Hey there,
I have been reading Jacob Bronowski’s Ascent of Man. The book’s core argument is that man shapes his environment as much as his environment shapes him. The teenager on TikTok and his caveman ancestor share their desire to express themselves in the mediums available. We create art because it gives us an escape from the realities around us.
Man is unique among creatures because our imagination allows us to empathise and live others’ stories without replicating their worlds. Society rewards artists for creating these make-believe worlds. JK Rowling (author of Harry Potter) and Epic Studios (makers of Fortnite) built billion-dollar fortunes by enabling make-believe worlds. But it has taken our society centuries to value art the way we do now. Here’s a quick journey to break down how that relationship has evolved.
Forex, bonds, and international trade boomed in 15th-century Florence, Italy. Many artists we admire today were patronized by the financiers of the time. Art was a symbol of status. Depending on an artist's work's rarity and stylistic complexity, a painting could retire a person. Art auction houses were set up in Sweden as early as the 17th Century. Around the same time, the Dutch were figuring out whether tulips were a good store of value.
About a hundred years later, auction houses like Christie’s and Sotheby’s became operational. However, art was still difficult to invest in before pooling capital became an option. In 1904, Andre Level created the first modern art fund. He and his friends bought contemporary art, held it for 10 years, and quadrupled their money. In the 1970s, the British Rail Pension Fund bought 2000 works of art across several categories, lending unprecedented credibility to art as an asset class.
Art went from a status symbol to an asset class. We created indices on art. Artnet (1989) and Mei Moses (2002) provided investors with benchmark datasets and kept track of prices in a severely fragmented market. The securitisation of art pieces was not a rampant practice yet. The modern world had ample cases where finance and art came to tussle against one another. Just a few years back, fans of Taylor Swift sent death threats to the studio that owned the rights to some of the work from her earliest days. The artist had tweeted that she was restricted from performing it again live.
A different instance of art and finance blending is that of Bowie Bonds. In 1997, the famous singer David Bowie partnered with Prudential Insurance Company to raise $55 million through a bond sale. Investors paid the money upfront for interest, which would be paid from the income generated by royalties of Bowie’s catalogue of 25 albums. Do you see a trend here? Almost everything we discuss in Web3, from buying art to streaming royalties, has happened in the past. We are not doing rocket science.
But there may be rocket fuel in that the infrastructure we are building can be used to move liquidity (dollars) and trace assets at a global scale. In the last century, an artist was discovered when their work was mentioned by a curator or shown at a large museum. The internet collapsed the cost of discovery for new artists. Blockchains enable the financial side of it. Anybody from any corner of the world could bid and own an artist’s work, as shown by Beeple’s $70 million sale of his NFT.
But it is one thing for a whale to acquire digital art and a whole different story for people to use them at a massive scale. Buying an NFT in the spot market, with no financing options, is akin to hoping everyone buys a house or car with no mortgages. The market for that is tiny. Even the iPhone required clever financing options involving carriers breaking down its cost for consumers to be onboarded. We have been looking at how the financial ecosystem around NFTs is evolving over the past few weeks. This piece is what came of it.
Sticky Traders, Stickier JPEGs
Let’s get a few things straight. A few promises around NFTs may not have entirely played out how we thought they would. For instance, artists, especially if they’re not creating visual art, have not been able to use these primitives to create a new source of income unless they already had distribution figured out. Some collections have done millions in sales, but have NFTs had the same impact Soundcloud had on music? Maybe it is too early to say (Yes, Audius exists).
Similarly, the gaming ecosystem has not taken kindly to NFTs branching into it. Sky Mavis has done an exceptional job with Axie Infinity and Ronin, and several studios are exploring how to better use on-chain assets within games. But ask somebody spending more than a few hours on a gaming console if they own an NFT in any of their games, and the answer may be a resounding no. However, things may soon change, as apps like Stepn and Axie Infinity gain permission from the App Store to be functional in specific markets, albeit with an Apple tax.
