Cryptographic assets may be the key to seizing the opportunities in intelligent agent AI.

Written by: Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton

Translated by: Jiahua, ChainCatcher

The Evolution of AI is Dominating Investment Narratives

AI has been evolving. Around the early 2010s, early capabilities like machine learning, natural language processing, and predictive analytics ignited the "big data" era, enabling people to process structured and unstructured data at speeds and volumes previously unimaginable. At that time, AI was more of a tool assisting human work.

By the early 2020s, generative AI emerged, elevating its uses and roles significantly. AI transformed from a helper into a "co-creator," capable of generating content, responding to various inquiries, and even completing parts of work on behalf of humans. The potential of generative AI has yet to be fully realized; its products are becoming stronger and are permeating more corners of everyday life.

As its influence grows, AI's status as a primary investment theme is indisputable.

On July 14, 2026, IBM's stock price plummeted by 25.2% in a single day. Prior to this, it warned that corporate technology budgets are increasingly directed towards AI infrastructure, while spending on traditional software and IT projects is being postponed or cut.

Today's S&P 500 Index concentration is at its highest level since the dot-com bubble of the late 1990s. The ten largest stocks by market capitalization are all AI concept stocks, collectively accounting for nearly 40% of the total market capitalization of the index. In contrast, during the internet bubble, this figure was only 25%, and in 1980 it was as low as 15%.

Institutional investors particularly view AI as a structural mega trend, heavily investing in AI infrastructure, data centers, and semiconductor stocks.

However, such positioning may not be sufficient to capture the next round of evolution in AI dividends.

The Rise of Intelligent Agent AI

The emergence of generative AI marked a leap in capabilities, effectiveness, and application scenarios compared to earlier AI tools. Today, intelligent agent AI is maturing and being widely adopted, potentially impacting everyday life no less than, or even more than, its predecessor.

Intelligent agent AI advances the interaction model from a passive, responsive chatbot to an autonomous system capable of perceiving its environment, formulating its plans, and executing multi-step tasks to achieve higher-level objectives without continuous human oversight.

Agents can interact directly with external software systems and code repositories, thereby redefining the role of AI. 38% of institutions report that by 2028, Agents will become team members like human colleagues, enhancing productivity and driving innovation together.

Following this trend, the tasks assigned to AI will become increasingly complex. Generative AI excels at gathering knowledge and organizing content; Agents will increasingly take on "transaction" tasks, initiating, tracking, completing assignments, and managing final outcomes.

Predictions suggest that by 2030, the scale of the agent business could reach between $3 trillion and $5 trillion.

For institutions, such transactions will primarily occur within enterprise software systems. Projections indicate that by 2028, 33% of enterprise software will incorporate intelligent agent AI, with up to 15% of everyday decisions being handled by these Agents.

Between software systems, small fees may be exchanged for computational power, API calls, data usage, and various services. This represents a novel interaction model that can record and settle accounts down to the level of each individual task.

Protocols for "Software Paying Software" are Emerging

Protocols supporting such "machine-to-machine" transactions are gradually emerging. Stripe and Visa have already launched the Machine Payment Protocol (MPP).

Open-source solutions are also gaining traction. In the early 1990s, the World Wide Web's designers allowed for a response code numbered "402," labeled "Payment Required," when establishing communication rules between browsers and servers.

Coinbase built upon this to create the "x402" protocol, allowing Agents to initiate and complete such payment instructions, and subsequently transferred the relevant intellectual property to the Linux Foundation, making it an open industry standard.

Today, major credit card networks, along with Web2 giants like Stripe, Shopify, Google, and Amazon Web Services (AWS), as well as an increasing number of Web3 service providers, have adopted this payment standard. The goal is to enable "software to pay software," with no human intervention required throughout the process.

In the coming years, Agent payments are likely to reshape consumer interaction. Projections indicate that by 2030, Agents will account for 15% to 25% of US e-commerce sales.

Currently, ChatGPT processes 2.5 billion inquiries daily, with approximately 53 million shopping-related queries initiated through AI platforms. OpenAI is also integrating the checkout process into third-party ChatGPT applications, such as Target, DoorDash, and Instacart.

How Blockchain Can Support Machine-to-Machine Transactions

To support these "machine-to-machine" transactions, a secure, autonomous, verifiable, and high-throughput accounting system is required.

Traditional credit cards and banking systems are not suitable for the small payments made by Agents due to their fee structures. A standard credit card transaction typically incurs a fee of 2% to 3%, plus about $0.30 in fixed costs; however, an Agent purchasing 1 second of computing power or executing a single data query usually costs only $0.001.

For intelligent agent AI to function effectively, it may rely on cryptographic technologies and blockchain, as these foundational tracks are inherently suitable for such scenarios. In fact, with the following characteristics, blockchain and cryptographic technologies are expected to serve as the foundational carriers for these transactions.

