Crypto x AI Stack

The convergence of cryptocurrency and AI is transforming the digital economy, creating the "Agentic Web," where AI agents drive economic activity and power on-chain applications, reshaping how technology and users interact.

TLDR Crypto x AI envisions an "Agentic Web" where autonomous AI agents operate on decentralized blockchain infrastructure. This convergence spans four key layers—compute, data, middleware, and applications—to enhance verifiability, censorship-resistance, and native payment mechanisms for AI. Key beliefs include crypto becoming the go-to payment method for AI transactions, generative AI creating smart contracts, and natural-language interfaces simplifying on-chain interactions. Despite technical and ethical challenges, this synergy promises a more transparent, accessible, and agent-driven digital economy, transforming how humans and machines collaborate.

The future of AI can be built on blockchain technology. Crypto's efficiency, borderless nature, and programmability, combined with AI, have the potential to transform how you and machines interact with the digital economy. This convergence increases accessibility, transparency, and expands use cases in emerging tech, giving you greater sovereignty over personal data. It also introduces the concept of the “Agentic Web,” where AI agents operating on crypto infrastructure drive economic activity and growth.

So what does this look like in practice? Imagine AI agents making transactions on crypto infrastructure. Picture software code generated by AI—such as smart contracts—leading to a surge in onchain applications and experiences. Envision yourself owning, governing, and earning from the AI models to which you contribute. Think about leveraging AI to improve both user and developer experiences within the crypto ecosystem, enhancing smart contract capabilities, and creating entirely new use cases. And this is just the beginning.

Unveiling the Future of Crypto x AI

As you consider the convergence of cryptocurrency and AI, it's clear that blockchain technology won't be necessary for advancing every layer of the AI tech stack. Instead, crypto can significantly enhance distribution, verifiability, censorship-resistance, and native payment mechanisms in AI, while benefiting from AI-driven user experiences on-chain.

The Vision of the Agentic Web

Crypto x AI has the potential to give rise to the "Agentic Web," a transformative paradigm where AI agents operating on crypto infrastructure drive economic activity and growth. In this vision, you might see AI agents transacting autonomously on blockchain networks. These agents could have their own crypto wallets to fulfill user intents, access decentralized compute and data resources at lower costs, or use stablecoins to pay humans and other agents for task completion—all directed toward achieving their goals.

Preliminary Beliefs Underpinning Our Thesis

Several key beliefs shape this thesis:

1. Crypto will become the preferred payment method for transactions between agents and humans, as well as between agents themselves.

2. Generative AI and natural-language interfaces will emerge as the primary way for users like you to transact on-chain.

3. AI will generate the majority of software code, including smart contracts, leading to an explosion of on-chain applications and experiences.

Exploring the Intersection of Crypto and AI

The intersection of Crypto and AI consists of two primary sub-segments:

1. Decentralized AI (Crypto → AI): Building generic AI infrastructure that inherits the properties of modern peer-to-peer blockchain networks.

2. Onchain AI (AI → Crypto): Developing infrastructure and applications that use AI to enhance both new and existing use cases on-chain.

Layers of the Crypto x AI Landscape

The Crypto x AI landscape can be thought of in layers:

1. Compute: Networks that supply GPU resources to AI developers.

2. Data: Networks enabling decentralized access, orchestration, and verifiability of the AI data pipeline.

3. Middleware: Platforms that support the development, deployment, and hosting of AI models and agents.

4. Applications: User-facing products, whether B2B or B2C, that leverage on-chain AI mechanisms.

An Introduction to Crypto x AI

The AI market has seen remarkable growth, with nearly $290 billion in venture capital investments over the past five years. According to the World Economic Forum, AI technologies could boost annual US GDP growth by 0.5-1.5% over the next decade. Applications like ChatGPT-4 are achieving record-breaking user growth and adoption.

As the AI landscape evolves quickly, challenges such as data privacy, a shortage of AI talent, ethical concerns, centralization risks, and the rise of deepfake technology are emerging. These issues are driving discussions at the intersection of crypto and AI, as stakeholders explore how combining these technologies might address these challenges.

AI v Blockchain

Crypto x AI merges blockchain's decentralized infrastructure with AI's capacity to mimic human cognition and learn from data, creating a powerful synergy. Blockchain redefines system architectures, data and transaction verification, and distribution methods.