Quite recently, Meta began plans to drop all integrations of NFTs from their products. NFTs simply did not capture the psyche of the average retail customer. Signalling ownership of a Bored Ape on Instagram does not have the same clout as owning other Veblen goods issued by older brands such as Gucci or Louis Vuitton. Part of this scenario could be that many scammers use these NFTs to signal their wealth. What was supposed to be a status symbol among early adopters of NFTs as technology slowly transitioned is now unfortunately associated with scams in the minds of retail user.
So where have we seen any traction at all? Much of it is concentrated around trading NFTs as a spot asset, much like any other cryptocurrency. According to data from Nansen, the number of wallets trading NFTs on any given day has grown from around 10,000 to over 150,000 at the time of writing. The top 10 NFT wallets generated by profit (in ETH) have spent around 14,000 ETH to create profits of around 62,000 ETH. The median return among the most profitable thousand wallets is around 92%.
Is this subsection of users trading across collections? The median wallet here traded around 33 collections. There was one wallet that engaged with over 1400 collections. The top 100 wallets in terms of collections engaged with had interacted with at least 200 separate collections. So, at least among power users, it has become common to interact with many collections, likely in hopes that one of them will blow up and give the kind of returns Bored Apes NFTs did.
As we saw earlier, the gradual financialisation of art as we know it took centuries to play out. For a sense of scale, privately held art is worth $2 trillion, and the lending business is only about $20 billion. At just about $10 billion, the NFT market is currently much smaller in comparison. Still, the financial rails enabled by blockchains may translate to much higher capital velocity. As Avichal Garg from Electric Capital recently pointed out, NFTs do roughly 30 times eBay's transaction volume until its IPO ($12 billion vs $350 million for eBay). While the quality of this metric is up for debate, it is worth noting that users are willing to spend money acquiring these assets.
A new breed of early-stage startups is looking to capture a portion of that volume. We map out a few of the notable names below.
Price Discovery Engines
As with most assets, marketplaces are the fundamental block of NFTs as a market, as they help with price discovery. There are three fundamental models we have seen here.
First is that of a spot market, the kind you have on OpenSea. Holders of an NFT simply list their asset at a price they are willing to sell for, while those looking to buy can place a bid at which they are willing to take custody of the asset. This model was further improved by the likes of Sudoswap when they ported over the AMM model used in DeFi.
Sudoswap facilitates NFT trading via AMMs where the price changes based on a bonding curve (a function that determines how the price changes based on traders’ actions). There are two options at their disposal – a linear curve where the price changes linearly or an exponential curve where the price increases or decreases by the same multiplying factor.
Say two users want to sell 5 NFTs each of the same collection starting at 4 ETH, but one wants to sell them in linear increments of 0.5 ETH, whereas the other wants to sell in increments of 10%. The graph below shows what the difference in pricing between the two pools would look like over time. Assuming these two are the only pools for the collection, the lowest 8 out of the 10 listed above will be offered to the buyer by default. The advantage of such a model was liquidity for traders with large trade orders to enter and leave the market at ease.
Soon it became evident that a thin layer aggregating the best price for an NFT could be built atop marketplaces. Gem and Genie allowed traders to source the best price for large orders of NFTs at the click of a button. Aggregators could, however, threaten marketplaces once it grows to a point where more users go to the aggregator instead of the market.
Consider how Amazon has been launching its electronics accessories in the recent past. Sensing this threat, OpenSea acquired Gem a year back but was too slow to launch a token. Blur noticed an opportunity and launched in Q3 2021, with a potential token airdrop as an anchor in their marketing. Before one realised it, much of the volume began aggregating around trader-oriented interfaces like Blur.
Once it has sufficient users, the natural extension for an NFT marketplace is to begin offering leverage. Accordingly, this act allows more users to generate higher volumes, potentially translating into better fees for you. This works out through a lending model, which we cover further below. A different method is through derivatives of the instrument, where you track the asset's price — and trade against it without having custody of the actual asset itself. (Some finance jargon is coming right up; do skip right to lending if it doesn’t interest you)
Let’s say Sid is bullish on Bored Apes Yacht Club (BAYC) NFTs, but he doesn’t want to risk 60 ETH. If the NFT price is around 60 ETH and his budget is only 10 ETH, then Sid can use a protocol that offers up to 10X leverage to take a position on BAYC without actually buying the NFT. These futures, like perpetual futures for fungible tokens, use the funding rates concept to balance the long and short sides and ensure the price tracks the collection’s floor price.
