Wang Yuehua: The greatest value of stablecoins lies in redefining the way global capital flows.

On August 27, 2026, the "Where Capital Meets Intelligence – CDDJAP 2026 Summer Accelerator Launch Ceremony" was successfully held in the three function halls of Cyberport, Hong Kong. Concurrently, a high-level forum themed "Where Capital Meets Intelligence" was also successfully held. Wang Yuehua, Partner of Draper Dragon and Partner of CDDJAP Accelerator, accepted an exclusive interview with Jinse Finance. The following is the transcript of the interview. The real difference in the future AI competition between China and the US lies not in model benchmarks, but in the entire AI ecosystem. AI competition will present two different models: the US will continue to promote Intelligence Frontier, while China will continuously promote Intelligence Cost Down. AI × Blockchain presents several noteworthy opportunities: Agent Economics formed by AI Agent + Payment; Identity, Wallet, and Ownership of AI Agents; Decentralized Compute and Data; Verifiable AI and Privacy; and Machine Economy. The greatest value of stablecoins lies not in the crypto market, but in redefining the way global funds flow. In the next 5 to 10 years, those companies with the potential to grow into the level of Google, Amazon, and Anthropic will not necessarily come from today's hottest sectors, but from those that can become the next generation of infrastructure or platforms. Jinse Finance: Draper Dragon has long invested in early-stage technology companies globally. From a global investor's perspective, where are the biggest market opportunities in AI, Web3, and blockchain right now? Wang Yuehua: The biggest opportunities right now aren't in any single AI, Web3, or blockchain sector, but rather where several technology cycles are truly converging. First, let's look at AI. In the past two or three years, everyone has focused on the large models themselves, but I'm more focused on the process of AI moving from "model capabilities" to "industry capabilities." In other words, opportunities will gradually extend from simply training larger models to AI Infrastructure, computing power, chips, data, inference costs, as well as AI Agents, Enterprise AI, Robotics, and Physical AI. Especially when AI enters the real world, its demand for semiconductors, sensors, edge computing, robots, and the entire electronics supply chain will increase significantly. This is why I believe Asia, especially Taiwan, will have a very unique position in the next wave of the AI ​​industry. In the past, the innovation center of AI was mostly in Silicon Valley, but truly turning AI into products, devices, and infrastructure requires a complete semiconductor and hardware supply chain, which is precisely Asia's advantage. The second opportunity lies in Web3 and blockchain. This industry is gradually moving from a more "crypto-native" stage to a true financial infrastructure stage. Among the most important directions for the future, I am still optimistic about Stablecoins, Payments, Tokenization, and RWA. Stablecoins have already proven one of the biggest applications of blockchain: redefining the way global funds flow. In the future, cross-border payments, corporate settlements, and internet-native finance will increasingly be built on new digital financial infrastructure. The same applies to RWA. I don't believe all real-world assets need to be tokenized, but financial assets—such as bonds, money market products, credit, fund shares, and various structured financial products—are naturally suited for issuance, trading, and settlement on the blockchain. The value it brings is not simply putting assets "on-chain," but making financial infrastructure global, real-time, and programmable. Thirdly, I think the most interesting long-term opportunity is the convergence of AI and Blockchain. AI brings intelligence and automation, but AI agents will independently conduct transactions, payments, contracts, or manage assets in the future. This requires a machine-native identity, payment, asset, and trust system, and blockchain is likely to become the infrastructure for this. Therefore, from Draper Dragon's perspective, we are not focused on any short-term hot topic, but rather on when these technologies will truly move from technological innovation into industrial infrastructure.

Golden Finance: The US-China AI competition is intensifying. What are the core gaps in AI development between the two countries? What are their respective core advantages?

