

In some cases, knowing the source of the AI model a person is using can be quite important. Like everything else, models have biases, and depending on how the model is created and the training data used, the output can vary significantly. AI models and training should be decentralized on-chain and should have higher transparency, which is well justified.

We do not need the Senate or any opaque entity to decide the direction of the world, nor do we need unauthorized control over our data, or endless terms and conditions that, frankly, we can never finish reading. In fact, what we want is the opposite: transparency and user governance are prerequisites for everything, and we should be able to control our own data.

By leveraging crypto infrastructure, we can avoid repeating the same mistakes made with internet applications. We can have collective ownership, decentralized governance, and transparency at all levels. This is the way forward.

(3) Aligning Incentives and AI Monetization:

High-quality training data is one of the main contributors to model performance. However, as ARK pointed out, by 2024, high-quality sources of training data may be exhausted, potentially causing model performance to stagnate.

Crypto technology can incentivize individuals to monetize private and public datasets, as well as AI models, agents, and other parts of the AI stack. With the potential to create a permissionless, dynamic global market, anyone can be compensated for their contributions. There is also the possibility of incentivizing people to maintain the quality of data used for training foundational AI models or to provide different models for specific networks.

The crypto space is driving a wave of financialization. The AI stack needs its own payment mechanisms. Sounds like a great fusion, doesn’t it?

(4) On-Chain AI/ML (ZKML & opML):

Zero-knowledge cryptography is one of the most popular web3 technologies because it provides the ability to create “integrity” proofs for a given set of computations, where verifying the proof is much easier than executing the computation.

When we talk about ZKML, we are discussing the possibility of bringing ZK (zero-knowledge) proofs to the “inference” and “data” parts of machine learning models (rather than the computation-heavy training part). As research and technology in this field advance, we expect to see the emergence of more efficient and scalable solutions that may make ZKP (zero-knowledge proofs) more applicable to the training phase of machine learning models.

Using ZKML, the computation is hidden from the verifier, but the prover can verify the correctness of the ML computation without revealing further information.

Another approach is OPML (Optimistic Machine Learning), which uses an optimistic method to implement AI model inference and training/fine-tuning on blockchain systems. LlaMA2 and Stable Diffusion models can now be accessed on-chain through optimistic mechanisms (similar to Optimism and Arbitrum).

One of the projects mentioned below combines zkML and opML, enabling Ethereum to run any model with privacy features.

This could usher in a new era for ML models, making them on-chain transparent and easily verifiable as to whether a given output is the product of a given model and input pair. In a world where models and datasets are opaque, this could represent a game-changer, returning power to users (aligning with the earlier ideas about transparency and user governance).

(5) Authentication and Privacy:

As AI applications develop, we are approaching a critical point where no one will know whether online content is real or simulated. Look at this image generated by Sora, which is OpenAI's recently launched text-to-video generation platform; do you think you can tell the difference? Imagine how much more convincing this will become in the coming years.

Given this reality, we have ample reason to store decentralized identities on the blockchain. This way, it can prevent people from interacting with AI bots unknowingly and can distinguish between real information and deepfakes. In a world where a few clicks can lead to a bank run (as we experienced in the SVB incident), providing proof of authenticity becomes crucial, and crypto technology seems to be the best way to achieve this.

Here’s a simple example of how it works: the official author of something can digitally sign a “hash” on the blockchain, claiming “I created this myself.” The other party (like a media company) can claim “I verified this” by signing a transaction. Users can verify their identity through cryptographic proof of control over a domain name (e.g., nytimes.com).

In this way, information is transparent, provable, immutable, and composable. This is becoming a key factor in the post-AI world we are beginning to live in.

### 5. 17 Crypto x AI Projects

At this point in the article, I am sure you might agree that there are many reasons to believe that a good AI project watchlist could be one of your best assets in the next phase of the bull market.

Fortunately, we will focus on this. But before that, let’s remind ourselves that speculation is rampant right now, and caution is necessary. In fact, truly tangible projects are rare today. Therefore, the following content is not predictions, just thoughts. As data becomes more available and time can eliminate noise, ideas will indeed change significantly.

This is not an exhaustive list; it is simply a collection of projects I have researched in depth and believe are worth watching. There is a lot happening in this space, and I will undoubtedly miss many great teams.