A crucial consideration here is how the protocol fetches floor prices (the lowest price an NFT can be sold at) because the difference between the market price on the futures platform and the index price (the collection’s floor price) determines the funding rate. Funding rates drive incentives for those long or short on the platform.
If the futures platform uses floor prices directly from the marketplace, it is exposed to manipulation (wash trading and spoofing are common among NFTs). Nftperp, a perpetual futures platform for NFTs, uses a True Floor Price that is resistant to the problems mentioned above. They filter out transactions that are outliers or seem like wash trading, and then use the time-weighted average price (TWAP) to estimate the correct asset price.
JPEG Morgans Of The Metaverse
Lending has existed as long as money. One may want to ‘borrow’ an NFT for several reasons. Three models have emerged. The first model involves lenders and borrowers matching on-chain, peer-to-peer. The second involves peer-to-peer lenders adding NFTs to a pool that provides loans for collateral. The last involves collateralised debt positions, as with MakerDAO and DAI.
Peer-to-peer is the simplest form of lending. Platforms like NFTfi allow borrowers to list their NFTs to take out loans against them. Once the borrower accepts an offer from one of the interested lenders, the NFT is put in an escrow account. It is released to the borrower if they repay the loan before the expiry period. If the borrower fails to repay the loan, the NFT is released to the lender. Recollecting debt collateral and giving it to a lender is something smart contracts are exceptionally good at as anyone that has had a loan liquidated in DeFi would know.
This is one of the simplest ways to facilitate NFT-based loans. Another positive is that the model doesn’t rely on external factors such as a floor price oracle, making it safer than other lending practices. The downside is that matching takes a long time, and there’s often liquidity scarcity.
The Web2 world has ample examples of peer-to-peer lending platforms like LendingClub, Zopa, and Prosper offering alternatives to banks. Borrowers enjoy better interest rates on these platforms, but they have not scaled massively. A possible explanation could be that since there is no government guarantee, lenders are apprehensive about giving out loans using these platforms. The risk-premium for issuing loans in peer-to-peer markets is high, but so are the default risks.
The P2P model of lending can be compared to order book-based trading. Borrowers and lenders must agree on the details before the loan is initiated. The time taken to find these matches makes the model difficult to scale. Decentralised exchanges (like EtherDelta) started with off-chain order books, but the lack of liquidity obstructed their scale. Uniswap was a real breakthrough because it did not mandate buyers and sellers to agree to a price. In a way, putting passive liquidity to use was Uniswap’s genius. Similarly, the P2Pool model is like the AMM, where users borrow and lend from a pool instead of another user.
This design is akin to a money market where users tap into the money market for loans, but their collateral is an NFT instead of ERC-20 tokens or ETH. BendDAO was among the first few projects to facilitate this, and it has a lending pool of NFTs. Think of something like AAVE, where users deposit/lend different fungible assets, or ETH, and can take out ETH (or other available assets) loans instantly. In BendDAO’s case, the NFTs a borrower deposits act as collateral. In case the value of the collateral falls below the defined threshold, the protocol needs to liquidate the NFT.
BendDAO also allows users to pay a down payment that is a certain percentage (varies based on a collection) of the price of the NFT and buy the NFT. The gap is bridged with a flash loan from AAVE. The flash loan is repaid by the protocol before the protocol recovers the loan from the buyer. The buyer also becomes the borrower. This mechanism relies a lot on price oracles, which, to be honest, is not the best of practices.
Scaling this model is an issue since external factors like NFT price and liquidity impact the collateralisation ratios. This means the governing body must vet new collection listings, which creates a bottleneck. A different challenge emerges when there is bad debt in the system. Assume a person gave NFTs as collateral to the system and took a loan. There needs to be sufficient liquidity if it is under-collateralised due to market volatility and needs to be liquidated. (Remember FTX’s FTT loan? I do. Every day.)
Rapid sales of the NFT can hurt the price of the floor, resulting in more losses for the protocol—a liquidity cascade of sorts, a scenario where selling off one NFT to recoup a bad debt leads to more bad debt entering the system. BendDAO faced a similar liquidity crunch in October 2022, and the protocol had to take corrective measures to ensure that lenders did not lose their ETH. You can read more about this situation in this tweet from February.
