Author: Karthik Senthil
Compiled by TechFlow
Foreword
As the narrative of smart contracts gradually loses its appeal, the powerful entry of AI has brought a breakthrough, driving an unexpected surge in heat and innovation. From the consensus layer of Bitcoin to the execution layer of smart contracts, and now to the AI-driven application layer, is the crypto industry witnessing the third technological leap?
KOL @karsenthil has published his views on Crypto X AI at different times, believing that we are on the eve of the next "takeoff" in the crypto industry.
The original content is divided into two parts, with a relatively scattered structure.
In the first article, the author expresses the view that AI will drive blockchain to enter the next technological leap. In the second article, he elaborates on the specific implementation paths of AI at the application and infrastructure layers, and the potential opportunities for investors and builders.
TechFlow has integrated and compiled the content, and the following is the complete content.
Crypto X AI Viewpoints (Part 1) - We Are Entering a "Leapfrog" Development
AI is driving blockchain to the next major breakthrough.
Each stage of blockchain development usually follows a similar trajectory:
A "disruptive" technological breakthrough triggers a new wave of innovation;
As more imitators join, technological progress gradually stagnates;
Then the next technological leap appears, pushing the industry forward again.
Original image from @karsenthil, compiled by TechFlow
The first major breakthrough in cryptocurrencies occurred at the consensus layer, with the invention of Bitcoin and Proof-of-Work (PoW) marking the beginning of this stage. From 2009 to 2014, this wave led to a more than 10,000-fold increase in the market capitalization of cryptocurrencies (from about $750,000 to about $7.5 billion).
The second leap occurred at the execution layer, with the emergence of smart contracts giving blockchain programmability. Today, the vast majority of blockchain infrastructure (such as L1, L2) and applications (such as tokens, stablecoins, DeFi) rely on this core innovation. From 2014 to the present, this wave has driven a roughly 500-fold increase in the market capitalization of cryptocurrencies, reaching about $3.5 trillion, with projects born in this stage accounting for about 43% (about $1.5 trillion) of the total market capitalization.
However, the current technological progress has stagnated again. Why is this the case? Here are my (possibly controversial) views:
The potential of smart contracts has been largely exhausted. Even the recent popular memecoins are just a recombination of existing technologies (such as tokens, bonding curves, NFT community hype), rather than a completely new invention.
Smart contracts have become the main bottleneck for user experience (UX). Current crypto applications require direct interaction with smart contracts, meaning users must understand the contract's location, functional meaning, and how to interact with it, as well as sign transactions and pay Gas fees.
Fortunately, the next technological leap has arrived - it brings entirely new innovations at the application layer by enhancing usability.
AI Will Become the User Experience Layer of Crypto Technology
The widespread adoption of any new technology requires a powerful "front-end" to simplify complexity and integrate functionality. Personal computers have graphical user interfaces (GUIs) and operating systems, the internet has web browsers and FAANG, and mobile devices have native apps and app stores.
AI will become the user experience layer for blockchain technology, providing users with orders of magnitude better experiences, thereby driving wider adoption. AI can solve the three major user experience challenges in crypto technology: user guidance, execution of complex operations (which large language models (LLMs) are particularly adept at handling), and function discovery. I predict that by 2030, 40% of the global population will have experienced on-chain transactions, and over 95% of those on-chain transactions will be completed through AI. At that time, people will use blockchain-powered applications unconsciously.
Original image from @karsenthil, compiled by TechFlow
To achieve this goal, AI will act as a bridge between the application layer and blockchain infrastructure, playing a role both upwards and downwards in the technology stack. In the future, applications will interact directly with AI agents, and these agents will aggregate and execute on-chain operations on behalf of users. Furthermore, smart contracts will evolve into "intelligent tokens" deeply integrated with AI, providing users with generative and customized experiences, rather than the current one-size-fits-all approach.
From the perspective of AI, the future of blockchain applications becomes clearer. For example, the next-generation financial super-app may use AI to proactively recommend and execute on-chain DeFi operations based on the user's intent and preferences (such as security, yield, etc.), combined with real-time information from prediction markets. Users won't need to understand the differences between L1 and L2, or the names of protocols and assets, or even the workings of cross-chain bridges. The early signs of this trend are already emerging.
Crypto X AI Viewpoints (Part 2) - Opportunities for Builders and Investors
So, who will be the biggest winners?
As AI-driven innovation in the application layer accelerates, the answer is clear: applications will still be the focus (of course, with the support of infrastructure, as this is still the crypto domain). As David mentioned in the excerpt below, we are already seeing a shift from the infrastructure cycle to the application cycle, and the addition of AI will further drive this trend.
@divine_economu: "In 2024, the crypto currency space reached two important milestones:
The first time that the hottest projects were application-led
The hottest protocols were those that innovated in ways to support the development of applications
This is the first time in crypto history that we've shifted from an infrastructure-led cycle to an application-led cycle."
I'm particularly bullish on the following four categories of crypto products, all of which are in the early stages of development and therefore have tremendous growth potential:
Aggregators, also known as SuperApps
I predict that the "FAANG of the crypto space" will emerge: these super-apps will integrate functionalities from AI agents that simplify the on-chain user experience (UX) and directly engage with users. At the same time, these apps will also vertically integrate the technology stack, not only enhancing their own application capabilities but also attracting developers by providing infrastructure (similar to Amazon or Google). Within their respective domains (such as search and advertising, finance, commerce, social, etc.), these applications will exhibit monopolistic characteristics. Just as FAANG companies now contribute about 20% of the S&P index, I expect this category of applications to account for a similar proportion of the crypto market by 2030. A conservative estimate of this market opportunity is in the hundreds of billions of dollars, while an optimistic estimate could reach the trillions.
