VC Perspective: Hyperliquid Incident Tears CEX and DEX's Secret War of Power

Original title: Exchange War Erupts: Hyperliquid vs. Binance & OKX - The Chopping Block

Original source:Unchained

Original translation: TechFlow

Guest:

• Haseeb Qureshi, Dragonfly Managing Partner
• Robert Leshner, CEO of Superstate & Lianchuang
• Tarun Chitra, Managing Partner, Robot Ventures
• Tom Schmidt, Dragonfly General Partner

Highlights of this issue

Hyperliquid 's JELLYJELLY Crisis How a highly anticipated DeFi project ultimately lost the trust of the market due to the use of false oracles prices to save itself.

Trading platform competition upgrade – Binance and OKX’s move to list JELLYJELLY perpetual contracts is interpreted as a precise blow to Hyperliquid.

Is a decentralized trading platform really decentralized? – The Hyperliquid incident reveals the problems caused by the concentration of validators’ power behind the so-called “decentralization” performance.

DNA on: Blockchain version 23andMe – Say Foundation proposes to protect genetic data through token thresholds; is this an innovation in privacy protection or a dystopian idea?

Decentralized Science (DeSci) scam – Tarun criticizes DeSci's philosophy again and explains why the risk of putting genetic data on chain is more serious than meme coin.

The Stable Coin Regulation Battle – The Stability Act and the Genius Act compete in Washington, who can gain the upper hand?

The possibility of stablecoins as narrow banks – The rise of cryptocurrencies may force the Fed to accept a financial philosophy they have resisted for 20 years.

The future bet of HLP deposits – The host bets with real money, will the amount of deposits rebound or continue to decline after the Hyperliquid collapse?

New trends in Memecoin and Olympus – Are those former "robbers" quietly making profits from broken vaults?

Tarun’s Failure Ranking – Why JELLYJELLY’s Failure is worse than MobileCoin, but at least it fits Hyperliquid’s brand positioning.

Hyperliquid event expands

Haseeb:

One of the biggest stories of the week is the drama that took place on Hyperliquid. For those who are unfamiliar, Hyperliquid is a new hot DEX and is now the first DEX by overall trading volume. They have made large-scale airdrops, which are loved by crypto retail investors for their size and fair way of launching.

In recent days, Hyperliquid has encountered a large-scale attack, surrounding a memecoin - JellyJelly, a memecoin with very low liquidity and has passed its peak period, but it is listed as a contract trading symbol by Hyperliquid. A trader opened a short position worth $8 million on JellyJelly, which was equivalent to 50% of Jelly's market value at that time, and was a very large short position. Next, the trader raised Jelly's spot price, resulting in his forced closing of his position.

So, why do traders do this? Why force yourself to be closed?

On Hyperliquid, when a position cannot be closed in a normal way, HLP (Hyperliquid's crowdfunding market maker) will take over the position and try to close the position in an orderly manner. The existence of HLP is crucial to Hyperliquid's liquidity, and it always provides liquidity to traders. However, due to the huge scale of the position this time, HLP was forced to short Jelly, but no one in the market was willing to take over this short position, which eventually led to the so-called "short squeeze".

This short squeeze is not just a prank by retail investors. In fact, the two major trading platforms, OKX and Binance, are also indirectly involved. When the market realized that Hyperliquid or its HLP was shorting Jelly on a large scale, OKX and Binance announced that they would launch JellyJelly contract trading within 24 hours.

Almost everyone thinks, "This is the war between trading platforms." The bosses of CZ and OKX have pointed their guns at Hyperliquid, which is a chance to hit them.

Hyperliquid's validators voted to remove JellyJelly and forcefully close the position at a level below the market price by artificially adjusting the price of oracle.

Haseeb:

Hyperliquid's decision has sparked widespread controversy. They believe that rather than letting the platform or HLP holders bear the losses, it is better to lock the price at a false low price through artificial intervention to ensure that the interests of HLP users are not damaged. However, this move caused Hyperliquid's token price to plummet, from about $21 to $15, a one-day drop of nearly 25%.

