# Blockchain and AI: Forging a New Digital Economy
Blockchain technology is increasingly becoming the foundation for advancements in artificial intelligence (AI), according to research by Coinbase Ventures. Cryptocurrencies decentralize the ownership of digital assets, enhance data accessibility, and foster transparent transaction environments. As cryptocurrencies merge with AI, they are setting the stage for a new digital economy. This synergy is evolving particularly around the concept of the ‘Agentic Web,’ where AI agents leverage cryptocurrency infrastructure to autonomously and transparently conduct economic activities, ensuring user-driven data ownership.
# The Road to the Agentic Web
The Agentic Web allows AI to operate independently on cryptocurrency networks, aligning economic activities with user intentions. For instance, AI agents can draft smart contracts reflecting user preferences or autonomously conduct various transactions. This evolution ensures data ownership and privacy while invigorating the digital economy. Coinbase Ventures believes this fusion of cryptocurrency and AI will enhance data accessibility, guarantee verifiability, strengthen resistance to censorship, and expand user-driven economic activities.
The decentralized and borderless nature of cryptocurrencies is crucial for the autonomous operation of AI agents. Cryptocurrency-based transactions can ensure regulatory compliance and trustworthiness when AI agents interact with humans. Cryptocurrencies provide a foundation for AI to efficiently utilize computing resources and handle on-chain verifiable data. Coinbase Ventures foresees AI agents becoming pivotal in economic activities within this emerging Agentic Web.
# Structure of the Crypto x AI Stack
The convergence of cryptocurrency and AI can be categorized into four main layers: Computing, Data, Middleware, and Applications. Each layer plays a significant role in maximizing the synergy between AI and cryptocurrency, paving the way for innovative services for users.
## Computing Layer
High-performance computing resources like GPUs are essential for AI model training and data processing. As AI models grow more complex, GPU resource shortages have become a concern. Distributed computing networks offering GPU resources have emerged to solve this problem. For example, systems that utilize surplus GPU resources to train AI models or perform distributed inference can offer increased accessibility to GPU resources at lower costs. These networks are evolving to convert physical GPUs into on-chain digital assets or marketplaces for trading GPU resources. Companies in this layer include Akash Network (AKASH, AKT) and Io.net (IO), though challenges such as limitations in AI tasks supported, dispersed GPU locations, and a lack of developer tools remain. Adoption in the mainstream market will take time.
## Data Layer
Extensive data is necessary to enhance AI model performance, yet current data resources are limited. New data sharing models like ‘Data DAOs’ are gaining attention to address this issue. Data DAOs allow users to share their data in exchange for rewards, enhancing data usage transparency and ownership. Additionally, tasks like synthetic data generation, data quality management, and preprocessing of training data can be conducted in a decentralized manner, improving data accessibility and quality. Though still in its early stages compared to traditional centralized data platforms, the data layer presents a significant opportunity to overcome the ‘Data Wall.’ Companies in this layer include Ocean Protocol, Sahara AI, Vana, and Ceramic.
## Middleware Layer
A foundational infrastructure is necessary to build a decentralized AI agent ecosystem. This includes platforms supporting AI model development and usage, networks for deploying open-source models, and systems for decentralized data processing. Open-source large language models (LLMs) are essential for on-chain AI implementation, enabling AI models to understand and process on-chain data. Technologies like Zero-Knowledge Machine Learning (zkML) and Optimistic Machine Learning (opML) provide ways for AI models to process data in a verifiable form while protecting personal information. While commercialized LLMs and AI agents on-chain are still in their infancy, Coinbase Ventures sees this category as a future core area of decentralized AI.
## Applications Layer
Various applications based on cryptocurrencies and AI agents are being developed. Initially focused on transactions and data analysis, these applications are evolving towards more sophisticated functionalities. For example, users can interact with AI via text-based interfaces to perform on-chain transactions or create customized applications in real-time. This application layer shows high potential for use in AI companion apps, natural language interfaces, security tools, and risk management agents. As the integration of cryptocurrency and AI accelerates, these applications are expected to develop into more complex and autonomous forms.
# Innovating Through Blockchain and AI
While blockchain and AI exhibit distinct technological characteristics, their convergence creates innovative synergies. The decentralized nature, transparency in data verification, and cryptocurrency-based decentralized transaction systems are crucial for AI agents to operate autonomously and independently in economic activities. Conversely, AI’s ability to process vast amounts of data and provide tailored experiences adds new value to the cryptocurrency ecosystem.
Coinbase Ventures believes the Crypto x AI Stack will combine the autonomy of AI with the transparency of cryptocurrencies to establish a new form of digital economy. They are exploring various investment opportunities in decentralized data collection, GPU-based computing resource activation, and the creation of agentic networks to support the advancement of the Crypto x AI Stack. The vast possibilities opened up by this new digital economy combining AI and cryptocurrencies anticipate active participation from developers and enterprises aiming to advance relevant technologies.
