Source: The Round Trip

Compiled by: Nancy, PANews

In early June this year, CSOP, together with HSBC and OSL, launched Hong Kong's first tokenized class of a HKD money market ETF, drawing widespread attention. Against the backdrop of Hong Kong's gradually clarifying regulatory framework, this move by CSOP — the largest ETF issuer in Hong Kong — is seen as a significant step for RWA moving from proof-of-concept to compliant real-world implementation, and also provides a representative compliance sample.

Recently, PANews spoke with Wang Yi, Deputy CEO and Chief Investment Officer of CSOP, about the logic behind implementing tokenized products, the compliance framework, on-chain technical pathways, and the development progress of RWA and HKD stablecoins. The conversation further extended to an investment perspective on AI CAPEX, semiconductors, and global technology asset allocation trends.

Wang Yi stressed multiple times during the interview that the key at this stage is not whether the technology can enable tokenization, but whether the ecosystem has a complete compliance closed loop. The money market fund becoming the first implemented product was not a coincidence, but was driven by a combination of regulatory certainty, asset stability, and a mismatch in market demand.

PANews: Why launch the first tokenized product at this particular time? Was it driven by structural changes in the market, signals of regulatory maturity, or a concentrated surge in institutional client demand?

Wang Yi: When we considered this, regulation was certainly the first step. In fact, earlier on, we had already seen that Hong Kong's regulators have made tremendous efforts in areas like Web3, on-chain, tokenization, and stablecoins.

Our decision on when to launch this product was actually based on Hong Kong's ecosystem environment. Because Hong Kong has relatively high local compliance requirements, with licensed exchanges; tokenization itself also requires upgradeable compliance capabilities.

So the launch at this time point is actually the result of multiple factors. We cannot simply consider whether there is market demand in isolation, because if the market has demand but the ecosystem does not support it, that would also be an awkward situation.

The core driver of this launch was seeing the maturation of the entire compliance ecosystem. For example, our partners, including HSBC's participation. HSBC has also been recognized as one of the main issuers of Hong Kong stablecoins (HKD stablecoin), and at the same time, OSL is a locally licensed exchange in Hong Kong.

Therefore, from the overall ecosystem perspective, we put compliant partners first.

PANews: Why choose a HKD money market ETF as the first tokenization pilot, rather than bond ETFs, stock ETFs, or other products? From the perspectives of technical adaptation and asset characteristics, are money market funds naturally more suitable for tokenization?

Wang Yi: As an ETF issuer, we would certainly consider tokenizing individual stocks, individual bonds, and even stock funds or bond funds. In offshore markets, many asset management companies are already doing this, such as the tokenization of US Treasury bond funds, or the tokenization of individual US stocks.

However, money market funds are a relatively more tried-and-tested direction currently. Looking at the underlying logic, in the Web3 ecosystem, during periods when the crypto market was quite hot, yields from DeFi protocols were very substantial.

But the current time point is actually very interesting. Overall DeFi yields have dropped significantly (previously commonly around 7-8%, or even 50-70%, now fallen to 1-2%). Between the worlds of TradFi (Traditional Finance) and DeFi, a mismatch has actually emerged in the safest assets, providing an entry point for on-chain tokenization of traditional assets. The returns on cash have gradually become attractive relative to DeFi yields.

We already had a money market ETF, so making tokenized shares was relatively smooth.

PANews: Does CSOP have clear plans to tokenize other product lines in the future?

Wang Yi: Definitely. Our first step now is the money market fund has been issued. Subsequently, we will continue communicating with clients to expand into more asset classes.

There is also quite interesting demand within the industry. For instance, some Web3 users have asked when a 2x long SK Hynix ETF would be tokenized; these are all within our scope of observation. Additionally, assets such as commodities and gold are also under our consideration. So the money market fund is definitely not the only product; the overall product line will continue to be enriched going forward.

PANews: How does a tokenized fund differ from a traditional fund? Are subscription/redemption and yields different?

Wang Yi: Based on the current compliance environment, HSBC remains our custodian, and fund operations still rely on the custodian's book-entry system.

On-chain recording, under Hong Kong's current rules, is primarily based on or solely referenced by the custodian's physical booking. The on-chain mapping is more of a record — essentially existing as a proof — but the final reconciliation still depends on the custodian's accounting books.

In principle, the subscription and redemption process is not significantly different from the normal operation of a money market fund, requiring the coordinated collaboration of various stakeholders. Overall, it is not very different from the original fund operating model.

PANews: Will it involve Web3 wallets, mnemonic phrases, private keys, and similar elements?