Wang Yuehua: I think that when discussing the gap between US and Chinese AI today, we cannot say that the US models are several years ahead of China. In fact, this gap has narrowed very rapidly in the past two years. Since DeepSeek, we have seen that even with limited computing resources, Chinese companies can achieve model capabilities close to the most advanced levels in the US through optimization of algorithms, engineering, and model architecture. Therefore, the real difference in the future US-China AI competition will not be in model benchmarks, but in the entire AI ecosystem. If we break down AI into several levels, I think the US currently has three main advantages. First is its cutting-edge fundamental innovation capabilities. The world's leading foundational models, AI research, and many new model architectures and technological paradigms are still primarily driven by US companies. OpenAI, Anthropic, Google, Meta, and the talent pool, universities, cutting-edge labs, and startup ecosystem surrounding these companies are crucial US strengths. Second is computing power and capital. AI is a very capital-intensive industry. The US has core chip and infrastructure companies like NVIDIA, AMD, and Broadcom, as well as the world's largest cloud computing platform and AI data center system, along with a very mature venture capital and capital market. Therefore, the US can continuously invest significant capital to explore the frontiers of AI. Third is its global software ecosystem and platform capabilities. From cloud computing, developer tools, and enterprise software to global internet platforms, US companies still maintain a very strong global distribution. This means that once new AI technologies emerge, they can quickly enter the global market through existing platforms. However, China also has several unique advantages. The first advantage is engineering capabilities and cost efficiency. Due to limitations in computing power and chip resources, Chinese AI companies prioritize model efficiency, inference costs, and engineering optimization. DeepSeek is a prime example. In many industries, what truly determines the large-scale adoption of AI isn't necessarily whose model has a 5% higher benchmark score, but rather who can reduce costs by 80%. The second advantage is application scenarios and commercialization speed. China boasts a massive consumer market and a very complete industrial environment encompassing e-commerce, payment, logistics, manufacturing, automobiles, and robotics. Therefore, once an AI technology emerges, it can quickly enter real-world commercial scenarios and iterate continuously. The third advantage is the manufacturing and hardware supply chain. In the next stage, AI will not only exist in the cloud. AI will gradually enter automobiles, robots, drones, factories, medical equipment, and various edge devices. As AI moves from Generative AI to Physical AI, China's capabilities in manufacturing, supply chains, robotics, and hardware productization will become increasingly important. So, the US currently excels at creating the "Intelligence" for next-generation AI, while China has a greater advantage in rapidly and cost-effectively transforming Intelligence into products and scale. The US creates new technological paradigms, China promotes large-scale applications, and Taiwan and the Asian supply chain truly transform these technologies into products that can be deployed on a large scale. The biggest investment opportunities in the future may lie precisely at the intersection of these three capabilities. Jinse Finance: China's large-scale models are gradually catching up with the US models, and there is even a trend towards them being neck and neck. If the performance of Chinese AI models is close to that of US models, but the cost is very low, could this change the global AI industry competitive landscape? Wang Yuehua: I think it is possible, and the impact could be very profound. Because if the performance of China's large-scale models is close to that of the US, but the cost of using them is only a fraction of that of US models, then the logic of competition will change. In the future, the competition will not necessarily be about who can create the smartest model, but about who can provide sufficiently good Intelligence to the most people and the most devices at the lowest cost. This will lead to three very important changes. First, AI's intelligence will become increasingly commodity-based. Today, we still consider foundational models a very scarce capability, but if more and more models can reach levels close to frontier models, while inference costs continue to decrease, the price of intelligence itself will drop rapidly. This is somewhat similar to the development of the computer industry in the past. Initially, computing power was very expensive, but later, as chip and cloud computing costs continued to decline, the real value came from the software, internet, and applications built upon it. AI is likely to repeat the same process. Therefore, the greatest value in the future may not necessarily lie in foundational models, but may gradually shift towards applications, agents, data, vertical AI, robotics, and physical AI. Second, low cost will greatly accelerate the global adoption