This incident has raised two core issues:

First, is Hyperliquid 's response in this case reasonable? Are there fundamental flaws in their mechanism design?

Second, does this reveal that Hyperliquid 's decentralization is not as high as it is advertised?

These issues have sparked heated discussions in the industry, with some centralized trading platforms (such as Bitget) publicly criticizing Hyperliquid's unfair approach. The competition among decentralized trading platforms has become increasingly fierce, and the DeFi field seems to be experiencing an important watershed moment.

So, what do you think of this HLP incident?

Tarun:

I think this incident did expose some of the flaws in the protocol design. Just like the Automated Market Maker, the AMM mechanism does not allow order rejection. For example, the initial Unicorn v2 and v3 have no flexibility at this point, and you cannot choose to accept or reject a specific order. This problem also exists in Hyperliquid's liquidity pool.

Hyperliquid's HLP mechanism is different from other platforms such as GMX's GOP and Jupiter's JLP. The operating logic of HLP is that if the user deposits ETH (ETH), the platform will allocate these ETHs to multiple assets for market making. For example, it is possible to make markets with 1% ETH, JellyJelly, 90% for Ethereum, and the rest for Bitcoin. These asset allocations are determined by off-chain algorithms, and users need to trust the asset allocation capabilities of the Hyperliquid team.

Obviously, they made some mistakes in mechanism design, such as not setting position limits, open contract ceilings, etc. If these limitations were present, the problem could have been alleviated without the need for emergency intervention. Hyperliquid has stated that these mechanisms will be repaired, including adding open contract restrictions and concentration restrictions.

This is exactly why I mentioned in describing this question that there are certain limitations in liquidity pools that do not distinguish between order types. Under the current mechanism, HLP cannot selectively process orders, that is, it cannot reject certain orders and only accept certain types of orders. If HLP can distinguish positions that are forced to close by third parties, the market can price based on the actual value of these positions, and HLP does not have to bear unnecessary losses. However, the current design allows HLP to automatically trade with these positions, a pattern similar to how Unicorn pools work. Therefore, they lack sufficient restrictions in strategy design. These strategies are actually run off-chain by the Hyperliquid team, rather than being executed openly and transparently on the chain.

I'm not sure what their code is specifically implemented, because most of the code is not open source. Although I can run a node, I can only get the binary file and cannot view the source code. In addition, many settings of the system are not clear and transparent enough. This incident clearly demonstrates that they have obvious flaws in their strategic limitations. I think this is an issue they admit needs to be fixed first. But from a market perspective, this also shows the value of having a more open strategy, rather than being completely closed as it is now. Because at present, the operating mechanism of HLP is almost transparent to the outside world.

As a depositor of an HLP, you don't actually know whether HLP has clear risk limits, such as whether it will automatically assume the position risk of the entire liquidity pool. You can't be sure whether they will intervene in the market by artificially adjusting oracle prices as they did in this incident. Although some related content is mentioned in the documentation, users cannot verify the actual operation of these mechanisms because the code is not open source. Even if the code is not open source, there is a lack of other verifiable proof to confirm its behavior.

I think the mechanism provided by Hyperliquid is different from the transparency that users expect in other protocols. In other protocols, users can have a clear understanding of the operating logic of the policy, although this transparency may require some compromise in efficiency and flexibility. Hyperliquid's strategy is not public, which does improve the efficiency of capital use, but also weakens the user's sense of trust. This trade-off is not entirely wrong, but obviously they make less ideal choices in some decisions. However, these problems are understandable and can be fixed.

Controversy over the decision to rescue the market

Haseeb:

Is it reasonable to rescue HLP depositors? Obviously, HLP may face huge losses in this incident. Do you think this is a wrong decision?

Robert:

I think this is a mistake, frankly. Trying to solve market problems after the platform's risk parameters are out of control, but it is inappropriate to close positions by human intervention in a way that makes the HLP pool profit. Because in the contract market, one party’s profit usually means the other party’s loss. In this case, the Hyperliquid team and the validators seem to have mistakenly chosen who is the winner and who is the loser.

