NFTs under MiCA: The Definitive Handbook

Non-Fungible Tokens (“NFTs”) have carved out a niche, offering a unique approach to the ownership and exchange of digital and tangible assets. Unlike crypto-assets such as Ether or ERC20 tokens, which are fungible and interchangeable, NFTs are distinct, each carrying certain data and attributes that potentially make them non-fungible and unique.

Usually, NFTs are embedded with metadata, anchored with their token smart contract (“SCs”) on a distributed ledger technology ("DLT”). The information within NFTs can range from digital artwork and music to virtual real estate and collectables, each potentially capturing authenticity and ownership details that are ideally verifiable and indisputable on-chain. They have allowed creators and collectors to offer a direct, secure, transparent medium to create, sell, buy, and trade digital assets.

This article unravels the technical aspects of NFTs and examines the regulatory implications of NFTs under the Markets in Crypto-Assets (“MiCA”) framework. The focus is on three token standards: Ethereum, ERC721, ERC1155, and the unofficial ERC404 token standard. By delving into the technical differences between these token standards, we highlight their similarities and differences. We discuss the technical implementation of three characteristics inherent to non-fungible assets for each token standard: their form, features and substance. These characteristics have a material impact on two features central to identifying NFTs: their uniqueness and fungibility. The legal argumentation established throughout this article builds on assessing these two features to determine which type of “unique and non-fungible" crypto-assets are regulated under MiCA.

1. What are “NFTs”?

NFTs have become popular as unique and non-interchangeable data that signify ownership of associated items, such as images, music, or videos. The token “ownership” is recorded and tracked on a distributed ledger. In this context, NFTs are comparable to digital certificates hosted on an immutable and permanent digital ledger. For this analysis, we consider NFTs as being made up of three parts:

NFT token (the form): The mapping of the balances of a token contract is the representation of the tokens themselves. Someone “owns” tokens when their balance in the contract is non-zero. In the standard NFT smart contract, ERC721, each token has a uint256 variable called tokenId, so for any smart contract (collection), the pair contract address, uint256 tokenId, must be unique. Furthermore, depending on how the metadata is stored, a string tokenURI may be included, which relates to the NFT metadata (discussed next).[1]
NFT metadata (the features): All the data associated with any NFT, which includes all the unique traits, NFT descriptions and the corresponding digital item, if applicable. The metadata may be stored on-chain or off-chain.
NFT rights (the substance): In some instances, the NFT may grant rights to their holders, such as membership rights or the right to listen to an audio track.

These concepts and their relationship are depicted in Figure 1 below.

Figure 1: NFTs Structure

1.1 The Form

Token standards on Ethereum are SC frameworks that define the rules and functions for tokens, ensuring compatibility and consistency across various projects and protocols. The Ethereum community has embraced several token standards, each with a different use case to fulfil multiple functions. These token standards are also instrumental in ensuring that smart contracts and decentralised applications remain composable.[2]

This section begins by outlining the most popular and essential NFT standard on Ethereum, ERC721, and comparing it to the fungible tokens on Ethereum, precisely the ERC20 standard. Building on this, we explore two token standards that blur the line between fungibility and non-fungibility, specifically the ERC1155 and ERC404 standards.

1.1.1 ERC721

ERC721 is a standard framework for delineating ownership of NFTs, characterised by their uniqueness. The ERC721 token standard is best understood compared to the ERC20 token standard, which defines a typical list of rules for fungible tokens on Ethereum. The ERC20 standard allows for creating tokens that are interchangeable with one another and do not possess unique attributes, making them ideal for cases where token uniformity and divisibility are necessary. Token ownership of ERC20 is recorded by directly mapping addresses to their respective balances, which tracks how many tokens each address owns. This mapping ensures that token transfers are accurately recorded and allows any network user to view the token account balance of a wallet address.

In contrast to the ERC20 standard, which employs a straightforward mapping to associate addresses with their holdings, ERC721 is distinguished by its incorporation of unique tokenIDs and tokenURIs. The ERC721 standard’s tokenIDs are a single non-fungible index, and the group of these technically non-fungible assets is deployed as a single contract with settings for the entire collection. This collection of digital assets is defined by its name and ticker, typically stored in variables within the token contract. NFTs may include additional metadata used to describe the characteristics and attributes of individual NFTs within the collection. This data is stored outside of the token contract (either on-chain or off-chain), and a link to this stored data is established through the tokenURI.

ERC20 smart contracts utilise a specific data structure to keep track of how many tokens each address owns. This data structure is known as a "mapping," which functions like a list that pairs addresses – a type of account identifier on a DLT – with the number of tokens they hold. When users want to know how many tokens a specific address owns, they call a function on the token contract known as the balanceOf function. In simple terms, calling a function can be thought of as sending a request to the token contract to perform a specific action. Here, the user provides the address they're interested in, and the function call prompts the token contract to retrieve and return the balance of tokens held by that address.

Figure 2 below provides a visual representation of this process. A user invokes the balanceOf function on the ERC20 SC by calling it and providing a wallet address. This transaction triggers a response from the token contract, and the function returns the token balance of the requested wallet address.

Figure 2: ERC20 balanceOf Function Call

ERC721 contracts utilise a similar but slightly different approach to maintain a record of asset ownership compared to ERC20 contracts. In ERC721 SCs, often used for unique digital items, there is a "mapping" known as owners. This mapping links each unique item, identified by a number called a tokenId, to the owner's address. For instance, if a user wants to know who owns the first item in a collection (with a tokenId of 0), they would call the ownerOf function. This function call utilises the owner mapping to look up and return the token owner's address with tokenId 0. Specifically, when the ownerOf function is called with a specific tokenId, it accesses the owner's mapping where each tokenId is associated with an owner's address. The function retrieves the address stored in this mapping for the given tokenId and returns it, effectively providing the ownership information for that particular token. Additionally, ERC721 SCs have another mapping similar to that found in ERC20 contracts, called balances. This mapping counts how many unique items a particular wallet holds from a specific collection. For example, if a user owns three different artworks from the same collection (each with a different tokenId), the balance mapping records a ‘3’ associated with their address. Like finding the owner of a specific token by calling the ownerOf function, another function known as balanceOf can be called by a user to see the number of unique tokens a specific wallet owns.

Figure 3 below displays the relationship between the balances and ownership mapping utilised by ERC721 tokens. The left-hand side of Figure 3 displays the mappings utilised when the ownerOf and balanceOf functions are called. The right-hand side shows the data returned when the function call is made.

Figure 3: ERC721 Function Mappings

The main difference between the ERC20 and ERC721 token contracts is the tokenId and how the ownership mapping is stored. However, many NFT collections utilising the ERC721 standard opt to employ an additional metadata extension that further distinguishes ERC721 tokens from ERC20 tokens.

This extension incorporates the tokenURI function, a unique feature of the ERC721 standard. The tokenURI function is integral to linking each token with its specific set of metadata. When this function is called with a particular tokenId, it returns a URL (“Uniform Resource Locator”) or a URI (“Uniform Resource Identifier”) that points to a location with that token's metadata. This metadata is a structured record that may contain details such as the NFT's name, description, image, and other attributes that provide context and characteristics about the specific. For instance, the code below shows the implementation of the tokenURI function for BoredApeYachtClub NFTs.

Figure 4: tokenURI code example [3]

To call the tokenURI function, the user provides the tokenId of interest. The function then determines the correct metadata URI: it first checks the existence of a token-specific URI; if absent, it combines the provided tokenId with a pre-established baseURI. The baseURI is a prefix common to all tokens, while the token-specific URI tailors the link to each NFT's distinct metadata. While the ERC721 standard specifies the existence of the tokenURI function, it does not define how or where the metadata should be stored or in what format it should be presented. The actual metadata can be sorted on-chain or off-chain depending on the implementation of the project. Common storage solutions include off-chain decentralised file systems like IPFS or on a DLT specialised for data storage, such as Arweave. The metadata can be stored in a JSON format, a lightweight and easy-to-use structure for storing and exchanging data.

