Editor's Note: On April 21st, at the highly anticipated RWA-themed forum of the 2026 Hong Kong Web3 Carnival, Xiao Feng, Chairman of Wanxiang Blockchain and Chairman and CEO of HashKey Group, delivered the opening keynote speech and officially released the 2026 "Token Economics White Paper". ( Related reading: HashKey Group Releases Third Web3 Economics White Paper: Reconstructing On-Chain Finance and Tokenization Infrastructure for the Era of Intelligent Agent Economy ) In his speech, Xiao Feng deeply analyzed the business model that integrates AI Tokens, blockchain Tokens, and privacy computing technologies such as fully homomorphic encryption. The full text of the speech is below:
Good afternoon, everyone!
After being bombarded with information yesterday and this morning, everyone is probably a bit tired. This afternoon is the RWA themed forum, and as the opening, we will officially release the 2026 "Token Economics White Paper".
Looking back, HashKey began publishing this series of white papers in 2023. In the 2024 version, we specifically proposed the "three-token model," namely equity tokens, utility tokens, and non-fungible tokens such as NFTs. The HashKey Group itself is actively implementing this model: we have our own utility tokens, we sell NFTs during specific events, and the Group also has an equity structure listed on the Hong Kong Stock Exchange.
Over the past decade, we've discovered that while a single token might be sufficient for the most basic public blockchain protocols, a single token layer (such as a utility token) is insufficient to establish a sound, comprehensive, and effective economic incentive mechanism for applications and B2B and B2C users. Utility tokens are primarily used for community incentives, while equity tokens are used to incentivize the founding team and shareholders. This forms the core of our Token Economics White Paper version 1.0.
The white paper has now reached version 3.0. This time, we focus on how the intelligent agent economy, driven by AI agents, can be integrated with cryptocurrencies and blockchain. Therefore, my presentation today is titled: "Innovation in the Intelligent Agent Economy Model"—exploring the disruptive innovations that the combination of AI tokens, blockchain tokens, and privacy computing technologies such as ZK (zero-knowledge proofs) and fully homomorphic encryption will bring to the intelligent agent economy.
From the perspective of business innovation, there are two main characteristics of blockchain technology.
- First, it is an open network that requires no trust and no permission. Participants do not need to undergo KYC or sign contracts beforehand, which is the core technical characteristic of blockchain-native business activities. However, this alone is clearly not enough.
- The second characteristic of blockchain is its openness and transparency. It's hard to imagine what enormous problems banks and other financial institutions with extremely high requirements for privacy protection and compliance would face if they put their entire business process on the blockchain. That would be "data running naked."
Neither financial institution data nor medical data with extremely high privacy requirements can be run directly on a public blockchain without any protection.
On the other hand, AI has indeed unleashed tremendous economic creativity to this day. Especially as AI has evolved from large models to AI agents, the industry is discussing how the intelligent agent economy can unlock more than ten times the commercial value in the future. However, this also faces the issue of data transparency. In the case of complete data transparency, the AI intelligent agent economy has significant flaws, and to address these flaws, privacy-preserving computing technology is essential.
Looking back at the development of blockchain, since the launch of the Bitcoin network in 2009, nearly 16 years of development have fully demonstrated its enormous commercial and economic value. However, due to the open and transparent nature of public blockchains, many compliant businesses cannot operate directly on them. Therefore, in 2015, traditional banks and regulatory authorities in various countries proposed the concept of "consortium blockchains/permissioned blockchains." The emergence of consortium blockchains has alleviated privacy protection issues to some extent—only permitted nodes can join and share data within the permitted scope.
However, this mechanism has significant flaws. Over the past decade, we've seen two major consortium blockchain organizations: R3, a global interbank blockchain, and Hyperledger, led by IBM. But after ten years, they haven't incubated any applications with broad commercial prospects. This led to the view that consortium blockchains might not be blockchains at all. In the context of the time, this did hold some truth.
However, with the rise of tokenization and the tokenization of traditional financial assets, consortium blockchains are making a comeback. As we understand it, almost all major global banks are already running a permissioned blockchain internally. However, these internal bank permissioned blockchains are often single-node and used only for internal verification; we call them "private blockchains."
Why was it able to return?
When a globally renowned bank offers tokenized services (such as deposit tokenization) to its customers, it doesn't need to address trust issues or require third-party node endorsement. Customers already trust the bank; they simply use tokenization technology within the bank's account system to complete a cross-border remittance from New York to Hong Kong in just two minutes. Without tokenization, this remittance might have taken two days. Therefore, private blockchains were the first to experience a resurgence.
However, private blockchains also face limitations. When customers of two different banks conduct cross-bank and cross-border remittances, a broader network across a single private blockchain is required, thus bringing consortium blockchains back into the spotlight. For example, SWIFT is collaborating with nine major global banks to explore how to use blockchain and deposit tokenization tools to solve cross-bank and cross-border fund transfers. This marks the resurgence of consortium blockchain technology. But in cross-bank collaboration, a core challenge remains: how much internal data are you willing to share with your partners?
At this juncture, a new technology has achieved a breakthrough: privacy-preserving computation—including zero-knowledge proofs and fully homomorphic encryption. These technologies can perform computations while protecting privacy, and the results of ciphertext computations are completely consistent with those of plaintext computations. In fact, these technologies have existed for some time; I recall that at the Ethereum DevCon held in Shanghai in 2016, speakers mentioned technologies such as zero-knowledge proofs and formal verification. However, to this day, they have not been widely adopted, primarily due to insufficient performance; their efficiency and cost cannot support commercial implementation.