HLP liquidity providers should be at risk. If the liquidation is successful, they will make a profit; if it fails, they will have to accept the loss. However, this closing operation made HLP profit, which also means that other market participants suffered losses. I think this goes against the principle of fairness in the market. If they have to interfere with the price, at least they should not choose the price that is beneficial to them. What is even more confusing is that they choose to be even lower than the market price at the beginning of the event, obviously in order to make themselves a winner.

Tom:

I agree. This behavior also leads to the subtle relationship between HLP as a product and the entire Hyperliquid platform. HLP is just one of the funds pools, and users can also choose other funds pools and run different off-chain strategies. HLP is positioned as Hyperliquid's "official fund pool", but in theory anyone can create a fund pool. Therefore, most people will not assume that HLP will enjoy any special treatment. However, this incident made people feel that HLP received some kind of preferential treatment.

Some people compare this incident with the "socialization of losses" or "automatic deleveraging" mechanisms of traditional contract trading platforms, but the two are not the same. In traditional mechanisms, when the overall market is below the margin level, the trading platform will freeze the position and spread the losses into the insurance fund. In this incident, HLP's loss status was only compensated through human intervention, and Hyperliquid itself did not face the risk of default. This makes people question why HLP can serve as a priority liquidity provider for trading platforms? If the HLP operation fails, why will it be rescued?

Robert:

And it was actually rescued in a profitable way, which is crazy.

Tarun:

That's true. And even more ironic, the holders of HLP decided the price of this bailout through governance votes, which indirectly brought profits to themselves.

Robert:

Can you explain in detail? How do HLP holders participate in the Hyperliquid verification process?

Tom:

HLP holders can delegate voting rights to a validator, but some validators need to go through KYC, so the mechanism is a bit complicated.

Tarun:

Verifiers control the price of oracle, so the price adjustment of oracle is determined through governance votes. In other words, HLP holders actually participate in the vote indirectly through governance.

Haseeb:

Yes, this has received a lot of criticism. Because the Hyperliquid Foundation has the absolute majority of voting rights for the HYPE token, in this case, the HYPE holders quickly completed the vote through commission. However, the entire process took only two minutes, from the beginning to the end of the vote, the so-called voters had little real say.

Robert:

I'm still a little confused about whether the launch of contracts on other trading platforms will have an impact on Hyperliquid. The contract market and the spot market are relatively independent trading venues. Even if the demand for JellyJelly contracts on Binance increases, it may not directly change the price of Hyperliquid or spot markets, because the spot price is controlled by the index, which also determines the capital rate of Hyperliquid. So, what is the specific mechanism of this impact?

Haseeb:

First, if Binance wants to go online to the spot market, they need to purchase actual spot inventory, which takes longer to complete. But the contract market is launched faster than the spot market, because the contract market does not require actual spot inventory. As long as there is enough demand, users can start trading contracts without immediately adjusting spot prices.

Robert:

Each long has a short position, and each long has a short position.

Haseeb:

That's right. But structurally speaking, it is easier to start the contract market. If you say, "Hey, I want to hit these guys as soon as possible, and time is limited." Then the quickest way is obviously a contract, not a spot.

The purpose of launching the contract market is to allow more people to participate in the short squeeze. If Hyperliquid is experiencing short squeezes, the opening of the contract market will exacerbate this trend.

Tarun:

This mainly depends on the dynamic changes in the capital rate. In this incident, the capital rate soared by 300% in a short period of time, resulting in extremely unstable markets.

OKX and Binance join

Haseeb:

Funding rates soared by hundreds of percentage points. This is a very fierce short squeeze, so the market expects this matter to subside soon. I guess Binance and OKX may remove the JellyJelly contract in a week or two because there is obviously no real demand for this product in the market.

Tarun:

No one will really need JellyJelly.

Haseeb:

However, I find it interesting that the mechanisms of this event can be difficult to understand, especially if you are not very familiar with the contract market, liquidity providers and the operations of HLP. Let's simplify it first. The essence of the matter is that Hyperliquid is trapped in a high-risk position, while Binance and OKX are trying to further weaken Hyperliquid 's position through market operations. More specifically, their goal is to force HLP, not Hyperliquid itself, into an insolvency situation.