After establishing the key differences between Ethereum's two primary token standards – ERC20 for fungible tokens and ERC721 for non-fungible tokens – we now delve into the complexities of contract designs that merge the features of both these contracts. Our focus shifts to the technical design of two notable standards: ERC1155, known for its multi-token capabilities, and the more recent, albeit unofficial, ERC404 standard.

1.1.2 ERC1155

Introduced in 2019, ERC1155 is a multi-token token standard on Ethereum. It represents a significant evolution from ERC20 and ERC721 contracts by enabling a single SC to manage multiple token types, including fungible and non-fungible tokens. The versatility of the ERC1155 token standard allows for more efficient transactions and reduces the gas cost for operations involving multiple token types.

ERC721 sets the foundation for NFTs by assigning a unique tokenId to each digital asset, ensuring every token is distinct and individually identifiable. This uniqueness allows for the creation and trading of individual digital assets. In contrast, the ERC20 framework operates on the principle of fungibility, where tokens do not require individual tokenIDs since each token is inherently identical to one another. Building on these principles, ERC1155 introduces a balanceOf function that caters to fungible and non-fungible tokens by utilising an additional tokenId field whereby the tokenId represents different tokens housed within the contract. For fungible tokens within the ERC1155 contract, tokens share the same tokenId, and the account balance reflects the quantity of each token, distinguished by its tokenId held by a wallet. ERC1155 also allows for the existence of NFTs; this is done by assigning a token to a unique tokenId, with the supply of that token set to one, ensuring that the tokenId individually and uniquely represents the token. This functionality underpins ERC1155's versatility, enabling a single contract to seamlessly manage an array of tokens, from fungible tokens akin to ERC20 to unique NFTs akin to ERC721, while maintaining distinct balances for each tokenId.

Because the state affecting all tokens is maintained within one contract, ERC1155 introduces additional functionality to transfer and create multiple tokens in one transaction. This is enabled through the safeBatchTransfer function, which allows for transferring multiple tokens simultaneously, and the mintBatch function, which facilitates minting multiple tokens in one transaction. This efficiency improves gas costs and has made ERC1155 a prevalent standard for gaming applications where the state affecting multiple tokens – perhaps representing different collectables, currencies or characters – needs to be maintained.

Projects utilising the ERC1155 framework can introduce additional functionality to their tokens by inheriting the ERC1155Metadata_URI contract extension within their primary ERC1155 implementation. This extension allows metadata to be added to tokens defined within the ERC1155 contract. Similar to ERC721 contracts, metadata for each token is defined through a URI that points to a JSON file containing the token's details, such as its name, description, image, and other properties. This JSON file adheres to a specified schema that ensures consistency and accessibility of the metadata, and the file can be stored either on-chain or off-chain, as discussed earlier. The linkage between the ERC1155 token and its metadata is achieved through the dynamic substitution of tokenIDs in the metadata URI. The extension specifies a function that returns a base URI containing a placeholder variable ID. When a request for token metadata is made, the contract replaces this placeholder with the tokenID, converting the base URI into a direct link retrieving the token metadata.

It's important to highlight that the ERC1155 contract design allows multiple tokens to share the same tokenId, leading to these tokens sharing a standard metadata URI and, as a result, the same metadata JSON file. This attribute renders these tokens non-unique since they are associated with the same metadata. For a token to be genuinely unique and non-fungible within an ERC1155 contract, it must be the sole token related to a tokenId, thereby exclusively linking to the metadata. This exclusivity ensures the token's uniqueness, as it is the only one represented by and connected to its specific metadata, distinguishing it from others in the contract.

1.1.3 ERC404

In February 2024, alongside multiple other token contracts on Ethereum, a new kid joined the block (pun intended). Known as ERC404, this unofficial and novel token standard aims to merge the fungibility of ERC20 tokens with the unique and non-fungible properties of ERC721 tokens in one contract – similar to what ERC1155 achieves. Although this may seem trivial initially, the ERC404 token standard is unique because it facilitates representative fractional ownership of NFTs. This feature has the potential to significantly improve the liquidity of NFTs as ERC404 tokens can be traded on decentralised exchanges (DEXs) such as Uniswap.

In the ERC404 standard, the contract is designed to manage both fungible and non-fungible tokens. The fungible tokens are divisible and can be owned in fractions, similar to ERC20 tokens, allowing for partial and flexible token transfer. The unique feature of the ERC404 standard is its conditional logic; when a wallet owns a certain number of these fungible tokens, the contract automatically mints an NFT (as an ERC721 standard token).

The architecture and features of the ERC404 token contract revolve around the minting and burning of the associated NFT according to the following rules:

If a wallet holds more than a pre-specified amount of the fungible token, the ERC404 contract automatically mints an NFT and sends it to the wallet.
If a wallet's number of fungible tokens falls below the pre-specified amount, the contract automatically burns the NFT.
Transferring the fungible tokens to another wallet will also transfer the pre-specified amount of NFTs.

These properties establish a link between the value of the fungible token and its associated NFT, effectively setting a "floor price" for the NFTs. This arrangement allows arbitrageurs to capitalise on any price discrepancies between the fungible token and the NFT by buying or selling the NFT in alternative markets whenever its price diverges from the valuation determined by the fungible token. Furthermore, the relationship between these two tokens allows traders to speculate on the price of NFTs without having to purchase the whole NFT, potentially leading to fairer and more efficient market pricing.

Let’s look at how the ERC404 contract establishes a link between its fungible tokens and its NFTs. The ERC404 documentation describes this relationship as a lossy encoding scheme where IDs and token amount data occupy the same shared space.[4] This shared space is represented through the contract’s handling of balances and IDs, particularly in how it manages the balanceOf mapping and the ownerOf mapping. Specifically, the balanceOf mapping tracks users' balances in a fractional representation typical for fungible tokens. Still, in the ERC404 framework, these balances also implicitly relate to the ownership of fractional parts of NFTs. On the other hand, the ownerOf mapping records the ownership of specific tokenIDs similar to the mapping in traditional ER721 tokens.

The relationship between fungible tokens and NFTs in the ERC404 contract is regulated by minting and burning processes, ensuring a balanced supply of each token type. Essentially, when a wallet holds above a certain number of fungible tokens, the ERC404 contract mints an NFT representing the accumulated value of those tokens. This NFT is then assigned to the wallet, with the assignment recorded in the DLT through an update to the ownerOf mapping, which links the unique tokenID to the wallet address. Conversely, if the wallet's fungible token count drops below the required threshold - perhaps due to a sale of the tokens - the corresponding NFT is burned. This results in updates to both the ownerOf and balanceOf mappings, removing the NFT's association with the wallet and adjusting the count of the remaining fungible tokens. Additionally, when an NFT is transferred to a different wallet, the equivalent number of fungible tokens moves with it. This transfer is documented by updating the ownerOf mapping to reflect the NFT's new owner and adjusting the balanceOf mapping to account for the latest distribution of fungible tokens.

Because the NFTs in ERC404s are ERC721 standard tokens, they follow the same design as typical ERC721. As explained earlier, the metadata and images pertaining to these tokens are accessed through the tokenURI, which can store the JSON file and the image on-chain or off-chain. For example, the first ERC404 project, Pandora, stores the metadata for its 10,000 NFTs on-chain with a URL pointing to the image hosted on GitHub. Another ERC404 project, DeFrogs, uses IPFS to ensure NFT metadata, such as JSON files and images. Table 1 below shows how the metadata for both projects is stored within each respective token contract for both projects.