There are tradeoffs to both the p2p model (scale, capital efficiency) and the p2pool/CDP model (limited coverage, oracle dependency). A third option, coined the “Automatic Tranche Maker (ATM)” has recently been introduced to the market by MetaStreet Labs.
In the ATM, lenders choose a price they will lend to an NFT collection. Different lenders can select different prices, which are then stacked on top of each other to offer instant liquidity to borrowers.
For example - three lenders want to lend to CryptoPunks, but all three have different risk tolerances. Let’s presume, Cryptopunks is worth 100 ETH, and you have a mix of lenders willing to offer different amounts of ETH for the NFT.
Lender A wants to lend up to 10 ETH per punk, Lender B up to 30 ETH, and Lender C up to 45 ETH.
All three lenders would deposit into the same CryptoPunk pool, indicating their own unique prices.
When a borrower comes to the pool, they’ll see instant liquidity at 45 ETH for their CryptoPunk, and happily take the proceeds. Behind the scenes, the three lenders each get their desired outcome .
Lender A takes the 0-10 ETH position, Lender B takes the 10-30 ETH position, and Lender C takes the 30-45 ETH position.
While the borrower will see a single interest rate, the Lenders will split the interest disproportionality, such that Lender C gets the bulk of the return (in exchange for taking on the bulk of the risk), with the share of interest waterfalling down to each subsequent lender in the pool.
The ATM model enables collaborative lending strategies via pooling, which provides lower capital costs than peer-to-peer lending markets while maintaining each user’s ability to set their risk profiles independently of other participants or 3rd party oracles.
Blur came up with a P2P lending design called Blend that takes care of some of these drawbacks of P2Pool lending. Blend borrows heavily from how loans work traditionally. It removes the need for oracles (as is the case with most P2P designs), and by default, there is no expiration date for these loans. The lender or borrower can get out at any time. The lender can initiate an auction whenever they want to get out of the position for whatever reason (say they find a higher interest rate elsewhere). The borrower has a stipulated time to respond by paying off the principal and interest.
If the borrower fails to comply, the loan is open for a takeover from bidders. If a bidder cannot be found, the lender takes possession of the NFT (collateral). This is like how loans work in real life. From the borrower’s perspective, if they want to get out, they can repay the loan at any point and get out of it. However, they must keep tracking whether their lender has initiated an auction.
The CDP-based model borrows a lot from MakerDAO. This type of lending uses a synthetic asset that a user can mint against their collateral NFT. JPEG’d was among the first protocols to launch CDP-based lending for NFTs. Just as users can mint DAI by locking collateral into Maker’s vaults, users can mint pUSD by locking NFTs into JPEG’d’s vaults. Borrowers can retrieve their NFTs by repaying loans (along with interest). The collections that can be used as collateral are a curated subsection of NFTs that are whitelisted, which means a user can get an instant loan against their NFTs.
Another advantage is that the marginal cost of issuing a new loan is zero. From the supply side, the model is scalable. However, it is not without flaws. It needs an oracle, which we know is risky as the oracle itself is susceptible to manipulation. Mandatory whitelisting creates a bottleneck and thus acts as a barrier to scale. And lastly, only the native stablecoin can be minted, which means that it requires wide acceptance for the project to scale and cannot ride on the network effects of the existing stablecoins.
It Hertz To Return Rentals
This model can be thought of as renting a house. When someone rents an NFT, they inherit all its benefits for the period. At the end of the rental period, the NFT is returned to the owner. Renting NFTs can have different use cases. Say you needed an NFT to play a particular game for a few hours: Instead of buying the NFT for hundreds of dollars, you could rent it out for a few dollars in exchange for splitting a portion of any capital gains with the NFT’s owner. This is how yields in games like Axie Infinity took off.
Guilds retained the ownership of the NFT, but players would use it in-game and generate revenue which was shared with the guild. This trend will likely pass on across ecosystems where NFTs have a sought-after utility. A different place this could play out is with Stepn. The app rewards users for walking with certain shoes. However, the (virtual) shoes can cost thousands of dollars. What if you could rent it to a third party, and they split the rewards with you? You would unlock a whole new level of smart contract-enabled capitalism.
