The Rise of NFTFi

NFTFi's Emergence

Article author: JOEL JOHN, SAURABH

Article translation: Block unicorn

I have been reading Jacob Bronowski’s “The Ascent of Man.” The central idea of the book is that the degree to which humans shape their environment is equivalent to the degree to which the environment shapes humans. A teenager on TikTok has something in common with his cave-dwelling ancestor in his expression of desires; it’s just that the mode of expression is different, and we create art because it allows us to escape from the reality around us.

Humans are unique among animals because our imagination allows us to empathize and experience the stories of others without replicating their world. Society rewards artists for creating these fictional worlds, and J.K. Rowling, the author of “Harry Potter,” and Epic Studios, the production company behind “Fortnite,” have amassed billions of dollars through the creation of fictional worlds. However, it took centuries for our society to value art as it does now. Here is a brief history to help parse how this relationship developed.

In 15th century Italy, Florence rose as a hub for foreign exchange, bonds, and international trade. Many of the artists we admire today were sponsored by the financiers of the time, and art became a symbol of status. Depending on the scarcity of the artwork and the complexity of the style, a painting could retire a person. Sweden established an art auction house in the early 17th century. Meanwhile, the Dutch were pondering whether tulips made for a good store of value.

About a century later, auction houses such as Christie’s and Sotheby’s began operating. But investing in art was still difficult before fundraising became an option. In 1904, Andr√© Level founded the first modern art fund. He and his friends bought contemporary art, held it for ten years, and doubled their assets. In the 1970s, the British Rail Pension Fund bought 2,000 assorted works of art, earning unprecedented credibility for art as an asset class.

Art has transformed from a symbol of status to a category of assets and we created the art index. Artnet (1989) and Mei Moses (2002) provide benchmark datasets and track prices in a severely dispersed market for investors. The securitization of art has not yet become a common practice and in the modern world, finance and art often clash. Just a few years ago, Taylor Swift’s fans threatened death to the recording studio that owned the copyrights to her early works. The artist tweeted that she was restricted from performing these works live.

Another example of the fusion of art and finance is the Bowie bond. In 1997, the famous singer David Bowie collaborated with Prudential Insurance Company to raise $55 million through bond issuance. Investors paid this amount upfront to receive interest, which would come from the income generated from Bowie’s 25 album copyrights in the future. Can you see a trend here? Almost everything we’re discussing in Web3, from buying art to streaming royalties, has happened in the past. The fields we’re involved in aren’t particularly complex or esoteric.

But the infrastructure we’re building can be used to flow liquidity (dollars) and track assets globally, and the potential of infrastructure and technology to drive the development of the art and finance industries is enormous. In the past century, an artist’s discovery was achieved through a curator mentioning it or exhibiting their work in a large museum. The internet has lowered the cost of discovering new artists and the blockchain has made it possible in finance. Anyone around the world can bid and own a work by an artist, just as Beeple sold his NFT for $70 million.

However, acquiring digital art by a whale (big holder) and using it on a large scale is a completely different matter. Buying NFTs in the spot market without financing options is like hoping everyone will buy a house or car without a mortgage. This market is very small and even the iPhone needed to work with carriers and clever financing schemes to spread its cost to consumers to make it popular. Over the past few weeks, we’ve been following the developments of the financial ecosystem around NFTs. This article is the result of that.

Sticky Traders, More Difficult to Leave JPEGs

Let’s get something straight, some of the promises about NFTs may not have been fully realized as we expected. For example, artists, especially those who do not create visual art, cannot harness these native features to create new revenue streams unless they have solved the distribution problem. Some NFT collectibles have indeed achieved sales of millions of dollars, but will NFTs have the same impact on the music industry as Soundcloud did? Perhaps it is still too early to say (but Audius, an NFT music distribution platform, is already changing the way music is distributed).

Similarly, the gaming ecosystem is not very receptive to NFTs entering it. Sky Mavis has done an excellent job in Axie Infinity and Ronin, and some studios are exploring how to better use on-chain assets in games. But if you ask someone who spends more than a few hours on their game console whether they have NFTs in their game, the answer may be a resounding no. However, the situation may change quickly as applications like Stepn and Axie Infinity gain app store functionality in specific markets, albeit with the need to pay Apple taxes.