Table 1: Metadata Storage ERC404 [5] [6]

While ERC404 may seem quite similar in functionality to ERC1155 contracts, a unique difference relates to the uniqueness of tokenIDs. Recall that in ERC1155, multiple tokens can share the same tokenId, representing fungibility between those tokens. Alternatively, a tokenID can uniquely be associated with one token only, representing a unique and non-fungible token. In this design, ERC1155 may include unique and non-fungible tokens, but this depends on its implementation and does not necessitate the existence of NFTs. ERC404 is different by design in that each fungible token in the contract represents a partial right to mint an NFT with a unique tokenID. In other words, NFTs in an ERC404 contract are necessary by design.

1.2 The Features

As mentioned above, there are currently two main approaches to storing the NFT metadata:

"Off-chain" refers to NFTs which deploy their smart contracts on-chain, though their metadata is hosted off-chain. The hosting can be either decentralised or centralised.
"On-chain" refers to NFTs which deploy the metadata on DLT: The metadata can be stored on a single-chain or cross-chain.

1.2.1 Off-Chain Metadata

Off-chain NFTs are the most common approach for metadata storage due to the DLTs' limited storage space, high network traffic, and, consequently, the high transaction fees associated with storing such data. For this reason, the data is typically stored on an external hosting service, such as a centralised server, a cloud storage or a distributed file system. Here is where tokenURIs [7] may be relevant. In NFTs, tokenURIs are just a simple API call that should return data such as the "name", "description"," image", and "attributes" of the digital asset stored in any of the mentioned hosting services.

Whenever an NFT is minted, the owner and metadata of each item may be queried, and the tokenURI provided should be resolved with JSON documents that may provide relevant data, such as the name, description, image, and other attributes.

Let us use the example provided by Cyfrin Updraft's course [8], where their tokenURI provides the following information when queried:

Figure 5: tokenURI example [9]

The metadata stored directly on the DLT is immutable, but data stored off-chain can be manipulated or become inaccessible. If any hosting service where specific metadata is stored is unavailable, the NFT metadata may disappear. Consequently, the ownership of the NFT form still exists, but the NFT features may no longer be available. This could potentially result in the NFT losing its market price.

1.2.2 On-Chain Metadata

Storing NFT metadata on-chain means that the data associated with the NFT, such as its attributes, image links, and other descriptive information, is recorded directly on a DLT. The main benefit of storing metadata on-chain is that the data is immutable and will exist perpetually as long as the DLT remains online. There are two forms of on-chain storage: the first pertains to NTFs that store metadata on the same DLT as the NFT token contract, and the second pertains to NFTs that store metadata on an external DLT (cross-chain storage).

In the first design, the NFT token contract (which defines the ownership and transferability of the NFT) and the metadata (which describes what the NFT represents) are stored on the same DLT. This design simplifies interactions, as smart contracts can directly reference the metadata without requiring cross-chain communication. Despite its simplicity, this method faces challenges in terms of scalability and the financial burden of storage, particularly on DLTs like Ethereum, where storing comprehensive metadata and images is relatively expensive. Nonetheless, certain NFT projects opt to keep data on the same DLT as the token contract, such as on Ethereum. A notable example is the OnChain Monkey NFT collection, which circumvents Ethereum’s high storage cost by using SVG formats to store images. SVGs, unlike the more common and bulkier bitmap formats such as JPEG or PNG, are text-based, enabling them to store images in a more space and cost-efficient manner. This method ensures that the metadata and image of OnChain Monkey's NFTs are fully hosted on Ethereum, ensuring their permanence and immutability while keeping storage costs manageable. Figure 4 below shows what the encoded OnChain Monkey NFT looks like.

Figure 6: OnChain Monkey #2012 Image SVG Code [10]

Figure 5 below displays the visual rendering of the same SVG code.

Figure 7: On-Chain Monkey #2012 Image Rendered [11]

1.2.2.1 Cross-Chain Metadata

Alternatively, the metadata for an NFT could be stored on a DLT different from the one hosting the NFT token contract. This approach is often used to reduce storage costs associated with hosting the metadata on more expensive DLTs such as Ethereum. Prominent projects in this space include Filecoin and Arweave.

Filecoin is a decentralised storage network that extends the Inter Planetary File System (IPFS) framework. IPFS is a peer-to-peer protocol for storing and sharing data in a distributed file system. Created by the same team behind IPFS, Filecoin builds a financial incentive layer on top of IPFS. It compensates network participants with Filecoin’s native token – FIL – for renting out their spare storage space, effectively creating an economy around data storage. Filecoin operates by allowing users to pay storage providers to hold their data. These storage providers then prove they correctly store the data over time through cryptographic proofs. This setup ensures data validation and retrieval, maintaining the integrity and availability of stored data.

However, it's crucial to note that data isn’t directly stored on the Filecoin DLT. Instead, it uses IPFS for data storage and distribution, with Filecoin providing a layer of economic incentives to ensure that data remains stored on IPFS nodes. Given this distinction, Arweave represents a more accurate example of an on-chain storage network. Unlike Filecoin, Arweave is designed for permanent data storage on its DLT, offering a different approach to on-chain data hosting.

Storage on Arweave is paid using its native token – AR. The cost of data storage is a one-time fee that covers the storage in perpetuity (a so-called storage endowment). This fee is calculated based on the size of the data being stored and the current price of AR. Once paid, the data is stored permanently, with no additional future costs or maintenance fees. Data can be uploaded and stored on the Arweave network through various web applications or APIs. Once uploaded, Arweave assigns a unique transaction ID to each piece of data, effectively creating a permanent and immutable link for the stored data. This transaction ID is also stored as a string in the tokenURI of the ERC721 token contract on Ethereum, creating a link between the token contract and the data hosted on Arweave. The content of this metadata, including the image, can be viewed by querying tokenURI within the token contract, allowing a representation of the data to be viewed on Ethereum-based applications such as OpenSea.

1.3 The Substance

Usually, owning an NFT refers to being identified as the owner of the relevant metadata and having the right to transfer the token to any person. This section will focus on the substance of the NFT. The substance can be seen as the rights the NFT may provide its holder. Therefore, we must address the following relevant questions for us to understand what the substance of the NFT is:

What rights does the form per se give to the NFT holder?
What are the rights related to the features embedded in the form of the NFT?
What other rights does the NFT give to its holder?

1.3.1 What rights does the form per se give to the NFT holder?

In most cases, the form provides the same inherent rights to most crypto-assets: the right to own[12] and transfer the token from the holder’s wallet to any other wallet. The transferability right, one of the requisites for a digital asset to be considered a crypto-asset, should not be deemed a relevant right for the “substance” assessment. The ownership of the token may be relevant depending on the answers to the subsequent questions. The reason is that the token may be comparable, in terms of its form, to a bearer financial instrument because the holder of the NFT is presumed to be the owner, and whoever is in its possession is entitled to any of the attached rights.

1.3.2 What are the rights related to the features embedded in the form of the NFT?

Now, we will ascertain if the features themselves may automatically provide intrinsic rights by being included on the form. Depending on its technical design, it may provide different answers to this question.

1.3.2.1 Off-Chain Metadata

The NFTs where the metadata is stored off-chain or cross-chain are the most complex. As described in the relevant section, these NFTs store their unique/relevant data on an off-chain hosting service. Most NFT projects rely on this solution and advertise it as comparable to a tangible asset; if you own the form, you have the right to use or benefit from its features or rights. As described in the relevant section above, the metadata can include different data: classically, it consists of a name, a description, a link to an image, the traits used in the image, and sometimes attributes when used for characters or items in a videogame. The challenge may arise when the only relevant data available in the form is a tokenURI pointing to where the data is hosted.