Wang Yi: That's a very good question. From the tokenization side to the distribution side, the entire chain operates in a compliance environment. This is also why a VATP (licensed virtual asset trading platform) is necessary. It does not mean that users can purchase directly just by having a wallet; rather, within a compliance environment, users need to hold our tokens through a compliant wallet.

PANews: Which blockchain was used for this issuance, and what token standard was adopted to implement investor access controls and transfer restrictions?

Wang Yi: We are currently using Ethereum, and the token standard is ERC-20.

Ethereum is relatively standardized, and Hong Kong regulators also have a requirement to use approved public chains. We have also seen attempts in the market using other chains like Solana, but considering integration costs and coordination among various stakeholders, our first batch prioritized Ethereum as the main chain.

PANews: While bringing traditional assets on-chain improves trading and settlement efficiency, it also introduces new risks related to network security, smart contracts, and custody. What specific measures have you taken to mitigate these risks and protect investors' rights?

Wang Yi: Under Hong Kong's current regulatory environment, on-chain trading activities or transfers still rely on the custodian's book-entry system as the benchmark. This mitigates certain risks to some extent, such as instances of token loss or theft. All transactions still need to go through compliant platforms.

Therefore, from tokenization to distribution, and then to the actual market operation level, there is a certain degree of isolation, along with regulatory protection mechanisms.

PANews: Currently, the product is mainly targeted at high-net-worth individuals and institutions. Will it be opened to retail users in the future? What about the secondary market?

Wang Yi: We will definitely proceed according to regulatory requirements. But the secondary market is a direction we hope to see. If the product is merely deposited to earn interest, its use cases are relatively limited. But if it can become a tool that is tradable and flexibly usable, then its application scenarios and utility will expand significantly, making it easier to integrate into the entire Web3 ecosystem.

PANews: Are other financial institutions in Hong Kong accelerating their entry into tokenization/RWA? What are the driving factors and concerns?

Wang Yi: Previously, our main concern was that the entire ecosystem was not yet fully mature, especially since several of Hong Kong's largest financial institutions had not yet participated at the time. Now, with HSBC acting as both the custodian and the stablecoin issuer, market infrastructure and participant confidence have been significantly boosted, enabling us to smoothly advance the product implementation.

Other institutions are also actively exploring related directions. However, the premise for RWA development is the existence of underlying assets that genuinely possess market demand and appeal. The reason we chose a money market fund was mainly because its ecosystem was already relatively mature and the product format is also highly attractive to Web3 users.

For us, the more important significance of taking this first step is to accumulate practical experience and industry implementation capabilities, deeply understand investors' real needs, and lay the groundwork for expanding more RWA products in the future.

PANews: The global RWA market has developed rapidly in the last two years, especially tokenized products for US Treasury bonds and commodities. Hong Kong regulators have also hinted at gradually opening up tokenization access for more underlying assets (including commodities). How do you assess the current development status and future potential of Hong Kong's RWA market? Compared to Western markets, what are the significant differences in the development path of Hong Kong's tokenization market?

Wang Yi: Each jurisdiction sets its regulatory standards based on its own market.

In Hong Kong, stablecoins and RWA are progressing in parallel. Large asset management firms in the West, such as BlackRock, Apollo, and Franklin Templeton, are moving very fast, but the scale is still very small relative to their total AUM.

For the Hong Kong market, compliance has always been the primary prerequisite for development. We can draw on the practical experience of overseas markets, but more importantly, we need to align with the needs of local investors and participating institutions.

For global financial institutions, how to handle differing regulatory requirements and terms across various jurisdictions is a new topic.

PANews: Will subscription and redemption in stablecoins be supported in the future? What are the application scenarios for stablecoins?

Wang Yi: We certainly hope to see such development and have been maintaining communication and exchanges with stablecoin issuers. Once a stablecoin is officially issued, one of the most concerning questions for the market is its practical application scenario.

From the current perspective, using HKD stablecoins to purchase crypto assets, or to invest in and hold RWAs, are very reasonable directions with strong development potential. After all, stablecoins are fundamentally a medium of exchange. On this basis, how to further expand the application ecosystem of stablecoins by integrating them with RWA and a more active secondary market is a direction worth focused exploration for the industry in the next stage.

PANews: Tokenization significantly shortens settlement cycles and may enable instant movement of assets and cash. In the long run, will this efficiency improvement have an impact on macro liquidity management, the money market fund ecosystem, and even broader market volatility?

Wang Yi: From a product design perspective, RWAs inherently have the potential for 24/7 trading, but the key question is: who will consistently provide liquidity? This is essentially a "chicken or egg" problem.