Recently, Meta began planning to remove all integration of NFTs from their products. NFTs have not captured the hearts of regular retail customers. Displaying ownership of a Bored Ape on Instagram has not been as impactful as owning other Veblen goods released by established brands like Gucci or Louis Vuitton, partly because many scammers use these NFTs to show off their wealth. The status symbol among early NFT adopters has gradually become associated with scams in the minds of retail users, which is very unfortunate.

So where have we seen some progress? A large part of it is focused on trading NFTs as spot assets, just like any other cryptocurrency. According to Nansen’s data, the number of wallets trading NFTs on a daily basis has grown from about 10,000 to over 150,000. The top 10 most profitable NFT wallets spent a total of about 14,000 ETH and generated profits of about 62,000 ETH, calculated in ETH. The median profit in the top 1,000 most profitable wallets is about 92%.

Are these users trading between different collectibles? Here, the median wallet has interacted with about 33 collectibles. One wallet has interacted with over 1,400 collectibles, and wallets in the top 100 for interacting with collectibles have engaged with at least 200 unique collectibles. Thus, interacting with many NFT collectibles has become a common practice, at least among power users, and likely reflects a hope that one of the collectibles will be a big success and provide returns similar to those of Bored Ape NFTs.

As we have seen before, art as a gradually financialized process that we know of took centuries to complete. Private art holdings are valued at approximately $2 trillion, whereas the lending business is only about $20 billion. And the current NFT market is only about $10 billion, relatively small. However, the financial infrastructure provided by blockchain may lead to higher capital velocity. As Avichal Garg of Electric Capital recently pointed out, before eBay’s IPO, NFTs were trading at approximately 30x eBay’s volume (NFTs at approximately $12 billion versus eBay at approximately $350 million). Although the quality of this metric is still up for debate, it is noteworthy that users are willing to spend money purchasing these assets.

A cohort of emerging startups is attempting to capture some of this trading volume, and below we list some projects worth watching.

Price-Discovery Trading Platforms

Like most assets, NFT trading platforms are a fundamental module of the market, helping with price discovery. Here we see three basic models.

The first is a spot market, much like the market on OpenSea. Users holding NFTs simply list their assets at the price they are willing to sell them for, and users looking to buy can submit bids for the asset they are willing to receive. This model has been improved upon by platforms like Sudoswap as they move to the AMM (automated market-making) model used in DeFi.

Sudoswap facilitates NFT trading through AMM with prices that fluctuate based on a bonding curve (a function that determines price changes based on traders’ behavior). They have two options: a linear curve, where prices change linearly, or an exponential curve, where prices increase or decrease by the same multiplier factor.

Suppose two users both want to sell the same series of 5 NFTs at a starting price of 4 ETH, but one user wants to sell with a linear increment of 0.5 ETH and the other wants to sell with a 10% increment. The chart below shows the price difference between these two pools over time. Assuming these two pools are the only pools for the series, buyers will default to the lowest 8 of the top 10 listed. This model provides liquidity for traders of large orders, allowing them to easily enter and exit the market.

Soon, people realized they could build a thin layer on top of the market to aggregate the best prices for NFTs. Platforms like Gem and Genie allow traders to get the best price for bulk NFTs with the click of a button. However, once aggregators grow to the point where users are more likely to go to the aggregator than to the market, it may pose a threat to the market.

Imagine the impact of Amazon’s recent launch of its electronic accessories market on its competitors. OpenSea acquired Gem a year ago, but was too slow to launch tokens. Blur saw an opportunity and launched its own platform in Q3 2021, using potential token airdrops as their marketing scheme to promote their market. Before people knew it, most of the trading volume had already begun to concentrate on trader-oriented platforms like Blur.

Once the NFT market has enough users, the natural extension is to start offering leveraged trading. Doing so can generate higher trading volume from more users, potentially bringing you better fees. This is done through a borrowing and lending model, which we’ll further explain below. Another method is through derivative instruments that track the price of an asset and enable trading without actually owning the asset. (There will be some financial terminology coming up, so feel free to skip to the borrowing and lending section if you’re not interested.)