If the link (tokenURI) between the form (token contract) and the features (metadata) is broken, it would probably mean that the form does not have any attributes or relevant features that would potentially differentiate it from another, apart from the tokenID and the corresponding number on the subdirectory of the link to the hosting server to point to the correct hosted JSON file. In this scenario, the attributes or images that would provide value to the NFT are suddenly irrelevant, almost as if they were non-existent. Using a more traditional example, it could be comparable to a printed digital photo where the ink would have magically disappeared, leaving only an empty piece of paper. Consequently, the piece of paper is like any other and is no longer a form of photo representation. This means that, even if the data may still exist in a computer, the paper no longer represents such printed data. The same can be argued with off-chain metadata, where the NFT no longer points to the relevant features, and therefore, the form is nothing but a balance on the smart contract. The form would be irrelevant if the tokenURI is broken.

However, a distinction between the centralised and distributed off-chain solutions is required:

Firstly, the metadata stored on a distributed solution is immutable since it is uploaded on a distributed hosting service; the same may not apply to centralised service hosts where the issuer can easily modify the features. Therefore, there are no intrinsic or immutable rights in relation to the features of a centralised solution.
Secondly, the probability of the tokenURI not pointing to the relevant metadata is vastly different between both solutions: on the distributed solution, it would require that decentralised hosting service, such as IPFS, would go offline or that the content is not pinned or available is more unlikely; on the centralised solution, the risks are far more common and possible to occur: the centralised solution is unavailable, the issuer forgot to pay the subscription plan or does not have the server available, internet may temporarily go down on the local server, data can be tempered with, amongst other common risks.

Even if the distributed off-chain solution may share many features of a cross-chain solution, it can never truly guarantee that the data will be available. For example, IPFS nodes can choose what data they want to store and/or distribute, and the data may not always be kept online since the nodes may unpin the relevant content.[13]

Nevertheless, in both cases, there is a possibility that the rights related to the features embedded in the form may be merely attached to the form and may become unavailable; consequently, depending on how the risk assessment is made, we may consider cross-chain metadata as being comparable to either on-chain or off-chain metadata.

1.3.2.2 On-Chain Metadata

The NFTs where the metadata is stored a single-chain are more straightforward to understand from where their non-fungibility may arise. As described, the features are embedded in their form and do not rely on external access to data points to give the NFT holder the unique features such a crypto-asset may hold. Since the features and the form are inseparable, the assessment will focus on whether the features on each NFT of the same collection are unique to each NFT and if they are sufficient to render the NFT as non-fungible with the other NFTs of the same collection.

However, the cross-chain solutions may have risks that are comparable to distributed off-chain solutions. Even though the metadata is stored on a DLT where it should be immutable and available at all times, this solution still relies on an external access to provide the relevant features or rights to its holder. Which means that, technically speaking, there is a possibility that the rights related to the features embedded in the form may be merely attached to the form and may become unavailable. Nevertheless, this is the relevant risk. Data availability should not be a concern, and therefore, it is highly unlikely that the external access is lost.

1.3.3 What other rights does the NFT give to its holder?

The last question arises regarding any rights that are not related to the features of the NFT but are marketed on promises or described in contracts. Here, the mere promise of a future relevant right may be comparable to the right itself, assuming the issuer or any relevant party promises such a right. These rights may range from consumer rights, such as granting discounts on services or goods, to financial instrument rights, such as coupon payments.

It is also relevant to mention that the only relevant rights should be those offered/marketed by the offeror of the NFTs. Similar to any other instrument or asset, if an unrelated third party decides to distribute rewards or give special benefits to the holder of a certain NFT, such “attached” rights should not be within the scope of this assessment. It could lead to absurd cases in which a third party subverts its intended substance. For example, if Tether decided to distribute 1 USDT for every holder of 1 Ether per month, it would not render Ether as giving a right to receive 1 USDT per month.

For the purpose of this article, we shall not address the impact of the rights in the crypto-asset legal classification. We shall address whether such rights are non-fungible or unique across a collection of NFTs. In the next chapter, we shall discuss these two concepts.

1.4 Non-Fungibility and Uniqueness

Before we explore the nature of NFTs under the MiCA framework, we must understand two critical concepts: fungibility and uniqueness. This section will examine these concepts separately and propose a practical method for determining whether certain assets, which might not initially appear related to crypto-assets, can be considered non-fungible and unique. However, grasping the concepts before delving into further details of the NFTs is of fundamental importance.

1.4.1 What is Fungible?

The Oxford English Dictionary defines fungible as "an item that can be replaced by another identical item" or "product or commodity that has been contracted for: that can be replaced by another identical item without breaking the terms of the contract. More generally: interchangeable, replaceable."[14]

The Cambridge Dictionary defines interchangeability as "able to be exchanged with each other without making any difference or without being noticed."[15]

A non-fungible asset is an item that cannot be replaced by a similar copy or exemplar that would fulfil the terms of a contract or have no impact. In order to evaluate if a similar item would make no difference, we need to consider both its functionality and features perspectives. For instance, if I were to replace a certain football ball with a similar one of the same brand and type, it would be easy to argue that they are interchangeable. The replacement ball should have a similar composition and be an indistinguishable replacement for playing the football match. However, an original artwork is not interchangeable with any of its copies. While its functionality as a piece of art remains the same, its distinct features make it one-of-a-kind.

Fungibility is related to interchangeability. When an asset is interchangeable, there should be no noticeable difference in quality, which should not impact its value or utility. Therefore, we may assess fungibility based on the two criteria: characteristics and utility.

1.4.1.1 Characteristics

This criterion should be straightforward regarding its visible features, distinguishing it from any other (similar) asset and rendering it unique. These visible features may include the following:

Unique visuals, such as a painting on a physical structure or sculpture, are impossible to replicate due to the nature of the creation process. In conjunction with the painting, the artist's style and technique create a unique piece. Fake copies of such pieces do not make the original non-unique or fungible since such copies are also unique.[16]
Unique location, such as in the case of real estate where, regardless of existing unnumerable replicas of a house blueprint, their location is unique and not comparable to their replicas. For example, house replica one may be near the sea, house replica two may be at the top of a hill and house replica three is in the middle of a big city. Even if the replicas may be the same (e.g. pre-fabricated houses), their location will significantly influence them to the point of making them unique. For obvious reasons, this feature should be limited to non-movable assets.

One can also discuss certain “traits/composition” that may also be a relevant feature. This would apply, for example, to tuned cars, where the components of a fungible car are altered to increase/decrease its specifications. In this case, the alterations the vehicle may suffer may render the vehicle to have a unique composition. However, in this scenario, the composition may be replicable, even if extremely rare, and, as a consequence, it may render it fungible.

Replicability is a relevant concept to discuss, as it will significantly impact assessing the characteristics of digital assets. As described in this sub-section, either the visuals or the location are tangible vectors that need to be replicable. For example, can a digital photo be considered to have unique characteristics? Even if the photo may have metadata describing and providing information about the rights and administration of the image, it is still infinitely replicable by copy-pasting onto a different directory from the one initially stored. Furthermore, even if one decides to print the photo, it may appear to be unique since it is the only printed available photo. Still, due to the nature of the digital asset, it can be replicated by multiple printing. Therefore, the mere possibility of reprinting should be sufficient to render the printed copy fungible and non-unique.

Here, we have raised an important consideration: the original data (photo) is unique, but its representation (digital storage or printed version) is replicable and, therefore, fungible and non-unique. Nobody else can take the same photo, but everyone can have access to a replica of the photo. How should we address this two-layered composition?

The original data and the representation must be seen combined; it is impossible to separate both to understand whether they are unique. In the case of a handmade painting on a canvas, the uniqueness results from the painting on a canvas, which is not replicable. A digital photo is unique, but since it may be replicable infinitely, it renders it fungible and non-unique.

1.4.1.2 Utility

The utility criterion is relatively straightforward: What is the purpose for which a particular asset was created? In some cases, there may be no purpose or utility; in other words, an asset doesn't need a designated utility. However, in most cases, things are created with a purpose. For example, football balls are produced to be played, cars are produced to be used as a means of transport, and books are made to be read.