From traditional market experience, liquidity in overnight trading is typically relatively limited, and supporting round-the-clock trading and services also means additional operational and market-making costs. Therefore, merely enabling 24-hour trading does not necessarily mean the market will have sufficient liquidity.

We believe that establishing an effective connection between TradFi and DeFi is a very positive direction. On the one hand, it can open new distribution channels for traditional assets; on the other, it can provide investors with more diversified asset allocation options when DeFi yields are declining.

However, how much impact this model ultimately creates still depends on the willingness and capability of market participants. Taking liquidity provision as an example, crypto-native market-making institutions are usually more familiar with the digital asset market, but may not necessarily have the ability to provide liquidity for RWA products; meanwhile, traditional financial institutions' market-making teams have clear advantages in related businesses but are often constrained by legal entities, regulatory frameworks, and operational mechanisms.

Therefore, in the short term, the key lies in how many institutions are willing to take the lead in providing related services. Only when the secondary market gradually builds sufficient liquidity and trading activity will there be opportunities to attract more large institutions and long-term capital to participate. At this stage, the entire industry is still in the infrastructure-building phase.

PANews: In the first half of 2026, global macro disturbances persist, with fluctuating interest rate expectations, geopolitical tensions, and technological transformation unfolding simultaneously. CSOP's ETF products have also seen strong growth in size. As CIO, which key investment themes are you currently focusing on, and what advice do you have for investors in terms of asset allocation?

Wang Yi: The core theme we are currently focusing on is the technology sector, particularly the AI theme.

Through our exchanges with global investors — whether clients, peers, or field visits — there's a very clear impression. For example, last year I visited South Korea about three or four times. You only realize one thing when you talk to them on the ground: there is a huge gap between reading reports from a distance and actually communicating on the front lines, and that gap is actually quite substantial.

Currently, everyone cares most about the AI CAPEX theme, because it represents the largest fund flow direction globally. Where capital goes is basically where the opportunities are. If an industry can't even attract capital, then its own logic is questionable.

From another perspective, people's judgment of AI essentially boils down to one question: is AI the next epic industrial revolution, a change on the scale of the Industrial Revolution?

Since last year, we have been advising investors to focus on downstream opportunities in the industrial chain and to try to avoid concentrated upstream risk exposure. The core logic is simple: cloud providers must continue to invest in CAPEX, which is almost certain. But the real investment opportunities lie not only on the "spending side," but on the side that "receives the capital expenditure."

For example, in the AI supply chain, NVIDIA sits at the core, and the constraints extend further into memory and computing infrastructure, such as storage manufacturers like SK Hynix and Micron, and even Samsung has recently had opportunities to enter this chain.

So we essentially follow the direction of the largest capital flows in the market to identify what investment opportunities they bring. The key to these opportunities lies in their sustainability — that is, how long they can last. Why can it be sustained? Is capacity constrained? What are the reasons for the inability to expand production? If capacity cannot be expanded, can prices go up? How much price increase can the upstream accept? And ultimately, to what extent can the corresponding market cap grow?

Of course, the risks here are also relatively high. So we emphasize one point: you must find those segments downstream that can generate real earnings and hold an indispensable position in the industrial chain.

In addition, we do discuss geopolitical factors, and clients often ask about these questions. But overall, the world is still somewhat fragmented at times. From an investment research perspective, everyone ultimately returns to a basic fact: the market respects real earnings. Whoever makes money deserves the corresponding valuation. If you have an industry position and technological barriers, the market will eventually price it in that direction.

PANews: How do you view the sustainability of AI tech stocks?

Wang Yi: Last year we already suggested that CAPEX is very important because it is the source variable of the entire industrial chain. But this doesn't mean you have to directly invest in CAPEX itself. The key is not whether you participate, but whether your judgment of its sustainability is correct.

There's really only one thing that needs to be confirmed: whether these cloud providers' CAPEX investment is still ongoing. As for their capability, it is not the core concern. As long as they can invest 300 billion this year, 500 billion next year, and 800 billion the year after, and the overall trajectory shows no significant shrinkage, no strategic pullback, and no change in plans, then the logic holds. Much of the market news fluctuations around CAPEX — such as changes in financing scale, rising costs, and short-term funding pressures — often go from "financing difficulties, rising risks" one day to "new capital, agreements signed" a couple of days later. Essentially, all of this is short-term noise with limited impact on the underlying trend.