Suppose Sid is bullish on Bored Apes Yacht Club (BAYC) NFTs, but doesn’t want to risk 60 ETH. If the NFT is priced at around 60 ETH and his budget is only 10 ETH, Sid can use a protocol that offers up to 10x leverage to trade BAYC without actually buying the NFT. These futures contracts, like perpetual futures for fungible tokens, use the concept of funding rates to balance longs and shorts and ensure price tracking of the series’ floor price.

The key consideration here is how the agreement captures the floor price (the lowest price of the NFT), as the difference between the market price on the futures platform and the index price (the aggregated floor price) determines the funding rate, which drives the incentive for long or short positions on the platform.

If the futures platform uses the floor price directly from the market, it will be vulnerable to manipulation (wash trading and other fraudulent behavior is common in NFTs). Nftperp is a perpetual futures platform designed specifically for NFTs that uses a real floor price mechanism that can resist the above problem. They filter out abnormal or wash trades and use a weighted average price (TWAP, the average price over a period of time) to estimate the correct asset price.

The JPEG Morgan of the Metaverse

Lending as a financial instrument has been around for a long time, and people may want to “borrow” an NFT for various reasons. There are three borrowing and lending models: the first is peer-to-peer borrowing and lending on-chain, the second is peer-to-peer lenders adding NFTs to a fund pool that offers collateral loans, and the last is collateralized debt positions, similar to MakerDAO and DAI.

Peer-to-peer borrowing and lending is the simplest form of NFT lending, and platforms like NFTfi allow borrowers to pledge their NFTs for a loan. Once the borrower accepts an interested lender’s quote, the NFT is placed in escrow. If the borrower repays the loan before the due date, the NFT will be returned to the borrower. If the borrower fails to repay the loan, the NFT will be returned to the lender. Smart contracts are excellent at recovering debt collateral and giving it to the lender, as anyone who has experienced a loan liquidation in DeFi will know.

This is one of the easiest ways to facilitate NFT-based lending, and another positive factor is that this model does not rely on external factors such as floor price oracles, making it safer than other lending methods. The downside is that matching takes a long time and there is often a shortage of liquidity.

There are many examples of peer-to-peer lending platforms in the Web2 world, such as LendingClub, ZoBlocking, and Prosper, which offer borrowers a different choice than traditional banks. Borrowers enjoy better rates on these platforms, but the scale has not expanded significantly. The possible explanation is that, without government guarantees, lenders are skeptical about using these platforms for loans. The risk premium for lending in peer-to-peer markets is high, but the default risk is also high.

The peer-to-peer lending model can be compared to order book-based trading. Before the loan is initiated, the borrower and the lender must agree on the details. The time it takes to find these matches makes this model difficult to scale. Decentralized exchanges (such as EtherDelta) initially adopted off-chain order books, but liquidity issues prevented them from scaling. The real breakthrough of Uniswap is that it does not require buyers and sellers to agree on a price. In a sense, Uniswap’s cleverness lies in utilizing passive liquidity. Similarly, the point-to-pool (P2Pool) model is similar to an automated market maker (AMM), where users borrow and lend from the pool instead of from other users.

This design is similar to a money market, where users can use a money market to borrow, but their collateral is an NFT rather than an ERC-20 token or ETH. BendDAO is one of the earliest projects to promote this model, with an NFT lending pool. It can be compared to AAVE, where users can deposit/borrow different fungible assets or ETH and immediately receive an ETH (or other available assets) loan. In the case of BendDAO, the NFTs deposited by borrowers serve as collateral. If the collateral value falls below a defined threshold, the protocol needs to liquidate the NFTs.

BendDAO also allows users to pay a certain percentage of the down payment (depending on the collectible) to purchase the NFT. This gap is bridged by a flash loan obtained from AAVE. The flash loan is repaid by the protocol before the loan is recovered from the buyer, who also becomes the borrower. This mechanism depends largely on price oracles, which, honestly, is not the best practice.

Expanding this model presents problems, as external factors such as NFT prices and liquidity will affect the collateral ratio. This means that governance organizations must review new collectibles for listing, which can lead to bottlenecks. Another challenge arises when there are bad debts in the system. Suppose someone gives an NFT as collateral to the system and borrows money. If the collateral is insufficient due to market fluctuations, there must be sufficient liquidity for liquidation. (Remember FTX’s FTT loan? I think about it every day.)