of AI. The most advanced models in the United States may still maintain a leading performance, but most companies do not necessarily need the most powerful models. A model with 90% or 95% of the capabilities of the most advanced models, but at only one-tenth or even less of the cost, may be a better business choice for many companies. Especially in Southeast Asia, the Middle East, Latin America, Africa, and a large number of SME markets, cost-performance will be very important. Therefore, China's global competitiveness in AI may not come from "replacing OpenAI," but from providing affordable AI, enabling more companies and developers worldwide to use AI. Third, it will change the value distribution of the entire AI industry. If models become cheaper, the real competitive barriers may shift from models to: data, distribution, applications, computing infrastructure, and the integration of AI with real industries. This is why we are increasingly focusing on AI agents, enterprise AI, robotics, edge AI, and physical AI. As intelligence becomes cheaper, the most important question becomes: what will you do with this intelligence? Whoever owns customers, data, scenarios, and the industry chain is likely to be the biggest beneficiary in the next stage. Of course, the United States still has very strong advantages, including cutting-edge research, top talent, GPUs, cloud infrastructure, and the global software ecosystem. Therefore, it is more likely that two different competitive models will emerge: the United States will continue to promote intelligence frontiers, while China will continuously promote intelligence cost reduction. From the perspective of global industrial development, both forces will accelerate the popularization of AI. If China can maintain this cost efficiency, it does indeed have the opportunity to redefine the competitive landscape of the global AI industry. Jinse Finance: AI is rapidly entering the application stage. What are your views on the combination of AI and blockchain, and what opportunities do you see? For example, AI agents, decentralized computing, and the intelligent economy? Wang Yuehua: The combination of AI and blockchain should not be for the sake of combination. AI itself does not need blockchain to become smarter. For large-scale model training and inference, traditional centralized infrastructure is likely more efficient in most cases. The truly interesting opportunity lies in the fact that as AI moves from copilot to agent, from "providing answers" to "representing people in performing tasks," blockchain becomes extremely important. Because if an AI agent can work autonomously in the future, it will encounter several very basic questions: Who am I? Whom do I represent? What assets do I own? How much money can I spend? How do I pay another agent? How do I sign contracts? How are the transactions I complete verified? Can I bear responsibility? Traditional internet and banking systems are designed for "people" and "companies," not for billions of software agents. Blockchain, however, inherently provides several capabilities: Identity, Wallet, Asset, Payment, Smart Contract, and Verification. Therefore, there are several noteworthy opportunities in the AI ​​× Blockchain paradigm. First is Agent Economics, formed by AI Agents + Payments. I believe this is currently the clearest and most likely direction to see large-scale applications first. In the future, agents will purchase APIs, data, computing power, software services, and even hire other agents to complete tasks. These transactions may cost only a few cents or fractions of a cent, and occur millions of times a day. Traditional credit card and banking systems are not suitable for this Machine-to-Machine Payment, but Stablecoin and Blockchain are very suitable. We have already seen the emergence of such infrastructure, such as the x402 protocol, which allows software to directly pay for APIs, data, and computing power using stablecoins. If this trend continues, a huge incremental user base for Stablecoin in the future may not be humans, but AI agents. The second is the identity, wallet, and ownership of AI agents. Today's AI agents are basically software within a platform. But in the future, agents will have their own wallets, permissions, and assets, becoming independent digital economic entities. Of course, the most important issue here is not giving AI "unlimited permissions," but rather, through smart contracts and cryptography, strictly defining what it can and cannot do, how it can be responsible, and what transactions require human approval, etc. Therefore, a large agent financial infrastructure market will emerge in the future. The third is decentralized computing and data. The biggest production factors of AI are actually three things: computing power, data, and algorithms. Today, these are highly concentrated in the hands of a few large technology companies. Blockchain provides a new possibility: organizing globally dispersed GPUs, storage, data, and other computing resources into a marketplace. And these three assets—computing power assets, data support, and algorithm assets—can all be tokenized, further revitalizing the AI ​​economy. The fourth is Verifiable AI and Privacy, which I am particularly interested in. As