Hey There,
Last week, Uniswap announced a fee for individuals using the product directly through their website. It is being referred to as an “interface fee”. At 0.15%, it is a small sum to be paid by anyone using the product until you realise that Uniswap does not own the inventory of assets on its exchange. They have little marginal cost for enabling each new transaction on the product. I thought it would be worth exploring as it marks a time when business models in the industry are evolving rapidly.
Before we dive into the unit economics of interfaces, it helps to revisit what I mentioned in my piece on aggregation in Web3. The nature of smart contracts allows protocol developers to source third-party liquidity by incentivising users with either tokens or platform fees. When you allow your money to stay at a bank, part of the promise is to receive a return on that idle capital.
But banks cannot print money out of thin air themselves. (You could argue central banks can, but the commercial banks can't.) They are required to find sources of yield that generate dollars. There are marginal costs incurred in generating yield. On the other hand, DeFi platforms see exponential revenue growth without costs surging.
Banks need to hire personnel as they scale from managing $1 billion in deposits to $10 billion in deposits. GMX or Uniswap does not see a similar cost growth as smart contracts facilitate most platform transactions. You don’t need human labour in proportion to the transaction volumes.
Incentives for Liquidity
Token incentives are a powerful mechanism to drive liquidity. Aave and Compound tokens were given to bootstrap a market for lending ETH and receiving a fixed yield. Similarly, Uniswap tokens incentivised people to put ETH and USDC into a pair on a smart contract. Unlike banks, protocols in crypto can mint assets to incentivise user behaviour. Blur took this to an extreme when it came to NFTs.
Their model incentivised users to put bids and asks closest to the spot price of an NFT. So, if you were a liquidity provider on Blur and an NFT was trading at $100, you would receive more points for providing liquidity at $99 instead of $90, which may have been a better price to acquire the NFT. Blur's model worked so long as the token incentives justified the losses incurred in providing liquidity for large NFT collection holders. There is a distinction to be made here between liquidity and efficiency.
An efficient market reflects the collective information market participants hold about an asset in the least amount of time. In this sense, a CeFi platform (like Binance) can be more efficient than a DEX (like Uniswap), as there's a minor lag depending on the blockchain where an exchange happens.
On the other hand, a liquid market is one where large position sizes can be purchased or sold at prices close to the spot price. Blur brought liquidity to an illiquid market and, by extension, arguably made it more efficient.
Take, for instance, this example of a trader who sold $9 million worth of Bored Apes in a single transaction. By allowing large holders to enter and exit an NFT ecosystem, the platform allowed NFTs to behave closer to tokens (in terms of price) than a more illiquid asset (like real estate). Why does this matter? The platform does not take the 'risks' on traded assets.
The risk is instead handled by users who hope to gain through the platform's tokens. As a token declines in price, the incentive for providing liquidity wanes rapidly.
Marginal Costs
Why does this matter? To understand, we must look towards Web2 aggregators like Uber or Spotify. When Uber launches in a new city, there are marginal costs in managing drivers. Surely, the firm does not often own the fleet or have marginal costs in hiring drivers. But there's labour involved in managing the fleet. Your operational expenses for fleets rise in proportion to the number of drivers you have.
Similarly, on iTunes or Spotify, the core commodity (music) has a fixed cost that rises in proportion to the size of your library. These costs increase proportionately to how extensive your library is, which depends on the number of labels and artists you sign.
The demand side for both these platforms does not add to marginal costs for them. Uber and Spotify can scale to hundreds of riders a day so long as they can manage their relationships with drivers or record labels. But the supply side has a fixed cost, which scales in proportion to how many offerings you have.
iTunes can lose out to Spotify if they don't have support for music in local languages in markets like India. Uber will quickly see users flocking to traditional forms of transport if they cannot provide drivers in meaningfully short wait times.
Web3 native products are unique because the teams behind platforms do not bear the marginal cost of liquidity. Initially, you take the cost through your native token given to users in an airdrop. Uniswap gave tokens. Aave gave tokens. These tokens, through their price and relative value in governance, incentivise users to continue being engaged in a product. But what happens if a platform's fees do not accrue to token holders? What incentive do users have to continue providing liquidity?
Projects like Aave and Uniswap are unique in that they could circumvent a critical level of liquidity to find PMF before interest in acquiring tokens waned. An AMM like Uniswap will always have users providing liquidity as long as
Token holders want a non-CEX avenue to trade, and
There is interest in receiving a passive yield on the asset.
Similarly, a platform like Aave will always have users providing liquidity if people want to put their idle crypto assets to work. The sustained rise (and eventual collapse) of DeFi in the quarters that followed Uniswap and Aave's launch helped them turn into products that saw enough volume to justify users putting their idle assets to work there.
Why does any of this matter? As token rewards wane, teams behind products like Uniswap must find new mechanisms to monetise themselves. Sure, there's the option of selling tokens, but it means giving up governance rights if you are truly decentralised, as you claimed when you issued the token.
Uniswap avoids switching on a protocol fee as it could mean liquidity and users flock towards zero-fee platforms. It also opens up their token to being interpreted as a security. OpenSea saw a meaningful threat from Blur when they launched a zero-fee model for NFTs.
At the crux of the interface dilemma is a simple question.
How do you monetise as a protocol when you don't own the core commodity your users flock towards you for?
Firms like Uniswap Labs (the entity that owns and develops the website at Uniswap) could generate revenue if they switched on the protocol fees. But they don't 'own' the protocol. The token holders do. You could argue that they could pass a proposal to switch the fees between the team and investors' tokens – but that would drive users towards their competitors.
So, in the interest of optics and relative market positioning, they go for what is now dubbed interface fees. Users who go to Uniswap from their native website will be paying 0.15% in fees each time they make a trade. Protocols are not new to the internet. SMTP has facilitated the transfer of emails for years. However, there are value-added services that private firms have built on this new protocol.
You can get a custom email hosted by Google for $15 monthly. An app like Superhuman can help you manage inbound emails for $30 a month. These are interfaces built on top of a free protocol (SMTP).