This behavior is very radical, right? Some people compare this incident with CZ's previous approach to FTX, but I don't think the two are similar. Because CZ had no reason to believe at the time that selling FTT would directly lead to FTX bankruptcy. If we look back at the Bitcoin hack, Binance and Bitget actually borrowed Ether to help Bybit fill the deficit and maintain its operations. So they behaved in that event completely differently than they are now treating Hyperliquid. I don't have a good theory at the moment to explain why Binance and OKX take such a strategy.

Hyperliquid's behavior implicitly conveys a signal, that is, HLP has some protection mechanism. If HLP loses too much, Hyperliquid will take the initiative to step in to protect it. Judging from the market reaction, Hyperliquid's price is very sensitive to changes in HLP situation, which confuses me. I 'm curious about the connection between the value of HLP and Hyperliquid. Maybe I've missed some key points about the HLP economic model.

Tarun:

In fact, HLP does not have a clear economic model. It is more like a pure liquidity provider and is not closely related to the core mechanism of Hyperliquid. But I find it strange that I prefer to think of the HLP pool as a debt instrument because it operates by raising funds from depositors.

Haseeb:

I think it's more like equity than debt.

Tarun:

No, HYPE is the equity. This is what makes people confused.

Haseeb:

I mean, from a market operational perspective, investors in HLP actually make all the profits. So it's more like equity than debt.

Robert:

That's true to some extent. Market makers use USDC deposited by users to trade in different markets.

Haseeb:

And the user will eventually get all the benefits, so this is not like debt in the traditional sense.

Tom:

I think the main reason for the decline in the price of HYPE token is that this incident has created uncertainty about the future of the trading platform. After all, if a trading platform has privileged liquidity providers and these providers will never bear the loss, then why do others still trade on this platform? This is actually a problem that all trading platforms with internal market makers will face: how big are these privileges?

Tarun:

There is a more pessimistic view that most of the liquidity of HLP actually comes from the Hyperliquid team itself. So they are unwilling to bear this part of the loss.

There is another reason to think of HLP as a debt instrument, that is, it obtains funds from depositors and uses them for market making activities in various markets. In a sense, it is playing a role as a "local lender". Similarly, agreements like Jupiter and GLP are explicitly lending agreements, in which they charge fees. HLP makes profits through expenses and price spreads. If the HLP defaults, such as in this case, the depositor will have priority claim rights.

So I think HLP is more like a debt holder, and HYPE is the real equity tool. Because HYPE is the core asset that can control key mechanisms such as oracles, and this control is the essence of equity.

FTX Moment?

Haseeb:

Hyperliquid is more like a trading platform, while HLP is a tool used by trading platforms for market operations. HLP can be regarded as equity in market operations, while HYPE is equity in the trading platform itself.

Robert:

I actually feel that we should learn from the FTX incident that the trading platform and hedge fund-like entities (such as those teams that conduct proprietary market making trading within the trading platform) should be completely separated. This will prevent conflicts of interest, right?

Tarun:

It is worth mentioning that the mechanism of Hyperliquid is very different from FTX. On Hyperliquid, I can view every transaction between HYPE and HLP at any time and withdraw funds at any time. This transparency makes it easier to be supervised. I agree with this, and I don't think Hyperliquid's approach is wrong in principle.

Haseeb:

If the trading platform provides protection to a market maker and clearly states that the market maker will not lose money, then the trading platform and the market maker are actually bound together. It's like the trading platform team itself is running this market maker. If you don't believe that this team can run the market maker well, then don't invest in the equity of the market maker.

Tom:

But the question is, how does this happen? For example, "We have a fund pool that can be started, and there is another market maker here to invest." On the surface, this seems to be an open choice, but in fact, it is just a unique market maker.

Tarun:

This is indeed a unique situation. If you look at other market makers, like Seafood, they are always losing money. I don't understand why people are willing to give him funds. His past records show that the losses are very serious.