AI increasingly participates in finance, healthcare, enterprise data, and agent-based decision-making, we need to know not only what answers AI provides, but also: What data did it use? Was the upstream data authorized? Was the data tampered with? Did the agent have the necessary permissions? Was the computation truly performed according to regulations? If not, did the agent make a mistake intentionally or involuntarily? How will accountability be determined? Therefore, a robust trust infrastructure for AI is indispensable. Finally, looking further ahead, I believe the most promising area is the so-called Machine Economy, or Agent Economy. Today, the global economy is primarily composed of people and businesses. In the future, there will be a third type of Economic Actor: the AI ​​Agent. Agents can create value, provide services, purchase data, purchase computing power, pay other agents, and manage assets on behalf of individuals or businesses. If billions or more agents are transacting with each other, the account systems, payment systems, asset systems, and settlement systems they require will likely not be entirely built on today's banking infrastructure. Therefore, AI provides machine intelligence, and blockchain provides machine identity, ownership, and money. When AI evolves from "thinking" to "acting and transacting," the true integration of these two technologies will begin. The next decade will see a gradual shift from today's Internet Economy to an Intelligent Economy involving humans, companies, and AI agents. Jinse Finance: Stablecoins are becoming a crucial infrastructure connecting traditional finance and the blockchain world. Where do you see the biggest application opportunities for stablecoins in the future? Payments, cross-border trade, or financial markets? Wang Yuehua: I believe the greatest value of stablecoins lies not in the crypto market, but in redefining the way global funds flow. The most obvious opportunity right now is in payments and cross-border settlements. Cross-border payments still face many problems: slow speed, high costs, numerous intermediaries, and low settlement efficiency between different national banking systems. Stablecoins essentially transform the US dollar into a 24/7, real-time, globally liquid, and programmable digital asset. Therefore, whether it's personal remittances, cross-border e-commerce, corporate payments, or cross-border trade settlements between emerging markets, stablecoins have a significant efficiency advantage. Especially in countries where access to US dollars is inconvenient and banking systems are underdeveloped, Stablecoin is no longer just a crypto product, but rather a kind of Digital Dollar Account. Secondly, there's B2B Payment and Global Trade. The truly massive flow of funds in global trade isn't in consumer payments, but between businesses. Supplier payments, cross-border procurement, international trade settlements, and treasury management—these markets are far larger than typical consumer payments. Today, it might take two or three days for a company to pay an Asian supplier from the US, going through multiple banks and incurring exchange rate fees, transaction costs, and reconciliation costs. If businesses could directly complete real-time settlements using Stablecoin in the future, it would have a significant impact on global trade finance. Therefore, the next phase of truly significant growth for Stablecoin is likely to come from Enterprise Payments, not just Consumer Payments. Finally, there's Financial Market Settlement. Today, traditional financial markets are enormous, but their underlying infrastructure remains quite traditional. Securities transactions might be completed in seconds, but actual clearing and settlement often take one or two days. If financial assets such as bonds, funds, money market products, and private credits are gradually tokenized in the future, then stablecoins will naturally become the cash leg for these assets. In other words, the future on-chain financial market needs two things: tokenized assets + tokenized money. Stablecoin is tokenized money. So from this perspective, stablecoin and RWA are actually two sides of the same trend. Without stablecoin, RWA will find it difficult to form a truly efficient on-chain financial market; and as more and more financial assets enter the blockchain, the scale of stablecoin usage will further expand. Another longer-term but very interesting direction is the AI ​​Agent Economy mentioned earlier. If future AI agents can purchase APIs, computing power, data, or other agent services themselves, they will definitely need a kind of money that machines can directly use. Credit cards and bank accounts are not designed for machines, but stablecoin is. Therefore, the users of stablecoin in the future may not only be billions of people, but may also include billions or even more AI agents. The real importance of Stablecoin is not that it makes crypto easier to trade, but that it transforms money itself into an Internet-native, programmable, global, and real-time infrastructure for the first time. Jinse Finance: In the next 5 to 10 years, which sectors do you think will produce the next wave of tech companies similar to Google, Amazon, and OpenAI? How can investors identify these