What truly determines the sustainability of CAPEX are essentially two factors: capability and willingness. Capability refers to whether the entities driving this CAPEX have the ability to continuously finance and expand their balance sheets. In reality, the leaders of this investment cycle are the world's strongest technology companies, which are top-tier globally in terms of cash flow, profitability, bond market financing ability, and refinancing capacity. Even if they continue to issue bonds, the scale is only a marginal increment relative to the entire U.S. IG market or Treasury system, so there is no hard constraint of so-called "unsustainable financing." If even these companies cannot sustain CAPEX, then there is virtually no entity in the entire system that is more capable.

But the real key is not capability, but willingness. For tech companies, CAPEX decisions are no longer purely financial issues; they have become existential questions in the face of technological evolution. If AI is defined as the next wave of technological evolution, then history is littered with examples of countless giants vanishing overnight. Today's trillion-dollar, globally leading companies could also be rapidly replaced in the next technological leap and fade into history. For tech companies, this is a matter of life and death: if you don't step up, you die, you fall behind. In the tech industry, there is no option to "gradually fall behind" — it's either lead or get eliminated.

So I would score willingness at 90 and capability at 80 to 90. They are truly very eager to get this done. Then the next thing we watch is how exactly they will execute it.

This brings us back to a more specific question: what is the biggest cost of an IDC (data center)? It's essentially chips, and within chips, the highest cost component is memory. At the same time, IDCs rely on a complete infrastructure system, requiring not only high-speed interconnected fiber optic networks but also highly dependent on stable and sufficient power supply.

The question is, does the United States truly possess the conditions for such long-term, concentrated expansion of IDCs? Electricity shortages are not the only contradiction; more critical is its grid structure: the three major grids are independent of each other, and power supply systems across states are not interconnected. Added to this are differences in state energy policies and voter demographics, creating significant institutional friction for power expansion. As a result, there are indeed growing voices questioning and restricting the continued expansion of IDCs within the U.S.

But as globalized companies, there is more than one way to solve this. Data centers do not have to stay entirely in the U.S.; they can be re-allocated globally. In fact, many companies already generate over 50% of their revenue from overseas markets. The Middle East may have been a priority in the past, but if geopolitical or stability concerns arise, they can turn to Europe or other parts of Asia. For example, Malaysia has lower energy costs and a government that actively courts investment, hoping to host more IDC capacity, which essentially provides another viable solution.

This is actually a dynamic process: when you hit an obstacle, you look for a solution and see if the cost is high. Originally, it was best to build everything in the U.S., but if that doesn't work, I'll go elsewhere.

And all these variables ultimately point to one core issue: whether CAPEX can be sustained. Take NVIDIA as an example. The core question about its sustainability is: during technological iteration, is the GPU still the only solution? Obviously it isn't now, but it remains the most central and largest-scale choice in the entire computing power system.

Therefore, judging its sustainability is not about "whether there are alternatives," but about "whether the cost of substitution is low enough." In other words, after I have already built a large number of IDCs, can I avoid using NVIDIA for computing power procurement? If not, or if alternatives are still significantly disadvantaged in performance, ecosystem, and software layers, then its structural position remains intact.

As long as the economic logic holds, the entire business logic holds. People only see the continuously rising prices, but more importantly, whether these prices bring real efficiency gains. Without efficiency gains, it would certainly be unsustainable; but now the demand for tokens is so huge that existing CAPEX simply cannot meet it — this is a classic supply-demand contradiction.

Demand is growing exponentially, while supply growth is relatively limited and constrained by multiple structural bottlenecks, such as wafer capacity and lithography machine supply. Taking ASML as an example, its annual output of lithography machines is itself a natural ceiling, and this supply rigidity forms the key logic that sets the constraints for the entire industry from the supply side.

This also explains why SK Hynix chose to cut some NAND capacity and concentrate resources on developing HBM. Although NAND itself remains profitable, under the constraints of limited production line and key equipment resources, it is impossible to expand across the board; they can only make structural trade-offs between different tracks.

From a long-term certainty perspective, HBM has stronger sustainability, mainly because IDC demand has distinct long-cycle characteristics (3–5 years), allowing demand and capacity to be locked in through long-term contracts, which differs from the short-cycle, high-volatility order structure of consumer electronics.

So going back to the original question, what exactly is sustainability? When people ask about sustainability, what they really care about is whether it can keep rising. But from our perspective, the core of sustainability lies in whether the underlying logic has changed. If the underlying logic hasn’t changed, then the pullbacks you see — including the correction a few days ago — are really just normal volatility. The underlying logic is what matters most. Because the logic hasn’t changed, the market still needs these things, and prices will rise when they’re supposed to. So I don’t think there’s much of an issue — the underlying logic remains unchanged for now.