The utility may only be relevant when it has a specific purpose, such as the right to claim a certain lottery prize or the unique opportunity to meet a celebrity on a 1-to-1 dinner at a place of your choice. Otherwise, it may fall under an irrelevant redundancy: most cars are meant to be used as a means of transport, and therefore, most cars have the same utility. As described above, if two cars have the same characteristics, they will undoubtedly have the same utility.

1.4.2 What is Unique?

The Oxford English Dictionary defines uniqueness as "the only one of its kind”. This definition provides a solid definition that distinguishes rarity from other definitions. The Oxford English Dictionary defines rarity as “a person or thing that is unusual and is therefore often valuable or interesting”. Rarity can increase the value of an asset, but it doesn't necessarily make it unique. For instance, a newly issued limited edition of a new Marvel Comic Book, with only ten copies, can be scarce and potentially very valuable. However, each copy is not unique since it can be replaced with any of the other nine copies, as they are all identical. Or are they unique in their way?

Since there is no simple way to ascertain what uniqueness may encompass, I shall propose a uniqueness assessment based on two non-cumulative criteria: condition and history. Please be aware that the fungibility criterion is also relevant for assessing a unique asset; however, as proposed with this assessment, there may be fungible assets unique to certain individuals and under certain circumstances.

1.4.2.1 Condition

As mentioned, uniqueness should not be limited to the characteristics or utility it can provide to an asset's owner/holder; it may also relate to the unique value of such an asset. Value is a highly subjective variable that may differ for every person based on the perception or relevance of each individual. In this regard, there is a relevant factor in ascertaining the value of tangible assets that may render them unique even if the characteristics or utility are fungible, which is the asset's condition. Based on the example above, let's assume ten years have passed since the release of the limited edition of the Marvel Comic Book. Currently, collectors are searching for one of the copies available, but they only collect assets that are as “good as new” and are willing to pay a considerable premium for such a book. In this case, even if the intrinsic utility of the book (reading the story) is the same across the limited collection of ten books, its condition (from poor condition to “good as new”) may have a significant impact on its value, rendering each book unique with different prices based on its condition. Another example would be Trading Card Game (TCG) cards, where the range of prices for the same card (e.g. Charizard [1^st Edition]) vary from 1,500$ to 400,000$.[17]

1.4.2.2 History

In addition to the earlier criteria, another subjective yet significant criterion could determine an asset's value. This criterion is the history associated with the asset, which might impact its value even if it appears fungible and non-unique. Let's consider the example of the football ball. Suppose a collector is looking for the football ball Eder shot during the EURO 2016 final to secure Portugal's victory against France. In this case, the history of the football ball would make it more valuable to the collector, rendering it a unique asset. Curiously, in this scenario, the condition of the football would not be a relevant criterion in ascertaining the asset's value since its history supersedes it.

1.4.2.3 Proof of Validity

It can be argued that while the perceived value of an asset is affected by its condition and history, these concepts only affect the asset's value to the extent that they can be substantiated. In other words, sufficient and convincing evidence of the asset's condition and history is required for these factors to influence its perceived value. Of course, people may have different interpretations regarding the validity of such evidence, and this is a consequence of the subjectivity of the evaluation process.

For instance, consider an artwork's provenance, which is essentially its documented history tracking its movement from its creation through various owners up to the present. This proof of lineage can enrich a piece’s context significantly for art historians and curators. The chronicle of the artwork might highlight associations with collectors or its role within notable collections, elements that potentially enhance its perceived value.

The market price of the asset can be interpreted as the average or aggregate of all opinions, taking into account the validity of the evidence affecting the asset's condition and history. This collective valuation represents a consensus that reflects the varying degrees of trust and interpretation placed on the available information by different market participants.

These asymmetries in opinions arise due to disparities in the information available to various parties, and this is why certain institutions, such as rating agencies, exist. These institutions provide credible information on points such as the authenticity of a painting or the history of a car, helping to mitigate the impact of such asymmetries on the perceived value of assets.

In the current assessment, we assume that there is sufficient proof of validity allowing the market to determine the true condition and history of an asset. However, this concept should be kept in mind when reading the following examples.

2. Proposed Solution for Tangible Assets

Our proposed solution may render any tangible assets as one of the following: non-fungible and unique, fungible and unique, or fungible and non-unique. In some instances, the uniqueness of an asset, based on its history or condition, may have such an impact on the market price perception that it can be comparable to an intrinsic characteristic of the asset, ultimately rendering it as non-fungible. However, such cases must be assessed on a case-by-case basis since it is a grey area that may be highly subjective. In any case, we tried to provide a solution for assessing these assets. Please find the visual representation of the process below.

Figure 8: Tangible Asset Assessment

2.1 Analysing Certain Assets in Terms of Fungibility and Uniqueness

The proposed assessment of both concepts can lead to confusion and lack of clarity; however, this is merely a thought process exercise to deconstruct the concepts to reach any conclusions, so please bear with us. In this subsection, we will introduce a new criterion that may result from the combination of the previous criteria, which has already been mentioned: market price perception. This criterion goes hand in hand with the first relevant point mentioned in the MiCA preamble regarding unique and non-fungible crypto-assets: “(…) It is not readily interchangeable, and the relative value of one crypto-asset with another, each unique, cannot be ascertained compared to an existing market or equivalent asset.” and with rarity. Let us use the same practical examples to test our proposed method; we shall analyse a used car, a Charizard [1^st Edition] card with PSA 10[18], the EURO 2016 final football ball, a medallion inherited from my grandmother[19] and finally, a handmade painting.

2.1.1 A Used Car: BMW 5 Series (E60) 530d (218 Hp) 2003, 2004, 2005

For our first example, let's assess a black BMW 5 Series (E60) 530d (218 Hp) 2003 with the specifications described in the footnote below.[20] Considering the fungibility assessment based on the two criteria, its characteristics and utility are fungible. This car can be replaced since it was mass-produced during 2003-2005, and a suitable replacement with the exact specifications should be possible. Therefore, this car's characteristics and utility are not exclusive, and consequently, the vehicle may be fungible.

However, the argument can take an exciting turn when assessing the car's uniqueness; even if it can be replaced, its condition and history may render it unique. It should be impossible to have a vehicle with the exact kilometres, engine condition, scratches, windows replaced, or even which roads it travelled, which owners it had, or unique stories that unfolded. All these make the car unique... but how can it be unique and fungible? The reason is that uniqueness, in some instances, can be somewhat irrelevant to the fungibility assessment. If the market price perception categorises certain conditions or history under different price ranges, it creates sub-classes of the same car, and the value may vary according to the condition or certain historical aspects (e.g. number of previous owners). However, if there is a market price perception that allows certain cars to be comparable to other vehicles with the exact specifications in sufficiently similar conditions or history, then its relative value can be ascertained when compared to an existing market or equivalent asset. Therefore, the car is unique in its own right but should be deemed as a fungible asset.

2.1.2 A Charizard [1^st Edition] Card with PSA 10

Let's consider the example of a Charizard [1st Edition] card graded PSA 10. This example is somewhat similar to the car example discussed earlier but with a difference. While there are many Charizard [1st Edition] cards available, they have the same characteristics and utility. However, their condition has a significantly higher impact on market price perception than BMW's. So much so that entities like PSA give grading ratings to the card solely based on their condition. We have mentioned that the market price can go from 1,500$ to 400,000$ depending on the rating. For PSA grading, 10 is the highest attributable grading possible (a perfect exemplar) and is extremely rare.