His funding pool is really interesting. But my point is that there is already an issue of reverse selection in these fund pools. It was not until this incident that people really realized that HLP and Hyperliquid were closely linked, and now this connection has become more obvious.

Robert:

I don't think they were separated before this event. The Hyperliquid team is operating a major market maker, which is the core market maker on the platform. Although the economic benefits belong to users, the owner of the trading platform is actually operating this major market maker and also controls the clearing mechanism.

Haseeb:

Yes, the Hyperliquid connection to the liquidation mechanism does put them in some privileged position. But on the other hand, it also forces them to take on high-risk positions that other market makers may not be willing to take.

Robert:

In a case similar to Alameda, whether they want to or not, they have to take on all the bad positions on FTX, including some high-risk assets. This ultimately led to the collapse of the trading platform. Although they are forced to take these risks, this is also a responsibility to some extent.

Haseeb:

Theoretically, the rationality of this arrangement is that Hyperliquid can continue to operate even if the HLP funds are liquidated to zero. This is the ideal design. If all the mechanisms are mixed together, the design of the entire system becomes unreasonable.

Tarun:

Sensoryly, I think it feels better to be defeated by MobileCoin than to be defeated by JellyJelly. MobileCoin is at least trying to be a real project, and JellyJelly is more of a VC joke.

Haseeb:

After this incident, you may think that HLP and Hyperliquid are closely linked. This may cause third-party market makers or liquidity providers to reduce their activity on Hyperliquid because they realize they are not at the same level of competition as HLP.

Tarun:

To be fair, I have observed a lot of market makers’ participation in decreasing. This phenomenon is not new, but now they have a clearer reason to reduce activity.

Haseeb:

On the other hand, you may see more capital flows into HLPs because people are now aware that the agreement may protect their investments.

Tarun:

We can use DeFi AMA as a benchmark.

Haseeb:

Will it rise or fall? At present, it has fallen.

Tarun:

The deposit amount yesterday was $1.85 million, $2.96 million three days ago, and $3 million on March 24. I think it's a good benchmark time. Now it has dropped to $1.85 million. I think there are two possibilities. One is that, as Haseeb said, funds will flow in because they are considered insurance-like products; the other is that the trading platform 's fees may be reduced due to a decline in confidence. I'm not sure which one will dominate.

Robert:

I think the risk has increased. If the incident is serious enough, the platform may step in and close the market, setting a settlement price so that HLP will not lose money. We just saw this in the JellyJelly incident. This is indeed a protection measure, but it also exposes the vulnerability of Hyperliquid's mechanism on small assets. The possibility of such an attack recurs has increased by at least an order of magnitude.

Haseeb:

I totally disagree. But no one will try this again now.

Tom:

Of course, HLP's strategy in trading platforms is obviously also changing to reduce risks. So, this does not mean that these events are completely independent.

Robert:

But this is not an isolated incident either. Two weeks ago, a similar attack occurred in the Bitcoin market, which was a very large asset. The exact attack parameters had already happened two weeks ago.

The Future of 23andMe and SEI

Haseeb:

Let's talk about DeSci, which recently had a big news in the DeSci space, the SEI protocol, a high-performance Layer 1 EVM chain, announced that they are making one of the boldest DeFi investments to date.

The SEI Foundation plans to acquire 23andMe, a genetic company that recently filed for bankruptcy. They promise to protect the genetic privacy of 15 million Americans and ensure that this data is safe for generations to come. They plan to migrate 23andMe's data to the SEI chain, return the data ownership to users through blockchain encryption technology, allowing users to decide how to monetize their data and share the benefits. It’s not just about saving a business, it’s about building a future where users still have control over the most private data.

Tarun:

Does anyone on the SEI team really understand privacy protection technology? I'm very suspicious. If you tell me that there are experts in the field of privacy protection on this team, I will find it more meaningful. But at the moment it seems that this is just a team that wants to spend a lot of blockchain funds, and their approach is even worse than those companies that simply make acquisitions.

Haseeb:

If they can do this correctly, would you support it? Maybe you can become their consultant and provide some professional advice.