opportunities in advance? Wang Yuehua: I believe that in the next 5 to 10 years, the companies that truly have the potential to grow into the level of Google, Amazon, and Anthropic will not necessarily come from today's hottest sectors, but from those that can become the next generation of infrastructure or platforms. There are five types of opportunities worth paying attention to. The first is AI Infrastructure and the next-generation Computing Platform. Today, we only see large-scale models and AI applications, but every change in computing paradigms gives rise to new infrastructure giants. The PC era had Intel and Microsoft, the Internet era had Google and Amazon, and the Mobile era has Apple. The AI ​​era will also require new computing, chip, data, memory, networking, inference infrastructure, and new developer platforms. Therefore, I believe one of the biggest companies of the future may not be a simple AI app, but rather the infrastructure for the entire AI economy. The second direction is AI agents and new software platforms. Today's software is basically "human-operated software." In the future, it may become "agent-operated software," or even agents directly completing tasks. If this trend holds true, many business models of SaaS, search, e-commerce, customer services, and even some financial services will be redefined. In the past, Google's core was organizing global information, and Amazon's was organizing global goods and cloud resources. In the future, a company may emerge that truly becomes the global operating system, distribution platform, or transaction platform for AI agents. This is a very large possibility and opportunity. The third direction is robotics and physical AI. In the past decade, AI has mainly existed within the screen; in the next decade, AI will increasingly enter the physical world. Robots, autonomous driving, drones, smart factories, medical devices, and home devices are essentially carriers for AI to enter the real world from the digital world. This is why we also believe that Physical AI may be a very important wave after Generative AI. This field is particularly worthy of attention from Asian investors because it is not just software, but requires deep integration of chips, sensors, motors, batteries, manufacturing, and supply chains. The fourth direction is next-generation financial infrastructure, including Stablecoins, Tokenization, Blockchain Infrastructure, and Machine-to-Machine Financial Networks that will serve AI agents in the future. If a large number of global assets, payments, and financial transactions become increasingly digitized, real-time, and programmable, new financial infrastructure giants will inevitably emerge. Today we may see Stablecoin as a cryptocurrency product, but soon it may be more like today's Visa, SWIFT, or Cloud Infrastructure. Fifth, another significant direction is the convergence of AI with traditional industries such as life sciences, energy, and new materials. The true long-term value of AI is not just about helping us write articles or code, nor is it merely about shortening the drug development cycle, improving energy efficiency, discovering new materials, or improving manufacturing efficiency. Once AI begins to accelerate real-world scientific discovery, the economic value it creates could far exceed that of today's Consumer AI. It's full of unimaginable possibilities. However, the real question for investors isn't knowing these directions, but rather: how to discover them before they become consensus? I usually look at a few things. The first is the Cost Curve. Before many great tech companies emerge, the industry often experiences a rapid decline in the cost of basic capabilities. For the Internet, it was bandwidth becoming cheaper; for the Cloud, it was computing becoming cheaper; for Mobile, it was sensors and chips becoming cheaper. Today, we are seeing the cost of Intelligence/AI rapidly declining. When a capability that was originally very expensive suddenly becomes 10 or 100 times cheaper, new products and companies that were previously impossible often emerge. The second is the Bottleneck. Every technological revolution creates new bottlenecks. The stronger the AI, the greater the demand for GPUs, Energy, Memory, Data, Networking, and Security. So we need to ask: if this trend grows 100 times, what will be insufficient? Many big opportunities lie precisely in these Bottlenecks. The third is to look at technology convergence. For example: AI × Semiconductor, AI × Robotics, AI × Blockchain, AI × Healthcare, AI × Manufacturing. Many innovations don't come from a single technology, but from the combination of two mature technologies. The fourth, and I believe the most important ability of a VC, is the willingness to believe in excellent founders before the market has formed a consensus. Companies that truly become Google, Amazon, or Anthropic often don't seem like the safest investments in their early stages. They are often too early, too unusual, and their business model is sometimes unclear. Therefore, venture capital is not essentially about predicting which industry will grow, but about finding those who see the future earlier than the market and have the ability to bring that future to life. So we invest in the future.