However, this is the perfect example to prove that rarity is insufficient to consider an asset non-fungible. Similar to the car, new classes of market value perception are created based on the card's condition (or grading). As a result, the card may be extremely valuable based on its rarity, but it is still fungible among the same class of cards (e.g. PSA 10). Therefore, the card may appear unique due to its condition, but it is still fungible with other cards in the same condition.

2.1.3 The EURO 2016 Final Football Ball

For our third example, let’s assess the EURO 2016 football ball used in the final between Portugal and France. Once again, this example is similar to the car above but has an interesting distinction that may lead us to a grey area. There are many EURO 2016 official football balls that were used and sold and are readily available for any person who may want to acquire one. This means that this specific class of football ball is fungible since its characteristics and utility are replicated across its production, and, therefore, they are easily replaceable.

However, in this specific scenario, one of the unique criteria may substantially impact the market value perception to the point where it may render the asset unique: it was the ball kicked by Eder which ultimately led to Portugal’s victory over France in the final. The history is so unique and may have a considerable impact on the assessment of the ball. The condition is not relevant at all. Even if it appears that it can be replaceable, the fact is that the history is so relevant that it made that ball “one-of-a-kind”; there was a single goal during that match, and it is not replaceable by any means. Further to the above, prospective collectors and even regular people have a market value perception that the ball is not just a simple ball. The relative price may be rather hard to assess because there is no “existing market or equivalent asset.” Consequently, since history is so relevant, one may even argue that history becomes a relevant asset characteristic, making it non-fungible and unique. This logic applies to many collector items.

2.1.4 A Medallion Inherited from My Grandmother

For our fourth example, let’s assess a medallion inherited from my grandmother. This example merely provides perspective on the history criterion described above. The same logic applies to the football ball; many medallions have the same characteristics and utility as ours. Let us assume its condition is as “good as new”, and its history is magnificent and relevant to us: it was how my grandparents' love story began while living miserably during a civil war. Its history profoundly impacts me, and it is truly a unique piece that no other medallion can replace. However, how is its history perceived on the relative value of the market? While it may be true that history may give it an immeasurable value to me, it does not mean that the average person or collector would have a distinct market value perception from other medallions. Therefore, if I were going to sell the medallion on a secondary market, its relative value would be comparable to other medallions with the same characteristics and utility, and, therefore, there would be an “existing market or equivalent asset.”

These two examples illustrate a considerable challenge due to the subjectivity of a condition or history's impact on an asset to the point of influencing its characteristics. What impact is required for such a criterion to override the characteristics that make an asset fungible? This is highly debatable; unfortunately, we cannot answer straightly. However, the market value perception may seem an acceptable measurement to assess if something is so unique that it is no longer fungible.

2.1.5 A Handmade Painting

For our final example, let’s assess a handmade painting. This is the simplest example possible since its characteristics are unique and not replicable, immediately rendering it unique and non-fungible. In these cases, it is not even required to analyse the condition or history; it can safely be stated that it is non-fungible and, therefore, unique.

2.2 Summary

We have demonstrated that while some cases are simple and easy to analyse, others may be more subjective and may require further discussion. In any case, the focus of this article is to analyse NFTs under MiCA, and this section was meant to be an introduction to the topics of fungibility and uniqueness. Therefore, we shall not delve any deeper into tangible assets. For a better understanding of this section, please find our comparison table of the examples provided.

Table 2 - Proposed Solution for Tangible Assets Comparison Table

3. Unique and Non-Fungible Crypto-Assets (UNFCA) and MiCA

MiCA establishes a unified regulatory environment across the European Union to safeguard investors through enhanced transparency measures and by implementing a comprehensive legal structure for issuers and service providers, which includes compliance with anti-money laundering regulations. The newly introduced regulation defines crypto-assets, asset-referenced tokens, electronic-money tokens and utility tokens, with the latter being a sub-category of ‘general crypto-assets', regulated under Title II but otherwise undefined. They are distinguished from one another and subject to different requirements depending on the risk they entail. They are classified depending on whether the crypto-asset seeks to stabilise its value by reference to other assets, should it be an official currency, any other value or right or no reference, failing which they classify as a general crypto-asset (unless they qualify as an exempted instrument under MiCA’s Article 2(4)).[21]

MiCA defines a "crypto-asset" as "a digital representation of a value or a right which may be transferred and stored electronically, using distributed ledger technology (DLT) or similar technology". [22]

This extensive interpretation of a "crypto-asset" underscores the regulators' intent to capture as many digital assets within its regulatory purview as possible. It will likely lead to a significant expansion of the types of digital assets that fall within the scope of regulated entities under the EU's jurisdiction.

3.1 Transferability

Non-transferrable digital assets do not meet the criteria to be classified as crypto-assets. Hence, such digital assets, which are exclusively accepted by the issuer or offeror and technically non-transferable to other holders, should be excluded from MiCA's purview. The SC must have a technical restriction in place to prevent the execution of any transfer function. For the purpose of this article, we shall restrict the analysis to the standards per se without any restriction on transferability. You may find more details on the transferability feature here. Therefore, transferrable digital assets satisfy the condition of being “transferable”.

3.2 A Digital Representation of Value on a DLT

While "representation of value" may seem unclear, the preamble of MiCA provides further clarification. It states that a "representation of value" includes an "external, non-intrinsic value attributed to a crypto-asset by the parties concerned or market participants." This value is subjective and based only on the interest of the purchaser of the crypto-asset. Therefore, any asset, including an NFT, that is transferable and has any value between two parties may be deemed a “crypto-asset” under the scope of MiCA.

For the scope of this article, we assume NFTs are deemed to be classified as “crypto-assets”.

4. UNFCA under MiCA

In light of Article 2's number 3, "This regulation does not apply to crypto-assets that are unique and not fungible with other crypto-assets." However, once again, there is no detailed description of what "unique” and “not fungible" mean. We have tried to define what fungible and “unique” may mean in a general context, and we shall base our understanding on our suggested process and on the information provided in the preamble of MiCA, as described below:

The value is attributable to each crypto-asset's unique characteristics and utility. It is not readily interchangeable, and the relative value of one crypto-asset with another, each unique, cannot be ascertained compared to an existing market or equivalent asset.
Represent services or physical assets that are unique and non-fungible (e.g. real estate).
The assets or rights represented should also be unique and non-fungible.
Fractional parts of a UNFCA should not be considered unique and non-fungible.
Large series or collections should be viewed as an indicator of fungibility.
A unique identifier for a crypto-asset is not sufficient to classify a crypto-asset as UNFCA.
Certain NFTs may be qualified as financial instruments.
Apparent NFTs whose de facto features or whose features are linked to their de facto uses would make them either fungible or not unique. (Substance over form)

4.1 Proposed Assessment for NFTs | UNFCAs

The proposed solution has strong influences from the Proposed Solution for Tangible Assets but with some significant differences: the condition and history may have little real impact, and there is a much more relevant vector into play, which is the technical implementation. From here, it is essential for us to understand the NFT's form, features and substance and how they are intertwined. Is it inherent to the representation? Can the characteristics be easily replicated as a photo? What rights does the NFT give to its holder? What makes them truly unique and non-fungible under MiCA? Once again, we try to provide a solution for assessing the NFTs.

Figure 9: UNFCA Assessment

4.2 Analysing NFT as UNFCA

Prior to analysing various examples of NFTs in detail, we will ascertain all the above points described in MiCA and go through every single one.

We have already discussed what unique and non-fungible may be from a general perspective and how relative value may impact an asset's assessment. Additionally, MiCA is very clear in some cases based on the substance over form of features approach: NFTs that are comparable to financial instruments should be treated as such, and NFTs that may represent services or physical assets that are unique and non-fungible (e.g., real estate) should be treated as such. As a result, these NFTs are considered UNFCA under MiCA and, therefore, are outside of scope, but other relevant laws may be applicable.