AI, in turn, enhances data computation, analysis, and content generation capabilities. This convergence has sparked both excitement and skepticism among developers, prompting exploration of novel use cases that could accelerate adoption of both technologies.

While "crypto" and "AI" cover broad ranges of technologies, their intersection can be broken down into two key sub-segments:

• Decentralized AI (Crypto → AI): enhances AI capabilities through crypto's permissionless and composable infrastructure. This unlocks use cases such as democratized access to AI resources (like compute, storage, bandwidth, and training data), collaborative open-source model development, verifiable inference, and immutable ledgers with cryptographic signatures for content provenance and authenticity.

• Onchain AI (AI → Crypto): brings AI's advantages into the crypto ecosystem. It improves user and developer experiences via large language models and natural-language interfaces, and enhances smart contract functionality. Adoption pathways include developers integrating AI models or agents into smart contracts and onchain applications, as well as AI agents leveraging crypto tools (like self-custody wallets and stablecoins) for payments and to commission decentralized infrastructure resources.

Though both segments are still emerging, the potential for "Crypto in AI" and "AI in Crypto" is significant. As compute infrastructure and intelligence speeds improve, these intersections are poised to unlock entirely new use cases.

Crypto x AI: A Key Unlock for the “Agentic Web”

The idea of AI agents operating on crypto infrastructure stands out as particularly exciting. This integration aims to create the “Agentic Web,” a transformative paradigm that could enhance security, efficiency, and collaboration in AI-driven economies, all underpinned by robust incentive structures and cryptographic primitives.

We believe AI agents can become significant drivers of economic activity and growth, eventually becoming the predominant “users” of both on-chain and off-chain applications. In the medium to long term, this shift may prompt internet-native firms to rethink their core assumptions and develop products, services, and business models that cater to a largely agent-based economy. While crypto isn't essential for advancing capabilities or solving challenges at every layer of the AI tech stack, it can play a major role in bringing more distribution, verifiability, censorship-resistance, and native payment rails to AI, while benefiting from AI mechanisms to power new on-chain user experiences.

Preliminary Beliefs

Crypto as the Preferred Payment Rail

Crypto, being internet-native and programmable money, offers clear advantages for powering an agent-based economy. As AI agents grow more autonomous and engage in high-volume micro-transactions — paying for inference, data, API access, decentralized compute, or data resources — crypto’s efficiency, borderless nature, and programmability position it as the preferred medium of exchange over traditional fiat.

Agents will also require unique, verifiable identities ("Know Your Agent") to comply with regulatory rules when transacting with enterprises and end-users. Low- fee blockchains, smart contracts, self-custody wallets, and stablecoins can streamline complex financial agreements between agents, while decentralized networks ensure trust and auditability of these transactions.

Generative AI and Natural-Language Interfaces

Generative AI and natural-language interfaces are set to become the primary way for you to transact on-chain. As natural language processing improves and AI’s understanding of crypto deepens, interacting through conversational interfaces will become the norm. You might simply describe a transaction intent in plain language (e.g., “Swap X for Y”), and AI agents will translate that into verifiable smart contract code, offering the most efficient and cost-effective execution.

AI Creating Software Code

AI is rapidly advancing in code generation, already transforming web2 software development. We anticipate this trend will soon dominate the crypto sector, lowering barriers for new and existing builders.

In the near term, the focus will be on easing development tasks. Looking further ahead, AI “software agents” may generate smart contracts and hyper-personalized apps in real time, based on user preferences, all stored and verified on-chain. This could lead to a rapid expansion of on-chain apps and experiences.

These beliefs suggest a future where the lines between AI and crypto blur, creating a paradigm of intelligent, autonomous, and decentralized systems. With that perspective in mind, let’s take a closer look at the enabling Crypto x AI tech stack layer-by-layer.

