Pantera: The Role of Crypto in the AI Revolution

Source: Pantera Capital October Blockchain Letter; Translated by 0xjs@Golden Finance

Crypto: The Tool for AI Gold Rush

Author: Matt Stephenson, Research Partner at Pantera Capital; Ally Zach, Research Engineer at Pantera Capital

"AI is infinitely rich, while Crypto is absolutely scarce."

Sam Altman's observation in 2021 has since become a catchphrase among enthusiasts of both technologies. At first glance, abundance seems to be more influential than enforced scarcity, suggesting that AI might be a more prudent investment. In fact, Nvidia's market value is greater than that of the entire cryptocurrency.

But Altman's remarks remind one of Adam Smith's "paradox of diamonds and water." Smith pointed out that while water is essential for survival, its abundance makes it nearly worthless.

In contrast, diamonds, although not very useful in practical terms, are valuable due to their scarcity. This paradox indicates that even if AI becomes as essential as water, its market value may still be limited. In comparison, the scarcity of cryptocurrencies is more strategically important and valuable than it initially appears.

Large language models (LLM) have achieved significant accomplishments, including passing the Turing test, and reportedly performing better than humans on standard IQ tests. But this raises a question: if humans cannot distinguish between humans and intelligent AI (in the Turing test), can they distinguish between intelligent AIs? If humans cannot tell apart, then the future improvement in AI performance may yield diminishing returns in terms of consumer-perceptible benefits.

Just as the leap in TV resolution from 4K to 8K shows barely noticeable improvements for the average viewer, the differences between high-performance AI models and slightly more advanced models may also be difficult for most users to perceive. This could lead to the commoditization of the majority of the AI market, where the most advanced models are reserved for specialized applications in research, industry, or government, while more cost-effective "good enough" models become the standard for everyday use. Top AI models may become "expensive niche products that mainstream consumers will never consider upgrading to."

Therefore, even if we speculate on the potential growth of AI, we should consider another option: the powerful capabilities of AI that are currently known already exist and will become increasingly commoditized. This is where the intersection of crypto and artificial intelligence ("Crypto x AI") truly comes into focus. The potential of crypto may not be a high-beta bet on the meme value of AI, but rather a mechanism for acquiring the practical value of a distributed future for AI. Once everyone has a 4k TV in their home, its value lies in what we do with them.

By acting as an important and reliable input for AI and the track for distributed AI coordination and trading, cryptocurrencies are closer to the conservative "shovel and pick" bet on AI. This may surprise investors who primarily view Crypto x AI as a volatile proxy for AI's potential growth. Interestingly, over the past six months, viewing NVIDIA as a proxy for AI growth sentiment, cryptocurrencies appear more like a hedge against AI growth sentiment rather than a high beta investment.

We will first evaluate the bright prospects of "AI agents" and how cryptocurrency technology will play a role. Then, we will discuss the potential of cryptocurrency technology in supporting the current inputs of AI: data, computation, and models.

AI Agents: Programs Using Programmable Currency

Author: Matt Stephenson, Research Partner at Pantera Capital

Last year, before most people were talking about AI agents on the blockchain, I collaborated with others to write a paper that was accepted by the top AI conference in the U.S., NeurIPS. Since then, I have had the privilege of attending crypto and AI agent events at universities such as Stanford, Columbia, Cornell, and Berkeley, where I delivered talks, in addition to participating in numerous tech and investment conferences. Next week, I will be speaking about AI alongside a professor from Oxford, the president of IEEE, and a member of GBBC, all aimed at better understanding, exploring, and communicating what the future of AI agents is and how it intersects with blockchain. Of course, I am also investing in this future, including investments in agent infrastructure like Sentient and other undisclosed positions.

The future has arrived. Although OpenAI states that AI agents will not be ready until 2025, in the cryptocurrency field, we already have AI agents conducting transactions and exploring the blockchain space. An AI agent that has promoted its own token (Note: Truth Terminal) currently has about $300,000, and by the time you read this article, it may become the first AI agent millionaire.

But what are these agents? How are they different from the "robots" we are more familiar with?

Agent is not just a robot

Defining "agent" is more nuanced than it appears. The definition of an agent in the field of artificial intelligence is not very practical: "anything that perceives its environment through sensors and acts upon that environment through actuators." Economists have a view of agents that is closer to what we want: "an agent is a person who acts on your behalf in a specific decision-making domain."

If an agent acts on your behalf, then a robot is essentially a difficult-to-communicate agent. First, you must write code for the robot to execute, which means communicating in a (programming) language that most people do not understand. And for those who understand the language, they still have to write programs that specify what the robot should do under various different conditions, which means the conditions must be specified in advance. Both of these are communication costs.