However, as far as I know, fully homomorphic encryption chips are expected to be launched in the second half of this year, with a performance of around 1,000 transactions per second. This is clearly sufficient to meet the needs of some commercial scenarios, as many scenarios do not require real-time computation, and waiting 10 minutes or even an hour is acceptable. With the support of fully homomorphic encryption, the integration of "blockchain token + AI token + privacy computing" is essential for achieving truly disruptive innovation in the business model of the intelligent agent economy. None of these elements can be dispensed with.
We can envision a resurgence of blockchain technology when privacy-preserving computation algorithms become efficient and cost-effective enough to support large-scale commercial use—at which point we may no longer need private or consortium blockchains. All data can be encrypted and directly uploaded to a public blockchain, with robust privacy protection technologies sufficient to meet the highest levels of global compliance requirements. This is highly likely to happen within the next three to five years. Yesterday, Ethereum founder Vitalik Buterin also shared Ethereum's roadmap for the next five years. He mentioned that Ethereum doesn't need to compete to be the "fastest chain," but only needs to adhere to decentralization and security. In the "blockchain trilemma," Ethereum focuses on the first two, leaving performance issues to hardware acceleration, algorithm optimization, and L2 and L3 networks to address specific scenarios.
Next, I'll give an example to illustrate how these three technologies can work together to create new business models.
Taking hospitals as an example, medical data is extremely valuable, but the requirements for privacy protection are also extremely stringent. In the future token economics model, all commercial institutions will become "token factories." With the support of technologies such as fully homomorphic encryption, hospitals can convert medical data into tokens. Anyone can access this data to calculate the desired feature results, but will absolutely not be able to obtain any individual's original private data.
Only the combination of these three technologies can truly disrupt traditional business models and reach the ultimate form of the intelligent agent economy. Even with only "AI Token + Privacy Computing," the business logic still holds, and hospitals can still innovate, but their business scope cannot expand globally. Digital products and services are inherently global in nature. Without the support of blockchain, demanders still need to negotiate with hospitals offline, sign agreements, and make payments through traditional banks. This is clearly not how a "token factory" operates.
What should a true "Token Factory" look like? It should leverage the permissionless and trustless nature of blockchain to transform hospital data into AI tokens, making them available globally. Any party with a need can access the data 24/7, just like using the Bitcoin or Ethereum network, without cumbersome protocol signing or KYC processes. Accessing data consumes tokens, and consuming tokens automatically completes payment to the hospital. This perfect combination of three elements represents the ultimate goal of the intelligent agent economy.
The same applies to individuals. Imagine your years of medical check-up and healthcare data are encrypted and stored on the blockchain. You could directly send a request to insurance companies worldwide: "My encrypted data is here. You can use actuarial models, within a homomorphically encrypted environment, to calculate my data and provide me with the most cost-effective and personalized insurance plan." This model will completely reshape financial services. There will be no more insurance brokers or intermediaries; you will no longer belong exclusively to a single financial institution, but you will simultaneously be a potential customer of all financial institutions. You can seek your "optimal solution" across the entire network in a trustless and permissionless manner.
Here, I'd like to clarify a common misconception: many people view AI Tokens as the monetary unit of the intelligent agent economy . This is incorrect. AI Tokens are "means of production" in the intelligent agent economy, not currency. Nvidia's Jensen Huang proposed a five-layer structure for Token Economics (electricity, computing power, large models, algorithms, and applications), which describes the production process of intelligent agent products or services. However, when you want to purchase or enjoy these services, you must pay with currency, and this currency can only be digital currency. This is because AI agents cannot use traditional human fiat currency; they require a programmable, infinitely subdivisible, and real-time settlement currency.
Imagine an AI agent calling a hospital's data API. It can't follow the traditional bank transfer process—"I'll transfer the money first, and you can provide the service tomorrow after the funds arrive." It must achieve real-time value settlement. Moreover, such machine-to-machine calls might only cost a few cents or even a fraction of a cent each time, and the transaction fees of traditional bank payment systems simply cannot support such small and frequent transactions. Therefore, programmable digital currency is the true "lifeblood" flowing through the intelligent agent economy.
Furthermore, the intelligent agent economy will also give rise to new digital assets. In this system, not only do currencies need to be tokenized, but assets do too. Why tokenize? Because without it, machines cannot understand or use them. Existing currencies like the US dollar, Hong Kong dollar, and renminbi are not programmable; only through tokenization can they be recognized and manipulated by AI agents.
In conclusion...
The AI-powered intelligent agent economy represents a trustless and permissionless business model, which will lead to a dramatic drop in business costs. In traditional society, maintaining trust in business requires an extremely large and expensive system—including accountants, lawyers, courts, police, and even prisons—all designed to prevent and punish business breaches of trust. The entire society shares the operating costs of this massive system. However, in a trustless blockchain business network, these costs will be completely eliminated.
This new business system, characterized by low cost and high efficiency, will inevitably give rise to new asset classes. The crypto world has "native digital assets" like Bitcoin and Ethereum, as well as "twin digital assets" (i.e., tokenized assets) formed by putting real-world assets on-chain. Similarly, the AI field will also see its twin digital assets and native digital assets. AI-native digital assets will undoubtedly emerge in large numbers in the coming years, building entirely new business models and calling for a completely new financial services system.
The financial system we are familiar with today is designed for "people," but in the future, a new generation of financial service systems, and even a completely new capital market system, will surely emerge, specifically tailored for AI, machines, and AI-native digital assets.
That concludes my presentation for today. Thank you!