A debatable topic is whether large series or collections should be considered an indicator of fungibility. As discussed above, rarity is not equal to non-fungibility, meaning a small or “large series” of the same asset should be fungible by default. It is fungible if more than one exists since it can be replaceable.

However, in a limited (or small) collection, one may argue that the asset is not “readily interchangeable,” but that does not mean it is not interchangeable at all. It can be an interesting point to the above arguments: how many copies must be available to make it “readily interchangeable”? Furthermore, should the relative value have an impact on the argument? In practice, exchanging a Charizard 1st Edition PSA 10 for another one may be easier than doing the same with a “5-million-dollar supercar”. This is comparable to opening a Pandora's box, which can create immeasurable difficulties in ascertaining what fungibility may be. The regulator may have created a vulnerability when mixing the size of the series or collections and the “readily interchangeable” together. Nevertheless, and in line with the arguments outlined in this article, this may not be relevant as they are mere indications. As such, if more than one copy exists, it should be deemed as fungible.

It is also vital to tackle the concept of collection. Even if, in most cases, collections have fungible collectables within the same collection (once again using the same example, there are numerous Charizard cards in the 1^st Edition collection. The remaining collectables of the collection are fungible as well since they were mass-minted), there may be specific scenarios where every collectable is a “one-of-a-kind”, meaning that each collectable is unique and non-fungible on its own. Therefore, it is essential to underscore that “should be considered an indicator”, but they do not automatically render a collection as fungible or not unique.

Finally, the other four indications are relevant for the NFT assessment since they are unique to the underlying technology that do not apply to other assets: a unique identifier to a crypto-asset is not sufficient to classify a crypto-asset as UNFCA, fractional parts of a UNFCA should not be considered unique and non-fungible, the assets or rights represented should also be unique and non-fungible, and NFTs whose de facto features or whose features are linked to their de facto uses would make them either fungible or not unique.

The topic of fungibility and uniqueness was already sufficiently complex to tangible assets, as discussed above, but the “NFTs” in general brought further discussion to the form (the digital representation on a smart contract) and the substance (the metadata and any potential rights associated with the NFT), which may blend with the characteristics and utility.

An exciting mention that may be relevant is that there is no such thing as a condition of an NFT; it does not deteriorate and may be maintained in perpetuity under the same conditions as it was minted.[23] Additionally, the history does not seem to have any relevant impact on assessing the fungibility of the NFTs.[24]

Therefore, if the condition and the history criteria may not have the same impact on the uniqueness assessment, what could be deemed a relevant replacement? I suggest the technical implementation of the NFT. The criteria to assess the fungibility of an NFT should be comparable to any other asset, but what can potentially make it unique may be related to its technical implementation.

The best approach to understanding the increased complexity is to provide and discuss use cases. We shall start with the simplest examples and move to the most complex ones, where serious questions may be raised.

4.2.1 ERC721: Unique Identifier: FULL SEND METACARD NFT

The first example to be assessed is the Full Send Metacard NFT.[25], and it is a great example for us to start with. According to the description of the NFTs, “the FULL SEND METACARD will give exclusive access to what FULL SEND does in the physical world and metaverse.” For simplicity‘s sake, I shall provide a screenshot of the relevant marketplace so that we can start building our argument.

Figure 10: FULL SEND METACARD Collection on OpenSea [26]

Based on our thought process in the previous section, it seems relatively clear that this NFT collection is fungible. As mentioned, the characteristics and utility are fungible across all the FULL SEND METACARDs since they provide the same rights to their holders, and the metadata are the same (as demonstrated in the example, the image/GIF is the same across the collection). Therefore, all the NFTs from this collection share the same metadata, and the asset or right represented is common.

Figure 11: FULL SEND METACARD NFT details on OpenSea [27]

In this example, the technical implementation is an ERC721, where the only differentiating factor is the tokenID amongst the different tokens of this collection. The metadata / tokenURI is the same for the entire collection. As described in the technical section, the tokenID is merely a different technical approach to storing token balance data, making it possible to identify the owner of a specific token by its tokenID. As a result, every single token in any ERC721 collection may be considered unique. Even if many tokens exist, they all have different numbers identifying them.

However, here is where the MiCA preamble may assist in the assessment of this NFT:

The assets or rights represented should also be unique and non-fungible: As we have seen, the metadata (the characteristics) and the right to exclusive access to FULL SEND (the utility) are fungible. They are fungible within the 10,000 NFTs minted by the relevant smart contract.
Further to the above, a unique identifier for a crypto-asset is not sufficient to classify it as UNFCA. This means that even if technically the unique identifier (tokenId) may be unique, substance over form takes precedence, and it should not be a relevant metric to ascertain its fungibility. Even if I lose access to my NFT with a unique tokenId and it is lost forever, it may still be replaceable by any other NFT from the same collection with a different unique Therefore, it is still fungible.

This leads us to our first conclusion: When the characteristics (the metadata) and the utility (the rights) are fungible and not unique, the technical implementation (the form) may be irrelevant if it solely relies on the unique identifier (tokenId) basis to support the non-fungibility of the token. To conclude, the FULL SEND METACARD NFT may be an NFT from a technical point of view, but it shall not be deemed a UNFCA under MiCA, and therefore, MiCA would apply to such crypto-asset.

4.2.2 ERC721: Unique Asset: A 1/1 Collection

The second example to be assessed is an NFT collection limited to a single mint. Even though I did not find any real example for this use case, it is still worth mentioning. Similar to the “handmade painting” example on the tangible assets assessment, since only an NFT exists, it would be relatively easy to argue that its characteristics, utility, or technical implementation would render the NFT non-fungible and unique, right?

However, the intrinsic nature of digital assets may present a new challenge. For this exercise, let us assume that the 1/1 NFT Collection issuer has used a cross-chain or off-chain solution to store the relevant metadata, with the image or relevant metadata being stored as an image and JSON file on the issuer's computer, which is normally the case when using such solutions. Imagine that one day, after the issuer realised the 1/1 NFT was a huge success, it decided to issue a new NFT collection with the same metadata. Suddenly, a new crypto-asset with the same characteristics or utility is circulating, potentially even with the same tokenId.

Firstly, we shall discuss how the unique identifier may not be restricted solely to the tokenId. In this case, if we merely focus on the tokenId, we would be bound to sustain that the newly minted crypto-asset was the same as the first one. However, we know for a fact that they are two different crypto-assets. Therefore, If the tokenId is the same (it should be 0 in both mints), what makes them different? The answer is the SC address. When a new SC is deployed, a new address is generated, even if the relevant data of the NFT is precisely the same.[28]

Figure 12: Smart Contract Address example on OpenSea

When the characteristics (the metadata) and the utility (the rights) are fungible and not unique, the technical implementation (the form) may be irrelevant if it solely relies on the unique identifier (address) basis to support the non-fungibility of the crypto-asset.

Secondly, we shall discuss how this approach may trigger particular concerns. Suppose the metadata or its characteristics are replicable, such as a digital photo in the example given in the Characteristics sub-section. The technical implementation (form) can be deemed entirely irrelevant in that case. This can result in an extreme conclusion that any NFT without on-chain metadata is replicable and, therefore, would not be fungible.

Similar to the EURO 2016 final football ball and the medallion, the characteristics (metadata) and the technical implementation (form) combined may substantially impact the market price perception to the point where they may render the asset unique. This means that prospective collectors and even regular investors have a market price perception that what may have value is the combination of characteristics with the technical implementation. However, the relative price may be relatively easy to assess since an “equivalent asset” already exists. Even though the market price perception may still be a relevant factor for the assessment of NFTs as UNFCAs, it would not apply to this scenario if the relative price is comparable.