Opportunities Within the Crypto x AI Stack Today

The drive to merge crypto with AI has created a rapidly evolving and complex landscape, with many builders seizing current market momentum. You can view the Crypto x AI environment as segmented into four key layers:

1. Compute: Networks focused on supplying latent graphics processing units (GPUs) to AI developers.

2. Data: Networks that enable decentralized access, orchestration, and verifiability of the AI data pipeline.

3. Middleware: Platforms and networks that support the development, deployment, and hosting of AI models and agents.

4. Applications: User-facing products, whether B2B or B2C, that leverage on-chain AI mechanisms.

Compute

AI demands enormous GPU resources for training models and executing inferences. As AI models grow more complex, the scarcity of cutting-edge GPUs—like those from Nvidia—leads to long wait times and rising costs. Decentralized compute networks are emerging to tackle these challenges by:

• Creating permissionless marketplaces for buying, renting, and hosting physical GPUs.

• Building GPU aggregators that allow anyone (for example, Bitcoin miners) to offer their excess GPU capacity for on-demand AI tasks in exchange for token incentives.

• Tokenizing physical GPUs into digital assets on-chain to "financialize" them.

• Developing distributed GPU networks tailored for intensive workloads such as training and inference.

• Establishing infrastructure that runs AI models on personal devices, akin to a decentralized version of Apple Intelligence.

Each of these solutions aims to boost GPU supply and accessibility with competitive pricing. However, mainstream adoption may be slow in the near-to-medium term due to varying support for advanced AI workloads, challenges with GPU co-location, and sometimes insufficient developer tools and uptime guarantees compared to centralized services.

Emerging Segments in Compute

Projects building in this space can be grouped as follows:

• General-purpose Compute: Decentralized marketplaces offering GPU resources for diverse applications.

  Examples: Akash, Aethir

• AI / ML Compute: Networks providing GPU resources specifically for AI and machine learning tasks, including aggregation, distributed training, inference, and tokenization.

  Examples: io.net, Gensyn, Prime Intellect, Hyperbolic

• Edge Compute: Compute and storage networks that power on-device large language models for personal, contextualized inference.

Data

Scaling AI models demands vast GPU and data resources. Large language models (LLMs) are trained on trillions of words from human-generated text. However, there's a finite amount of public, high-quality data available—Epoch AI even estimates that such language sources might be exhausted by 2024. This scarcity raises concerns that a lack of training data could bottleneck model performance, potentially causing a plateau.

To address these challenges, data-focused Crypto x AI firms have several opportunities:

• Incentivize Data Sharing: Users can be encouraged to share their private or proprietary data. For example, “Data DAOs” are on-chain entities where data contributors can benefit economically while governing how their data is used and monetized.

• Synthetic Data and Web Scraping: Tools can be developed to generate synthetic data assets from natural language prompts, or to incentivize users to scrape data from public websites.

• Data Preprocessing Incentives: Users could be incentivized to help pre-process datasets for training models, such as through data labeling or reinforcement learning from human feedback.

• Permissionless Data Markets: Establishing multi-sided, permissionless data markets allows anyone to contribute data and get compensated.

Although many emerging players are exploring these opportunities, centralized incumbents with established network effects and data compliance regimes pose challenges for decentralized alternatives. Nonetheless, the data layer for decentralized AI presents a significant long-term opportunity to overcome the “Data Wall.”

Emerging Segments and Projects in the Data Layer

• Data Marketplaces: Decentralized protocols for sharing and trading data assets.

  Examples: Ocean Protocol, Masa, Sahara AI

• User-owned / Private Data (incl. DataDAOs): Networks designed to incentivize collection of proprietary datasets, including user-owned private data.

  Examples: Vana, NVG8*

• Public and Synthetic Data: Platforms for scraping data from public websites or generating new datasets via natural language prompts.

  Examples: Dria, Mizu, Grass, Synesis One

• Data Intelligence Tools: Applications for querying, analyzing, visualizing, and extracting insights from on-chain data.

  Examples: Nansen, Dune*, Arkham, Messari*

• Data Storage: File storage networks for long-term archiving and relational database networks for managing structured, frequently accessed data.

  Examples: Filecoin, Arweave, Ceramic*, Tableland*

• Data Orchestration / Provenance: Platforms that optimize data pipelines for AI applications, ensuring verifiable authenticity and provenance of AI-generated content.

  Examples: Space and Time, The Graph, Story Protocol*

• Data Labeling: Incentive-based networks for human contributors to create high-quality training datasets, improving reinforcement learning and model fine-tuning.