Especially in the DeFi (or so-called DeFAI) field, I believe this is a killer application scenario: imagine a next-generation one-stop financial super-app, where users can seamlessly access all financial assets on-chain, get investment advice or ideas, analyze market sentiment in real-time, and quickly execute investment decisions. Another exciting direction is a "crypto version of Google", where, by designing algorithms similar to "PageRank", it solves the problem of discovering crypto applications and assets, while generating revenue through advertising or innovative value flows.
The winners in this category will create extraordinary results, as they will have a key advantage that Web2 super-apps lack: Tokens. Tokens are the only tool proven to have strong product-market fit (PMF) in the crypto space, able to attract users, unite believers and investors, and occupy mindshare.
Agents as SaaS
I'm excited about AI agents that can perform exceptionally well in a particular domain. These agents can be used through aggregators or combined with other agents, just like today's SaaS products or financial products. For example, imagine a fully autonomous agent that takes liquidity provider (LP) funds and makes top-tier investments in the crypto markets (being both a top 1% high-liquidity trader and able to participate in the best investment opportunities), while charging lower management fees than ETFs or funds. Or an agent that can achieve high returns in prediction markets or sports betting. Or tools like @aix_bt that can provide high-quality market and investment research data. These agents will allow users to access previously hard-to-reach markets (such as on-chain US dollars or real-world assets (RWAs)) and provide advanced investment strategies (such as quantitative trading or venture capital).
@Loopifyyy: "The first AI agent that can actually perform on-chain transactions and truly work for me, I would unhesitatingly invest my entire net worth in it. It solves the user experience (UX) problem, and now I can use the blockchain with just a simple prompt, regardless of whether it's cross-chain or not."
This is not limited to the financial domain. I can imagine a future where we have an AI doctor that is specifically trained on a patient's personal profile, can charge fees through crypto payment channels to insurance companies, and prescribe low-risk medications. Or an AI insurance agent that can find the cheapest home insurance for your house. Frankly, we still have a long way to go to realize these scenarios (most agents can't even perform basic on-chain interactions yet).
However, as these agents continue to innovate in customer acquisition, value realization, and pricing mechanisms through their native tokens (e.g., users need to hold 100 AIXBT to access premium services), the opportunities in this field are virtually limitless. As this trend deepens, I believe we will also see tremendous growth opportunities for platforms dedicated to trading and managing AI agents (similar to Ebay or OpenSea for agent marketplaces).
AI-Native Infra
The most important infrastructure opportunities of the future (e.g., next-gen L1s) will no longer focus solely on optimizing for speed or cost, but on attracting users by significantly enhancing the user experience (UX). This enhancement will be achieved by building the core architecture around AI agents and AI-driven smart contracts, and natively supporting the following functionalities: efficient on-chain reasoning capabilities (see Section 4), verifiable off-chain reasoning capabilities through Trusted Execution Environments (TEEs), smart accounts that support semi-autonomous AI agent operations (with built-in safeguards to represent users), access to compute resources and model training capabilities, and functionalities that enable bi-directional value flows between agents, driving innovation in agent collaboration and economic models.
Similar to the current era of decentralized applications (dApps), many of the agents mentioned in the previous Section 2 (especially the long-tail agents) will choose to deploy on these new L1s, rather than managing their own infrastructure, while enjoying the network effects of proximity and composability. I'm also excited about the potential of these next-gen L1s, as they may redefine value capture mechanisms, Maximum Extractable Value (MEV), and consensus mechanisms (e.g., can agents become validators?).
This does not mean I'm pessimistic about Ethereum, Solana, or other mainstream L1/L2 ecosystems. In fact, these ecosystems will also gradually introduce similar functionalities in the coming years. But I believe that the new L1s born in this era will be more aligned with the needs of contemporary developers, and thus have tremendous growth potential. Projects like ai16z and Virtuals have already shown early signs of this trend, and also indicate the massive opportunities to become a winner in this space.
L1 innovation will continue and remain strong.
Intelligent Assets
Currently, some of the popular applications in the crypto space, such as stablecoins, Non-Fungible Tokens (NFTs), and ERC-20/SPL governance tokens, are deterministic and static assets. They perform well in accomplishing their predetermined goals, but what if users could own dynamic, self-optimizing intelligent assets that aim to achieve specific objectives (such as increasing holders or enhancing value)?
Imagine smart contracts that can dynamically call models during on-chain execution, allowing assets to perform the following: adjust token supply, release schedules, burning or staking mechanisms, or even modify other parameters that currently require hard-coding or rely on social consensus to change. Each token could even be customized based on the holder's preferences, providing users with a whole new level of personalized experience.
I expect the early exploration of these intelligent assets will be focused on the NFT and Decentralized Autonomous Organization (DAO) domains. For example, NFTs could be fully generative in all aspects, not just the media content. Or a governance token could automatically draft proposals or represent users in voting based on the protocol's history and user preferences.
As the technology matures, the primary application scenarios for this category may shift towards the financial domain. For instance, imagine that Ethena's USDE stablecoin could dynamically adjust its synthetic US dollar strategy based on macroeconomic conditions. This would be an exciting future!