For example, suppose you have a friend going abroad, and you ask them to buy a souvenir for you. If your friend were like a robot, they would ask you to write a program detailing exactly what souvenir they should buy for you. But what if your friend were like an agent? Then you could use language to make your request, trusting that your friend would get you what you want. Using language, without needing to specify your preferences for the gifts you might receive abroad, can reduce communication costs. Clearly, this is a better agent.

Understanding the conditions in advance (because you must program them) limits the practicality of robots as agents. Then, simply because robots must be programmed means they are out of reach for those who do not program. We will turn to AI agent modeling as a reduction of these communication costs and the corresponding release of economic value.

Despite the high communication costs of existing robots, it seems that over $2 trillion in cryptocurrency stablecoin transactions each month are conducted by robots. As robots become better agents, perhaps able to trade USDC and USDT based on relative risk like you do, we should expect this number to increase.

AI agents will use encryption technology.

One reason AI agents are beneficial for cryptocurrency is that they help alleviate the notorious user experience issues associated with it. The complexity of blockchain interactions, wallet management, and decentralized finance protocols has long been a barrier to widespread adoption. AI agents can act as intuitive interfaces, converting user intentions into the precise technical operations required on the blockchain. They can guide users through complex transactions, explain risks, and even suggest optimal strategies based on market conditions and user preferences.

Another reason is that agents cannot have bank accounts, but they can transact using wallets. This restriction of the traditional financial system aligns perfectly with the spirit of cryptocurrency. In the crypto world, agents can operate without needing permission from a central authority. They can interact directly with smart contracts and decentralized protocols, holding and managing digital assets on behalf of users. This opens up new possibilities for automated wealth management, round-the-clock trading, and personalized financial services that operate entirely within the crypto ecosystem.

Finally, a mature agent ecosystem means that agents need to transact and coordinate with each other. Modern smart contracts, as programmable and always-on international legal systems, are well-suited for this task. AI agents can leverage cryptographic infrastructure to participate in complex multi-party transactions and agreements. They can negotiate terms and execute transactions within the parameters set by human principals, and even resolve disputes. This creates a new paradigm of autonomous economic activity, where agents can form temporary alliances, pool resources, and collaborate to complete tasks that humans cannot or are unable to manage directly.

We believe these activities will add value to the crypto infrastructure. However, there are also indirect effects that make crypto itself better. For example, due to attention constraints in crypto, the decentralized autonomous organization (DAO) has remained inactive. A DAO actively managed by an AI agent network (where each agent represents the interests of DAO voters) will change the game. These agents can analyze proposals, allocate resources, and execute strategies at speeds and scales beyond human capability, all while adhering to the core principles and goals set by their human creators.

AI agents and cryptocurrencies are not just a perfect combination; they are two technologies that need each other. Agents need programmable money to operate autonomously in the digital economy. Cryptocurrencies need AI to enhance user experience and fulfill their promise of bringing a financial revolution to everyone. As this synergy develops, we may see core blockchain infrastructures like Solana, Ethereum, Near, and Arbitrum become major beneficiaries of this new agent-driven economy. They are poised to achieve this by facilitating agent transactions, hosting decentralized applications for interactions with them, and providing the secure and transparent environment necessary for coordination among agents. With the increase in agent activities, these networks may see a rise in transaction volume, increased demand for their native tokens, and strengthened network effects. This is not just about technological compatibility; it is about creating a new economic paradigm where AI and cryptocurrencies work together to make finance more efficient and accessible, and it might even feel a bit sci-fi.

Cryptography Technology Supports Current AI

Author: Ally Zach, Research Engineer at Pantera Capital

Imagine that you are on the verge of a major breakthrough, only to find that the tools you need are out of reach. Innovation often feels like this—a journey filled with peaks of breakthroughs and valleys of challenges. Take the automotive industry as an example; the quest for more efficient engines once hit a dead end. Engineers were eager to push the limits, but the necessary materials did not yet exist. Progress stalled until new alloys and composite materials reignited the engine of innovation. Similarly, new technologies such as blockchain may unleash potential that AI has yet to develop.

Over the years, the development of AI has been gradual, first with slow progress and then rapid advancement, resembling an S-curve. In 2017, we achieved a key breakthrough with the emergence of Transformer-based architectures, as outlined in the influential paper "Attention is All You Need." These Transformers revolutionized the processing of sequential data in models, enabling efficient training on large datasets. This sparked the rapid development of powerful new LLMs and generative AI models.