Tarun:

If you want to do this, most of the other bidders are computing biology companies, such as AI drug research and development companies. Their goal is to use 23andMe data to train AI models. This incident has caused controversy because many users are worried about their data being abused. For example, I bought 23andMe’s service eight years ago, and their privacy policy promised not to share data with third parties, but now the company is in bankruptcy and this data may be used to develop drugs, and I did not agree to such a purpose. This concern is understandable. So there are two core issues here: one is privacy protection, and the other is the way of data monetization.

People are concerned about the terms of service or privacy, monetization. One of the goals of all those who are trying to acquire a company is pure monetization, like these computing biopharmaceutical discovery companies. Then there are some people like nonprofits trying to bid, and of course the DeFi world participants.

If blockchain can really bring breakthroughs in data monetization, it might be like the ICO boom in 2017, but I doubt it will fail again. If they can truly find a way to protect privacy and monetize data, it is worth looking forward to. But at present, it is not enough to just claim that "let users have their own data" because so far, I have not seen any successful examples. This reminds me of Tom’s previous complaints, people complaining that the studio has not monetized content through blockchain, but that is not the core of the problem at all.

Tom:

That's true. And I'm curious how they completed the acquisition, and as far as I know, 23andMe still has $200 million in debt. Unless they design a very complex financing structure, or attract investors with SEI tokens.

Tarun:

The problem is that most other bidders are large companies, and SEI doesn’t seem to have a high chance of winning. Emotionally, however, many people hope that the company will have a better destination than being acquired by bidders who monetize data. If SEI could come up with a plan that preserves the original terms of service and protects privacy, I think they could try it. However, this also means they need to rely on validator support, because they are actually borrowing the future benefits of validator.

Robert:

From a macro perspective, these data are currently stored in the database of a bankrupt company. Generally speaking, migrating data to a blockchain is not safer than existing methods. In fact, this may increase the risk of data breaches. Of course, this depends on the security measures adopted, such as encryption technology, zero-knowledge proof (ZK), etc. But overall, I don't think this will significantly improve privacy or security.

Haseeb:

Suppose, according to Tarun, the current situation is that a company acquires the database and processes the data at will. This situation is obviously not what we want to see, but there is indeed such a risk in theory. I have always been skeptical about the concept of "data ownership". For example, some people propose to encrypt the data and put it on the blockchain, and authorize others to use the data through the decryption key. But I've never seen this approach really solve the problem.

Tarun:

This is exactly what I worry about. If blockchain practitioners lack understanding of privacy protection technology, trying to deal with such problems will often backfire. For example, they may spend all their funds but leak data due to operational errors. Worse, these data may be used by some countries to develop biological weapons.

I prefer teams that focus on underlying encryption technology to do this, such as teams that study zero-knowledge proof or homomorphic encryption. But even these teams, they may not be good at commercializing technology on a large scale.

Haseeb:

It has been three months since you announced the war against DeSci, how is it going?

Tarun:

To be honest, I have given up. For example, Bio Protocol has almost disappeared now. I think everyone realizes that most of these projects are scams.

My point is that the DeSci craze is essentially a better package of meme coins. It has changed a brand to attract people who are disgusted with traditional meme coins. These people are still speculators who pursue trends. DeSci works more like donating to nonprofits, but lacks mechanisms to verify its public welfare.

Stablecoin legislation: Genius Act and Stable Act

Haseeb:

Now Congress is advancing a new stablecoin bill. Before this, we mentioned the Genius Act proposed by Kirsten Gillibrand, and now there is another bill proposed in the House called Stable Act, proposed by French Hill.

Robert:

These two names sound like a combination of "stable" and "genius", a bit like "stable genius", don't you think? Maybe that's where it's inspired, like quoting Trump's "stable genius" remarks.

Haseeb:

The new bill is called Stable Act. To compare Genius Act and Stable Act more clearly, the main difference between them is that Genius Act is more friendly to industry development. It allows banks and non-bank institutions to issue stablecoins, and state regulators can also participate in regulation, not just federal agencies. It also supports interoperability and allows payment of income in some cases, which overall encourages innovation and growth of stablecoins.