There isn't sufficient attention within the ‘metaverse’ as we used to know it to support a large NFT rental ecosystem. The incentives are simply not there, but fashion is one of the places it could take off. Say you own a limited edition gun that costs $400k, like this one in Counter-Strike. There is a credible case to argue that a third party would rent it from you to stream it on Twitch. Now, there’s the caveat that Counter-Strike is a multi-decade-old brand with millions of eyeballs. Web3 native games are simply not there yet.
A different place this could work is with social clubs. Instead of a single person spending tens of thousands of dollars to access a digital-first community like Foster, multiple people could pay to buy an NFT, which in turn lets them access the community. Fan clubs of artists could pool together to purchase NFT-based tickets that are valid for a season, and individuals in the club could rotate based on who can attend an artist’s events. All of these are hypotheticals that require more retail adoption of these primitives. For now, all we have is the technology.
Beyond Speculation
In what may be one of the weirdest comebacks in technology, vinyl records now sell more than physical CDs do. The reason is simple. If you stream music from a platform like Spotify, you stand to lose all of it if the relationship between an artist and the platform (Spotify) sours. Many of my favourite tracks by Jay Z were off Spotify for a while when he was trying to scale Tidal into a meaningful business.
Owning physical records means you can continue to access the music long after the platform is down. If anything, it shows a desire to own assets directly linked to artists people admire. You can see a similar trend in gaming. Quite recently, a gun in CS: GO sold for $400k. Many behavioural patterns we get excited about in Web3 exist in the traditional realms.
Passing on royalties or ownership of digital goods does not necessarily require issuing a random token that must be speculated on. Userbases in Web3 don't scale because we often get distracted by the incentives (tokens) instead of being obsessed about what would onboard and retain users.
The challenge is that without a critical mass of users, you cannot bootstrap a functional economy in an app. The assumption is that building subpar products is often acceptable because the customer is like a speculator, and incentives (like tokens) will blindside consumers long enough. But I don't see that happening in a 4% interest rate environment. The industry will evolve only if we build relatable products with blockchain rails the user does not see. Tools like account abstraction are already enabling users to do that.
Other technological evolutions that may accelerate the process are occurring. One example is the recently introduced ERC-6551 standard for token-bound accounts. This helps create a smart contract wallet for every ERC-721 token (NFT). It means NFTs can own assets and interact with applications. For instance, you could play a game where an NFT represents your character. And that character, in turn, could accumulate tokens as in-game rewards. So if you transfer the NFT, you transfer all the assets accrued while playing the game. Or you could buy NFTs representing entire token portfolios, such as the one shown below.
I have been thinking if better infrastructure always translates to more economic activity. Ironically, one place this has happened is the city I live in. Dubai became a hub for global trade by improving infrastructure and creating economic policies that attracted entrepreneurs and investors. And they did it for over a century. Blockchains are similar to cities in that they are empty and often enable very little economic activity. But reducing fees (monetary policy) and tooling (infrastructure) could onboard developers (entrepreneurs & investors) that want to build atop these blockchains.
NFTs need an application to do with it what Pokemon GO did for AR and ChatGPT did for AI. Applications like Axie and Stepn are examples of what happens when an application makes it easy for users to interact with these cryptographic primitives. But I fail to believe we can come up with only two examples. There’s space for more.
Consider dropping in your deck here if you are building applications that can onboard a few million users to these on-chain primitives. We’ll see you next with some work we have been doing on Web3 social networks.
Joel John
Disclosure: Entities I am associated with have exposure to Metastreet. One of the firms mentioned under lending. Connor from Metastreet also helped formulate the thesis during the early stages of this piece.
If you liked reading this, check these out next:
Telegram and Pitch Decks
Join in with ±3000+ researchers, investors, founders & overall great human beings. We don’t exactly talk much, but it would help you stay close to what we are focusing on & connect with others building cool things.
Fill out the form below if you are a founder building cool things and in the process of raising money or looking for feedback on what you are pursuing. We like the builders.
Solid write up. Love that chart of adoption. All the pieces are there - we just need someone to bake that cake