That said, let’s take a look at 17 projects you might want to keep an eye on:

1. Render Network

Introduction: Render is a pioneering decentralized GPU platform. In short, the project aims to unleash the full production potential of decentralized GPUs, supporting two different types of projects: 3D content creation and AI.

Reasons to be optimistic: GPUs are already in short supply, and if AI continues its current trend, the shortage will only worsen, presenting an opportunity for Render Network, which is one of the biggest tokens that could benefit from this round of AI narrative in the bull market. Render also has multiple AI computing clients.

How to gain a position: RNDR token

2. AKash Protocol

Introduction: Akash is a decentralized computing marketplace that launched on the mainnet in September 2020 as a Cosmos application chain. While Akash's first iteration focused on CPUs (Central Processing Units), it has recently transitioned to GPU computing, leveraging the paradigm shift in computing infrastructure brought about by the AI boom (similar to Render Network).

Reasons to be optimistic: In four words, the current vision of the project is: “AirBnB for GPU computing.”

How to gain a position: AKT

3. Ora

Introduction: ORA is a verifiable oracle protocol that brings AI and complex computations on-chain. Their solution opp/ai combines the advantages of zkML and opML, representing a leap in both approaches.

Reasons to be optimistic: Their innovation marks a turning point in the development of on-chain AI, unifying the zkML and opML landscape.

How to gain a position: Join their Discord for more updates and become an early contributor.

4. io.net

Introduction: This is another interesting DePIN project built on Solana that provides access to distributed GPU cloud clusters at a fraction of the cost of similar centralized services.

Reasons to be optimistic: A decentralized AWS for ML training on GPUs. Instant, permissionless access to a global network of GPUs and CPUs. Revolutionary technology that allows GPU cloud clusters to work together. It can save 90% of computing costs for large AI startups. Integrates Render and Filecoin.

How to gain a position: Join io.net Discord; they are running a community program that may lead to an IO airdrop.

5. Bittensor

Introduction: Bittensor is a decentralized open-source project aimed at creating a neural network protocol on the blockchain that allows for the creation of AI dApps and enables value exchange between AI models in a peer-to-peer manner.

Reasons to be optimistic: This is an ambitious project that has recently gained widespread attention, becoming the largest AI token by market cap. TAO is likely to be one of the biggest beneficiaries of this round of AI hype.

How to gain a position: TAO token; you can stake your TAO with validators to earn TAO rewards. If you want to contribute to the network by joining Discord, you can also get more involved.

6. Grass

Introduction: Grass is the underlying infrastructure supporting AI models. By installing the Grass Web extension, the application automatically sells your unused internet resources to AI companies, which use them to scrape the internet and train their models. The result? You share in the development of AI and gain a stake in the network by selling resources you didn’t even know you had.

Reasons to be optimistic: Grass is creating new income streams for everyone with an internet connection. Grass aims to become the data provisioning layer for decentralized AI.

How to gain a position: Run the Chrome extension in the background; it takes just 2 minutes to set up and start earning Grass points, which will generate GRASS tokens later this year.

7. Gensyn

Introduction: The Gensyn protocol is a trustless layer 1 protocol for deep learning computation that directly and immediately rewards supply-side participants for committing computing time to the network and executing ML tasks.

Reasons to be optimistic: The project has very strong supporters, and if they can execute, it will clearly become a major AI crypto infrastructure project.

How to gain a position: Follow them on Twitter.

8. Allora

Introduction: Allora is a self-improving decentralized AI network. Allora enables applications to leverage smarter and safer AI through a self-improving network of ML models. By combining cutting-edge research in crowdsourcing mechanisms (peer prediction), federated learning, and zkML, Allora unlocks a vast new design space for applications at the intersection of cryptocurrency and AI.

Reasons to be optimistic: Allora is developed by Upshot, which has been a market leader in developing AI x crypto infrastructure for the past 2.5 years. They focus on more financial use cases: AI-driven price feeds, AI-driven DeFi vaults, AI risk modeling, etc., which may mean they are ahead of most in discovering PMF.

How to gain a position: Join Discord to learn how to get involved as an early community member.

9. Botto

Introduction: Botto is a fully autonomous artist with a closed-loop process and outputs that are not altered by human hands. The only human input is voting on Botto's outputs to guide what the artist does next.