By contrast, Stable Act is more stringent. It stipulates that only banks or approved bank subsidiaries can issue stablecoins and must be subject to direct supervision by the Federal Reserve. In addition, it limits more types of reserve assets, does not allow payment of income, and imposes a two-year ban on algorithmic stablecoins, although existing stablecoins will have transition periods.

Robert, you recently participated in some lobbying activities on the Washington DC legislation. How do you think the bill is accepted?

Robert:

I happened to go to Washington DC on Tuesday and Wednesday to meet about 15 House members. Obviously, the most concerned topic is stablecoin legislation.

Robert:

I think both sides have shown great interest in formulating legislation that is beneficial to the crypto industry, and there is not much controversy. Stablecoin legislation is the most pressing issue at the moment, as both sides recognize the need to establish a legal framework for the operation of stablecoins. This legislation is relatively simple, so it could be the first major crypto legislation. While there are still some differences between the House and the Senate, overall I don’t think these differences will be a barrier. There may be more discussions about the market structure after the stablecoin legislation, but the current focus is still on the stablecoin itself.

At the same time, there are also many discussions about the market changes that may be triggered after the passage of stablecoin legislation. Overall, the industry generally believes that this legislation will lay the foundation for future market structure. While this goal may take some time, the current discussion and focus is almost entirely on the stablecoin legislation itself.

It should be added that I feel their high recognition of stablecoin legislation when communicating with some MPs who tend to support cryptocurrencies. Although this may have a certain perspective deviation, overall, the differences between the House and the Senate are expected to be smoothly reconciled, and the prospects for the advancement of legislation are very optimistic.

Will stablecoins become the "Trojan Horse" in the crypto world?

Tarun:

I first heard about the concept of "Narrow Bank" in 2009. At that time, many people were discussing narrow bank legislation. While this makes me look a little old-school, at that time everyone was discussing this model: Should you have a restrictive bank that can only provide very basic types of returns.

Haseeb:

Can you explain what a narrow bank is?

Tarun:

The definition of narrow banks has evolved over time, but the core idea is to make banks simple. Especially after the financial crisis, it has been proposed whether banks should be subject to stricter regulation, such as restricting their participation in transactions or other complex activities? Or is it possible to create a bank that only provides basic services, such as deposits and loans, without involving other complex businesses? Interestingly, many early fintech applications were somewhat like "pseudo-narrow banks". They allow users to deposit money, but there are few earning products. Users may purchase Treasury bonds indirectly through these platforms, or provide Bitcoin services like Square, but these platforms themselves do not participate in complex investment activities, such as proprietary trading or bond portfolio investment.

In a sense, many bills on stablecoins remind me of the concept of narrow banks. The stablecoin itself has no profit, and the way the bank license is used behind it makes me very interesting. This narrow banking idea has been almost 20 years since it was first proposed, and it has finally been realized through blockchain technology. I think history is repeating itself, but the pace is very slow. After all, there was no new bank establishment or new bank license issued for ten years in the United States.

Robert:

My understanding is that the core of narrow banks is to keep all deposits in the Fed’s discount window and maintain 100% liquidity. This way, banks do not need investment analysts or loan specialists, and all deposits are handed over to the Fed to earn interest rates, and then paid to depositors after deducting a certain fee. In a sense, this is basically a branch of the Federal Reserve.

这种模式是一种全储备银行，具备100% 的流动性，无需担心流动性风险。理论上，只需要十几名员工就可以管理一个规模庞大的银行体系。但人们反对窄银行的原因在于，它会与现有的商业银行竞争。商业银行通过贷款扩展货币供应，而窄银行只是将资金存放在美联储，这会减少抵押贷款等优质资产的流动性。

Haseeb：

我认为美联储拒绝窄银行的原因是，它削弱了美联储直接干预货币供应的能力。 虽然抵押贷款仍然可以由私人放贷者提供，但一旦市场完全转向私人放贷，美联储将失去对货币供应扩张的直接控制权。