Quickly selling NFTs may have a negative impact on the floor price, leading to more losses for the protocol and creating a liquidity cascade effect. In this situation, selling one NFT to make up for bad debt may result in even more bad debt entering the system. BendDAO faced a similar liquidity crisis in October 2022 and had to take corrective measures to ensure that borrowers did not lose their ETH. (You can look up news about BendDAO’s bad debt risk issue in February.)

There are trade-offs between P2P models (scale, capital efficiency) and P2Pool/CDP (collateralized debt position) models (limited coverage, reliance on price oracles). MetaStreet Labs recently introduced a third option called the “Automatic Tranche Maker (ATM).”

In ATM, borrowers choose the price at which they are willing to lend to the NFT series. Different borrowers can choose different prices, and these prices are stacked together to provide instant liquidity to the borrower.

For example, suppose there are three borrowers who want to lend to CryptoPunks, but each of these three borrowers has a different risk tolerance. Assume that CryptoPunks is worth 100 ETH, and you have a group of lenders who are willing to lend different amounts of ETH to the NFT.

  • In this example, borrower A is willing to lend up to 10 ETH per CryptoPunk, borrower B is willing to lend up to 30 ETH per CryptoPunk, and borrower C is willing to lend up to 45 ETH per CryptoPunk.
  • These three borrowers deposit funds into the same CryptoPunk funding pool, each with a different asking price.
  • When the borrower comes to the CryptoPunk pool, they will see that their CryptoPunk can immediately obtain liquidity of 45 ETH, and then happily receive this amount. Behind the scenes, all three borrowers get what they want.
  • Borrower A gets a position of 0-10 ETH, borrower B gets a position of 10-30 ETH, and borrower C gets a position of 30-45 ETH.

In the ATM model, the borrower sees a single interest rate, and the creditors disproportionately share the interest to give borrower C the majority of the returns (as an exchange of taking on the majority of the risk), with the share of interest gradually allocated to each subsequent creditor in the funding pool.

The ATM model provides collaborative lending strategies through a fund pool, which achieves lower capital costs than peer-to-peer lending markets and allows each user to independently set their risk configuration without the influence of other participants or third-party oracles.

Blur proposed a P2P lending design called Blend, which addresses some of the shortcomings of P2Pool (individual-to-fund pool) lending. Blend largely borrows from the way traditional loans work. It eliminates the need for oracles (which is the case in most P2P designs) and these loans are, by default, without an expiration date. Creditors or borrowers can exit at any time. Creditors can initiate an auction at any time to sell their position for any reason (such as finding a higher interest rate elsewhere), and borrowers have a set amount of time to respond by paying back the principal and interest.

If the borrower fails to fulfill the loan contract, the loan will be taken over by a bidder. If no bidder can be found, the creditor will take ownership of the NFT (collateral), which is similar to how loans work in real life. From the borrower’s perspective, if they want to exit, they can pay back the loan and terminate the loan contract at any time. However, they must be vigilant to ensure that the creditor does not initiate an auction.

The CDP (collateralized debt position) model borrows heavily from MakerDAO’s design. This type of lending uses a synthetic asset that users can mint with their collateral NFT. JPEG’d was one of the earliest protocols to introduce CDP lending for NFTs. Just as users can mint DAI by locking up their collateral in Maker’s vaults, users can mint pUSD by locking up their NFTs in JPEG’d’s vaults. Borrowers can retrieve their NFTs by repaying the loan (plus interest). The collectibles that can be used as collateral are a subset of screened NFTs, which means that users can get a loan for their NFTs instantly.

Another advantage is that the marginal cost of issuing new loans is zero. From the supply side, the model is scalable. However, it also has some drawbacks. It requires oracles, and we know that oracles themselves are vulnerable to manipulation. The mandatory whitelist setting creates bottlenecks and obstacles to scaling. Finally, only local stablecoins can be minted, which means that the project needs to be widely accepted for scalability and cannot leverage the network effects of existing stablecoins.