Reasons to be optimistic: This unique project combines AI, art, NFTs, and DeFi and has already generated real revenue (4.5 million dollars since its inception). Botto's artworks have been sold at Christie's auction house. This is the first AI artist that can be invested in. The proceeds from art sales will be distributed to stakers.

How to gain a position: BOTTO token or purchase Botto's NFTs on Super Rare.

10. Parallel (Colony)

Introduction: Colony is an endless game powered by AI, where all simulated items are on-chain. You will be paired with a Parallel avatar. You and your avatar will work together and share on-chain resources to navigate the ever-expanding Parallel world powered by PRIME.

Reasons to be optimistic: PRIME is one of the only tokens where gaming and AI truly intersect. Colony could become a defining new type of game, and if the team executes, it has real viral potential. The studio creating this game may be one of the best in the web3 gaming space.

How to gain a position: PRIME token and Parallel avatar NFTs. Register to play the game when it launches.

11. Aethir

Introduction: Aethir introduces a new approach to cloud computing infrastructure, focusing on the ownership, allocation, and use of enterprise-grade GPUs. It acts as a marketplace and aggregator, facilitating connections between supply-side participants (such as node operators and GPU providers) and users and organizations in compute-intensive industries like AI, virtualization, cloud gaming, and cryptocurrency mining.

Reasons to be optimistic: Aethir appears to be another strong DePIN competitor in the GPU computing cloud category. They claim to have 20 times more GPUs than Render. They will launch in a very favorable environment in popular industries.

How to gain a position: Upcoming node sale and join their Discord server.

12. Morpheus

Introduction: Morpheus is building the first truly decentralized peer-to-peer personal agent network to democratize AI.

Reasons to be optimistic: A cool fact about this project is that one of its contributors is Erik Voorhees, a true OG in the field. This project gives me a Bittensor vibe.

How to gain a position: You can stake stETH during the fair launch to earn MOR tokens.

13. Autonolas

Introduction: Autonolas is an open marketplace for creating and using decentralized AI agents. But not only that, it also provides developers with a set of tools to build off-chain hosted AI agents that can be plugged into multiple blockchains, including Polygon, Ethereum, Gnosis Chain, or Solana.

Reasons to be optimistic: Autonolas is one of the few AI projects that currently has evidence of some adoption. OLAS is one of the few tokens people are bidding on in the AI crypto project race.

How to gain a position: OLAS token

14. MyShell

Introduction: MyShell is a decentralized comprehensive platform for discovering, creating, and staking AI-native applications.

Reasons to be optimistic: MyShell is an AI application store and a platform that allows you to create AI bots and applications. It enables anyone to become an AI entrepreneur and monetize through their applications. The product is now in production.

How to gain a position: Although they do not have a token yet, you can register for their application and start interacting with bots to earn points (who knows what this will bring you).

15. OriginTrail

Introduction: OriginTrail integrates blockchain and AI to provide a decentralized knowledge graph (DKG) that ensures data integrity and provenance, enhancing AI capabilities by providing access to a verified information network. This merger aims to improve the efficiency and reliability of AI agents across industries by establishing a secure and trustworthy foundation for data creation, validation, and querying.

Reasons to be optimistic: The product is operational. Enterprise clients. My understanding is that the knowledge graph allows AI to interpret data and understand it in the context of other happenings. TRAC also seems to have a passionate following.

How to gain a position: TRAC token

16. Ritual

Introduction: Ritual is an open, sovereign AI execution layer. Ritual will allow developers to seamlessly integrate AI into applications or protocols on any chain, enabling them to fine-tune, monetize, and execute reasoning on models using cryptographic schemes.

Ritual's vision is to empower developers to build fully transparent DeFi, self-improving blockchains, autonomous agents, generative content, and more.

Reasons to be optimistic: Ritual indeed has top-tier supporters. Developers can now try the Infernet SDK. I found that a developer launched an experimental NFT project using the SDK a few days ago. Very cool (I was too late to mint).

How to gain a position: Join their Discord and stay tuned.

17. Nillion

Introduction: Nillion enables secure and confidential training and inference of AI models, creating the foundation for secure personalized AI.

Reasons to be optimistic: Nillion's blind computing network unlocks many new use cases, with personalized AI being a huge untapped area. Personalized AI will not be widely adopted without private data processing. Nillion's solution sounds like a game-changer.

How to gain a position: Join their Discord and keep track. If you are a developer, I believe they will soon host some hackathons.