Robert：

从另一个角度看，美联储可能反而获得了更多干预能力，因为隔夜利率的调整会影响所有市场主体。

Haseeb：

如果还存在储备比例的话，确实如此。储备比例是第二个杠杆，这是一个非常强大的杠杆，可以瞬间改变货币供应。提高利率或降低利率显然是有零下限的，虽然，技术上你可以突破，但美国不会进入负利率。但这只是一个更慢的机制，进入市场，现在作为银行，你可以利用储备中的一切去进行投资，这种变化会更快。

Robert：

这让我想起了Genius Act。Tarun 说稳定币像窄银行，但我觉得它们并不完全相同。

Haseeb：

我认为他在说的是Stable Act，特别是因为它不允许收益。为什么在Stable Act 中不允许收益？大概是因为他们不想让它与商业银行竞争。

Robert：

Tarun 的意思可能是，这种限制让稳定币更像窄银行。但窄银行的核心是，它允许支付完全流动的利息。

Haseeb：

所以，如果收益不被禁止，你可以用稳定币创建一个窄银行。所以在Stable Act 下，你无法真正创建一个与商业银行竞争的窄银行。但在Genius Act 下，你基本上可以有一个稳定币，只持有国债，并返还所有的国债收益减去20 个基点或其他收益，这最终会成为一个非常简单的商业模式。

你可以说这就是Tether 的模式，显然他们不支付收益，但如果他们支付，那将是一个令人难以置信的商业模式。它的劳动效率非常高，他们有大约90 名员工，管理着超过1000 亿美元的资产。所以这是一项相当不错的业务。

Haseeb：

我觉得这个观点很有道理。稳定币或许是以更容易接受的方式重新引入窄银行概念，同时还能带来地缘政治优势。相比之下，窄银行只会对商业银行造成冲击，却无法对美元国际化带来任何帮助。稳定币的优势在于，即使它与商业银行竞争，它也能通过国际化扩大美元的整体市场规模。而窄银行无法做到这一点，因为它们在商业银行和窄银行之间是零和博弈。

从政策角度看，这就是为什么稳定币可能更受青睐的原因。但我也同意你的观点，Tarun，当中央银行家或银行高管审视这个问题时，他们可能会倾向于让银行执照持有者垄断稳定币的发行。这其实是「监管捕获」的一种体现，即通过限制市场参与者来保护既有利益。

Robert：

你对最终法案的看法是什么？你认为它会更像Genius Act 还是更像Stable Act，限制更少还是更严格？

Robert：

我认为在除收益方面之外，它将会更不严格。我认为目前的商业银行部门不希望看到稳定币上有收益。

Haseeb：

这让我想到银行业的一些奇怪现象。比如，我的Chase 银行账户，为什么我的现金不能自动转入货币市场账户赚取收益，而必须依赖我主动去操作？如果银行能自动完成这些操作就好了。但事实上，很多人不会主动操作，导致现金闲置在那里。这种现象其实很普遍，虽然用户可以通过一个按钮将资金转入货币市场账户，但很多人就是不去点。因此，券商通过这种懒惰赚了不少钱。

Robert：

券商的主要收入来源之一就是利息差。

Tom：

我听说过FTC 曾对花旗银行进行调查，因为他们推出了两种几乎相同的储蓄产品，但其中一种的利率较低。这说明银行通过信息不对称获利，而稳定币在某些方面规避了这种问题。

Robert：

你无法轻易降低新客户的利率，但你可以推出第二种产品，而所有旧客户仍然保持在没有上升的利率上。

Haseeb：

但讽刺的是，如果你将这视为现金账户，稳定币，即使你没有获得收益，比如说在Tether 或USDC 上没有获得收益，单纯在市场上借贷也能给你带来相当高的收益。

Tom：

目前市场借贷的收益率在5% 到6% 之间。窄银行的优势在于用户可以自由选择自己的风险配置，而不是由银行替你做决定。比如，你可以选择投向私人信贷或代币化的国债，而不是被银行捆绑在一起。如果用户愿意，他们可以自己去操作。

Haseeb：

这确实很合理。如果稳定币真的从银行系统中抽走了存款，我猜这可能是因为它能让用户的现金无论多懒惰都能产生收益。