  Examples: Sapien, Kiva AI, Fraction.AI

• Oracles: Networks utilizing AI to provide verifiable off-chain data for on-chain smart contracts.

  Examples: Ora, OpenLayer, Chainlink

Middleware

Realizing the full potential of an open, decentralized AI ecosystem requires you to consider the new infrastructure being built. Some high-potential areas that builders are exploring include:

• Employing open-weight large language models (LLMs) to power on-chain AI use cases while simultaneously building foundational models that quickly understand, process, and act on on-chain data.

• Developing distributed training solutions for large foundational models (e.g., models with 100B+ parameters). While this has been seen as a distant goal due to technical complexities, recent breakthroughs by groups like Nous Research, Bittensor, and Prime Intellect are working to change that narrative.

• Leveraging privacy-enhancing techniques such as zero-knowledge or optimistic machine learning (zkML, opML), trusted execution environments (TEEs), or fully-homomorphic encryption (FHE) to enable private and verifiable inference.

• Enabling open, collaborative AI model development via resource coordination networks or building agentic networks and platforms that use crypto infrastructure to enhance AI agents for on/offchain use cases.

While progress is being made on these fundamental infrastructure primitives, production-ready on-chain LLMs and AI agents are still in their early stages. This dynamic is unlikely to change in the near-to-medium term, as it depends on the maturation of underlying compute, data, and model infrastructure. Nevertheless, this category is very promising and is a core focus for investment strategies—driven by the long-term growth and demand for AI services.

Emerging Segments and Sample Projects in Middleware

• Open-weight LLMs: AI models with publicly accessible weights that anyone can use, modify, or distribute freely.

  Examples: LLama3, Mistral, Stability AI

• Onchain Model Creators: Networks and platforms that enable the creation of foundational LLMs for on-chain use cases.

  Examples: Pond*, Nous, RPS

• Training & Fine-tuning: Networks and platforms that offer incentivized and verifiable on-chain mechanisms for training or fine-tuning models.

  Examples: Gensyn, Prime Intellect, Macrocosmos

• Privacy: Projects employing privacy-preserving mechanisms for AI model development, training, and inference.

  Examples: Bagel Network, Flock, ZAMA

• Inference Networks: Systems that use cryptographic techniques and proofs to verify the correctness of AI model outputs.

  Examples: OpenGradient*, Modulus Labs, Giza, Ritual

• Resource Coordination Networks: Networks designed to facilitate resource sharing, collaboration, and coordination in AI model development.

  Examples: Bittensor, Near*, Allora, Sentient

• Agentic Networks & Platforms: Platforms that support the creation, deployment, and monetization of AI agents for both on-chain and off-chain environments.

  Examples: Morpheus, Olas, Wayfinder, Payman*, Skyfire*

Bottom Line

The convergence of cryptocurrency and artificial intelligence promises to fundamentally reshape the digital economy. By integrating blockchain's decentralized, verifiable infrastructure with AI's capacity for learning and autonomous action, we are on the cusp of realizing the “Agentic Web” — a paradigm where AI agents drive economic activity, transact seamlessly, and generate self-evolving on-chain applications.

This transformative synergy, articulated through the Crypto x AI stack, spans foundational layers from compute, data, and middleware to user-facing applications. Each layer presents unique opportunities: decentralized GPU networks can alleviate resource constraints for AI development; innovative data protocols can democratize and verify vast datasets; advanced middleware can nurture privacy, collaborative model development, and sophisticated on-chain AI services.

Key beliefs underscore that as AI agents become primary economic actors, cryptocurrency will evolve into the preferred payment mechanism, while natural-language interfaces and AI-generated smart contracts simplify and accelerate on-chain interactions. These trends promise to drastically lower barriers to entry, enhance user sovereignty over data, and stimulate unprecedented growth in on-chain experiences and decentralized applications.

While technical and ethical challenges remain, the ongoing development of decentralized AI infrastructure and on-chain AI use cases highlight a promising trajectory. The anticipated benefits — improved verifiability, resistance to censorship, and incentivized collaboration — are poised to unlock a more accessible, transparent, and agent-driven digital ecosystem. This evolving landscape not only spurs innovative applications but also redefines how individuals interact with technology, enabling a future where human and machine collaboration is seamless, autonomous, and economically empowering.