Automatic generation and execution of contracts. Payment Agents will generate tokens to complete purchases and settlements. Each token contains a set of transaction rules: which merchants can accept the token, the maximum amount that can be spent per transaction, and the token's expiration date. Once a purchase is completed, the one-time token will automatically become invalid. Blockchain can hold, send, and receive these tokens and operate strictly according to the rules outlined in the tokens.

Decentralized identity verification. Each Agent has a unique identity that can be cryptographically verified. Each token it generates carries its own credentials, and this credential must be used to sign blockchain transactions. When validating transactions, the blockchain will check these credentials, and if an identity is deemed illegitimate, the consensus mechanism will block the transaction.

Fully auditable. Every decision made by Agents on the blockchain, every transaction, and every data exchange can be recorded on an immutable ledger, publicly retrievable by anyone through a blockchain explorer, ensuring accountability and transparency.

Access to decentralized computing power and data. Through blockchain, Agents can leverage distributed computing resources (like GPU networks) and data, lessening their reliance on centralized, private cloud infrastructure, while also helping to lower the operational costs of high-frequency trading models.

Speed and settlement. Bitcoin can only process about 7 transactions per second, while Ethereum handles about 75, but new high-speed public chains have significantly increased these numbers: Aptos can achieve up to 12,933 transactions per second (TPS), Solana is at 6,284 TPS, and BNB Chain is at 3,252 TPS.

This speed is comparable to the Visa network, which typically processes 1,700 to 10,000 transactions per second during normal operations. However, even such a comparison underestimates the efficiency of on-chain systems. The blockchain records and completes settlements within that TPS time window; whereas Visa only records transactions, with actual settlements taking 1 to 3 business days.

With these characteristics, blockchain will play a crucial role in facilitating consumer-level transactions of intelligent agent AI. Conversely, the growth of intelligent agent AI is likely to become a "killer application" driving blockchain's proliferation.

How to Invest in the Opportunity of Intelligent Agent AI

Currently, to capture the growth dividends from AI, investors typically buy stocks of AI concept companies and related industry chain stocks, or become LPs in private equity funds, or invest in energy companies and data centers that support AI operations.

However, to seize the opportunity in intelligent agent AI, these portfolios may need to extend their exposure to native tokens of public chains, as well as project tokens issued by on-chain applications and projects. Several forces are driving this shift, roughly as follows.

Cryptocurrency demand will rise. To record a transaction on a certain chain, Agents must use the native token of that chain to pay transaction fees. For example, to record on Solana, they need to pay with SOL. As Agent payments increase, the demand for native tokens of public chains supporting these transactions may surge, creating value for every token holder. Initially, this demand will likely come from small payments between enterprise software systems through machine-to-machine interactions.

The blockchain ecosystem will expand. The more transactions occur on a chain, the greater the demand for its native tokens, leading to an increasing flow of funds into that chain's treasury. Blockchain foundations will allocate these treasury funds to developers as grants, encourage them to develop applications on the chain, offer "bug bounties" to developers who discover security vulnerabilities, and incentivize those who validate transactions on the network. The more funds available for allocation, the more likely the ecosystem will grow and become more secure, attracting more developers to build applications, issue their tokens for project financing, and share ownership of applications.

Web3 applications will capture market share from Web2. As more applications migrate to the blockchain and talent continues to flow in, the advantages of Web3 applications over Web2 will become increasingly apparent. This has already happened in Web3 gaming: the gaming industry is shifting from the Web2 "single-player" model to the Web3 "player-owned economy."

Click-to-earn applications have already attracted hundreds of millions of users globally. Players can now trade and sell gaming items (NFTs) in secondary markets across platforms, truly owning and monetizing the assets they accumulate in games. Similar experiences and ownership transformations may unfold in a host of consumer-grade applications, further fueling market interest in the tokens associated with these projects.

The flywheel effect will begin to spin. Protocols embedding Agent payments into blockchain applications are already emerging, potentially bringing a flywheel effect to these newly issued tokens. Users only need to instruct their Agents to handle transactions and payments without having to set up wallets, buy tokens, or manage various tokens and cryptocurrencies themselves.

For users, the experience of using a Web3 application may appear no different from using a Web2 product. However, because the tokens involved here have both utility value and represent ownership, they can obtain more benefits within the Web3 ecosystem.

To some extent, this transition will mirror the shift from Web1 to Web2: from the static websites and webpage servers of Web1 to the cloud-based businesses and interactive applications of Web2. In both transitions, whether in terms of the underlying technology providers or businesses built on those tracks, the main players have shifted from established incumbents to a new set of players driving growth in the new era. This time, the torch will be passed to blockchain and various decentralized applications and projects.

Right now, investors haven't fully grasped how to capture the value generated by blockchain and its ecosystem. They are accustomed to a centralized business world dominated by companies: to share in the value generated by a company, they simply buy its stock.

However, I believe that over the next few years, it will become increasingly clear that to capture the value of decentralized networks and businesses, investors will need to purchase related cryptographic assets. Such assets are likely to become significant holdings in investment portfolios, especially for those wanting to seize the new opportunities presented by intelligent agent AI.