NFT Leasing

This model can be thought of as renting a house. When someone leases an NFT, they enjoy all of its benefits during the lease term. At the end of the lease term, the NFT will be returned to the owner. Renting an NFT can have different uses, for example, if you need an NFT to play a specific game for a few hours: you can rent it instead of spending hundreds of dollars to buy that NFT, rent it for a small fee, and share a portion of any capital gains with the NFT owner, which is how the yield in games like Axie Infinity grows rapidly.

The guild keeps ownership of the NFT, but players use it in the game and generate revenue, which is shared with the guild. This trend could continue in ecosystems where NFTs have popular practical uses, and another possible scenario is in Stepn, an application that rewards users for walking while wearing specific shoes. However, (virtual) shoes may be worth thousands of dollars. If you can rent it to a third party and share rewards with them, you will open up a whole new level of smart contract-driven capitalism.

In the “metaverse” we know, there is currently not enough attention to support a large NFT leasing ecosystem. Simply put, there are currently no corresponding incentive mechanisms, but it is one of the areas that may take off in vanity. For example, if you own a limited edition gun worth $400,000, like a gun in Counter-Strike. There is reason to believe that a third party will rent it from you and then show off on Twitch. Of course, the premise is that Counter-Strike is a brand with decades of history and millions of viewers, while Web3-native games have not yet reached this level.

Another possible area is social clubs. Instead of one person spending tens of thousands of dollars to visit a digitally-focused community like Foster, several people can pay to purchase an NFT and use it to enter the community. Fans of an artist can pool money to buy NFT-based tickets that are valid for a season, and individuals in the club can rotate based on who can attend the artist’s events. These are assumptions that require more retail users to embrace these native features. Currently, we only have technology without users.

Speculation Beyond

Perhaps one of the strangest comebacks in tech is the resurgence of vinyl records, which now outsell physical CDs. The reason is simple. If you stream music from platforms like Spotify, you risk losing access to all your music if the relationship between the artist and the platform (like Spotify) sours. When Jay Z tried to make Tidal (a music streaming platform) a meaningful business, some of my favorite Jay Z tracks were temporarily removed from Spotify.

Having physical copies of records means you can still listen to music after a platform shuts down. This suggests that people want to own assets that are directly related to their favorite artists. Similar trends can be seen in the gaming world as well. Recently, a gun in CS GO was sold for $400,000, and many of the behaviors we’re excited about in Web3 exist in traditional domains too.

Passing royalties or ownership of digital goods does not necessarily require issuing a speculative token. The Web3 community cannot scale because we’re often distracted by incentives (tokens) and ignore the importance of attracting and retaining users.

The challenge is that you cannot launch a functional economy in an application without a sufficient number of users. Assuming that building subpar products is acceptable is like treating customers like speculators, and incentives (such as tokens) will only mislead consumers for so long, especially in a 4% rate environment. Only by building associated products with blockchain technology invisible to users can the industry develop, and tools like account abstraction have already enabled users to do this.

Other technological advances that could accelerate this process are also happening. One example is the recently introduced ERC-6551 standard for accounts bound to tokens. This helps create smart contract wallets for every ERC-721 token (NFT), which means that NFTs can own assets and interact with applications. For example, you could play a game where an NFT represents your character. And that character can earn tokens as rewards in the game. So if you transfer the NFT, you’re transferring all the assets you’ve accumulated in the game. Or you could buy an NFT that represents an entire portfolio of tokens, like the example below.

I have been thinking about whether better infrastructure always translates into more economic activity. Ironically, this has happened in my city. Dubai has become a global trading center by improving infrastructure and implementing economic policies that attract entrepreneurs and investors. And they have been doing it for over a century. Blockchain is similar to cities in that they are both often empty and rarely promote economic activity. But lowering costs (monetary policy) and providing tools (infrastructure) can attract developers (entrepreneurs and investors) who want to build applications on these blockchains.

NFTs need an application, just as Pokemon Go did for augmented reality and ChatGPT did for artificial intelligence. Axie and Stepn are examples of what happens when an application enables users to easily interact with these cryptographic primitives. But I don’t believe we can only come up with two examples; there are more practical spaces for NFTs.