Despite the progress made in AI development, significant bottlenecks in data, computing, and model generation must be overcome to achieve the next leap. Combining AI with blockchain technology can help decentralize resources and democratize access, making innovation accessible to global contributors.

data

Data is the lifeblood of AI and is the fuel that drives its accuracy and reliability. High-quality, representative data is essential for building effective models, but obtaining this data is challenging due to privacy issues, restricted access, and inherent biases. Additionally, users are increasingly reluctant to share personal information, which makes data collection resource-intensive and often hindered by trust issues.

Blockchain technology offers a promising solution by introducing a decentralized, secure, and transparent data aggregation method. Platforms like Sahara align with our long-term strategy to advance AI decentralized infrastructure, enabling individuals to contribute data and monetize it while retaining control. Moreover, token economics incentivizes high-quality contributions by appropriately rewarding users. This approach helps address privacy concerns by allowing users to own and control their data. It democratizes data access, empowering small businesses that previously lacked resources to compete with large tech companies. By securely incentivizing data sharing, blockchain-based platforms commodify data, enriching the available data pool and potentially leading to more robust and equitable AI models.

However, despite its innovative nature, blockchain-based data aggregation is not an independent solution for AI development. When used in isolation, practical challenges such as scalability, data quality assurance, and integration complexity can limit its effectiveness. With vast datasets and mature infrastructure, large tech companies still have significant advantages that decentralized platforms find hard to compete with.

Therefore, solutions including blockchain-based ones introduce new ways of data collection and collaboration, serving as a complement rather than a substitute for traditional methods. The synergy between decentralized efforts and mature technology leaders can foster partnerships that leverage the strengths of both sides, promoting innovation and inclusivity in AI development.

calculation

The rising cost and scarcity of GPUs have posed significant obstacles for small businesses in AI development. Since the outbreak of the pandemic, GPU prices have continuously risen due to strong demand and supply chain issues, leading large enterprises to increasingly monopolize the use of essential hardware. This limits innovation, as many startups and researchers require assistance to afford the tools for training advanced models. This reduces the diversity of AI research and slows the progress of small institutions.

However, Crypto has the potential to create a fair competitive environment by commodifying computing power. Platforms like Exo and io.net are democratizing access to GPUs through decentralized markets, allowing anyone to access or lend computing resources. Individuals with idle computing power can offer it on the network in exchange for rewards. The commodification of high-performance computing allows a broader range of innovators to participate in AI development, breaking down the barriers that once restricted access to advanced tools.

In the future, as the supply of GPUs increases, the decentralized computing market may directly compete with traditional cloud services. These platforms lower the barriers to access and provide cost-effective alternatives, enabling broader participation in the AI ecosystem. However, ensuring users have access to reliable computing power remains a challenge. Validating GPU standards and maintaining consistent, secure resources are crucial for building trust and preventing fraud. While decentralized solutions may not replace traditional services, they can offer competitive alternatives, as flexibility and cost are more important than guaranteed performance.

model

Today, AI development is often concentrated within a small number of organizations such as OpenAI, Google, and Facebook. This concentration limits opportunities for global innovators and raises concerns about whether AI can reflect diverse human values. Centralized control may lead to models that embody narrow perspectives, overlooking the needs and viewpoints of a broader user base.

A transformation is taking place through the decentralization of AI development power on platforms. Platforms like Sentient and Near align with our vision that AI will increasingly operate on a crypto track, democratizing development by creating open-source, community-driven ecosystems. Utilizing blockchain technology, they transparently manage contributions, ensuring that developers are recognized and compensated through token rewards. This enables anyone to build, collaborate, own, and monetize AI products, ushering in a new era of AI entrepreneurship. Illia Polosukhin, co-author of the groundbreaking paper "Attention is All You Need" and co-founder of Near, is working to foster an open environment for developing general artificial intelligence (AGI) through crowdsourcing efforts. Initiatives like this aim to integrate AI development with broad human values.

These platforms act as catalysts for change, driving a competitive yet collaborative AI economy. By expanding participation, they encourage a flourishing of diverse ideas, leading to more innovative solutions and the potential reduction of bias in AI models.

Crypto x AI provides a unique opportunity for the democratization of AI development, but it also brings significant challenges. Balancing large-scale collaboration with the need for high-quality, expert-driven work is crucial to ensuring that models are robust and ethical. By decentralizing data access, computing power, and model development, crypto breaks down traditional barriers, enabling talent from around the world to participate in AI's development. This influx of diverse perspectives fosters collaboration and builds a more inclusive ecosystem. Embracing this collaborative model can not only accelerate innovation but also ensure that the global community shapes the future of AI.