Now, let us imagine that it was not the issuer who issued a replica but another person who “copy-pasted” the metadata or even uploaded it (by first copying the JSON file data, downloading the image file, and re-uploading it). In this case, the original SC address would render the minted crypto-asset unique in its own right. The SC address can be comparable to the technical implementation of a seal of authenticity, thus making it the original NFT. So, in theory, the non-official replicas could be deemed unique and non-fungible on their own since the market price perception would see them as merely an unofficial/irrelevant NFT when compared with the original piece. Similar to what happens with handmade paintings/art, but it is still not the same.

Therefore, it is still being determined whether the mere ability to replicate the metadata should be sufficient to disregard the NFT as non-fungible and unique. However, if there is an NFT replica minted from the same issuer, such NFTs could be unique from a technical point of view, but they shall not be deemed to be a UNFCA under MiCA, and therefore, MiCA would apply to such crypto-asset.

4.2.3 ERC721: On-chain Single-Chain Metadata: CryptoPunks

The third example to be assessed is the CryptoPunks collection.[29] CryptoPunks is a pioneering collection in the NFT space that stores the images for its 10,000 NFTs as SVG images, benefiting from the relatively small size of each image at 24x24 pixels. However, it should be noted that on-chain metadata doesn't solely relate to image storage. Other characteristics, depending on the NFT design, are stored directly within the DLT as metadata.

In the case of CryptoPunks, the CryptoPunkData contract holds crucial information about each punk. This contract can be queried to retrieve on-chain metadata. For example, if we query the punkAttributes function within the contract, passing the integer 0 to receive the metadata for the first tokenID, the function returns a string indicating the characteristics of that particular punk. The response looks like this: "Female 2, Earring, Blonde Bob, Green Eye Shadow".

These attributes are stored within the CryptoPunkData contract as encoded identifiers in a byte array mapped to each punk's index. The contract uses mappings, such as assetNames, to convert these identifiers into human-readable strings that describe each punk's unique features. When the punkAttributes function is invoked with a specific punk index, it deciphers these identifiers by accessing the corresponding entries in the assetNames mapping. This process compiles and returns a comprehensive, comma-separated string that encapsulates the full spectrum of a punk's attributes.

Figure 12 below shows the query and response when calling punkAttributes for the first CryptoPunk at index 0.

Figure 13: punkAttributes [30]

These NFTs are the prime example of the convergence of the form, features and substance under a single asset, which can render it comparable to a tangible handmade painting collection by an artist: all of them different amongst themselves and without the possibility of replicating them, making them truly unique and non-fungible. Each tokenID has unique on-chain punkAttributes; therefore, from the SC address and tokenID, one can query the unique features associated with the corresponding NFT. There is no tokenURI or dependency on a third-party hosting service to render the metadata. The features are the form, and its substance is inherent to it.

As a result, the characteristics and technical implementation of the Crypto Punks NFTs render them unique. Partially similar to the example of the handmade painting, when the characteristics are unique, and the crypto-asset does not provide any rights deemed comparable to financial instruments rights, any other potential utility is irrelevant for the assessment of such crypto-asset. Therefore, considering our assessment, these NFTs shall be deemed to be UNFCA under MiCA; thus, MiCA would not apply to such crypto-assets.

4.2.4 ERC721: Off-chain Distributed Metadata: Bored Ape Yacht Club

The fourth example to be assessed is the Bored Ape Yacht Club (BAYC) collection. The BAYC collection consists of 10,000 NFTs, each providing its holders access to members-only benefits, the first of which is access to THE BATHROOM, a collaborative graffiti board.[31] Unlike the previous example, the metadata for BAYC NFTs is stored off-chain on the distributed IPFS file-sharing network.

As discussed previously, the structure of BAYC NFTs utilises a baseURI, which points to a directory containing the JSON files defining the metadata for all BAYC NFTs. The metadata for each NFT is stored in a sub-directory defined by its tokenId. Figure 13 below shows the query and response for the first BAYC NFT at index 0.

Figure 14: tokenURI [32]

The response to the query returns a string representing the IPFS link to the first BAYC NFT. Figure 14 also shows the JSON file with the metadata hosted within this IPFS sub-directory.

Figure 15: JSON file [33]

The IPFS string relating to the first field, 'image', provides another IPFS link hosting the image associated with the BAYC NFT at index 0. Figure 15 below shows the image stored at this location.

Figure 16: A BAYC image metadata [34]

From our assessment perspective, each NFT from the BAYC has unique characteristics and no relevant utility.[35] From a technical perspective, we are required to discuss two main points:

Firstly, we are bringing the unique identifier for discussion again for a different reason: we have ascertained that the unique identifier can be the tokenID and in some extreme use cases, the smart contract address (as explained on the A 1/1 Collection example); however, it may be important to understand the impact of the tokenID on a cross-chain metadata tokenURI.
Secondly, if cross-chain metadata implementation should be deemed as irreplicable, immutable and readily available;

On the first point, and as described above, the collection shares a constant baseURI and a variable tokenID to ascertain the tokenURI. One can argue that the features embedded in the form are merely dependent on the unique identifier; thus, this can lead to a concerning approach where, according to the MiCA preamble, “a unique identifier for a crypto-asset is not sufficient to classify a crypto-asset as UNFCA.” However, we do believe this approach should be limited to the scenarios comparable to the FULL SEND METALCARD collection, where the characteristics and utility are fungible and non-unique and the only difference between the NFTs is the tokenID. As a consequence, the tokenID is not sufficient to make them non-fungible. In the BAYC case, the unique characteristics may be linked to the tokenID, but the metadata provided via the tokenURI is different between all NFTs, potentially making each one of them unique and non-fungible. Nevertheless, this argument may be influenced by the second point as described below.

On the second point, we understand that the relevant attributes are not embedded in the form. Once again, the BAYC has a tokenURI which points towards an offline distributed storage provider (IPFS) where the metadata is kept. As previously discussed in “The Substance” sub-section, the main point of concern is related to the possibility that the rights related to the features embedded in the form may be merely attached to the form and may become unavailable. Consequently, it would render the NFT itself irrelevant.

The decentralised off-chain raises similar concerns to cross-chain solutions. The correct approach to assess these types of NFTs purely based on the metadata availability is highly debatable. Depending on the risk framework of the data availability on both solutions, such NFTs can be deemed as UNFCAs or non-UNFCAs.

4.2.5 ERC721: “Fractionalised” NFTs: Particle | Banksy Love is in the Air

The fifth example to be assessed is the Particle | Banksy Love Is In the Air collection. This collection is owned by Particle, which acquired the piece at auction from Sotheby's on 5/13/2021. Subsequently, the work was donated to the non-profit Particle Foundation, where it will remain in perpetuity. According to the Particles NFTs description,[36] NFTs are digital Collector Cards representing ownership of physical fine art masterpieces. They allow you to decide what happens to the art, and owning more NFTs gives you more control over it.

Figure 17: Particle / Banksy Love is in the Air Collection on OpenSea [37]

This NFT collection is similar to some of the other examples from a technical implementation; each NFT has a unique tokenID and a corresponding tokenURI pointing to an off-chain metadata hosting service. However, new challenges arise from the characteristics and utility provided to the NFT holder.

Once again, the MiCA preamble may assist in the assessment of this NFT on four different points

Represent services or physical assets that are unique and non-fungible (e.g. real estate).
The assets or rights represented should also be unique and non-fungible.
Fraction parts of a UNFCA should not be considered unique and non-fungible.
Apparent NFTs whose de facto features or whose features are linked to their de facto uses would make them either fungible or not unique. (Substance over form)

We have ascertained that the Particle NFTs represent a unique and non-fungible physical asset. The Love is In the Air, created by Banksy, is a “physical fine art masterpiece”, and therefore, it shall fall under the category of having unique characteristics that render it unique and non-fungible. Consequently, the collection may seem to represent a physical asset.

However, the second point mentioned may be broader and makes us question the first one: any asset (it does not require being physical) or rights should be unique and non-fungible. If we do a more detailed analysis, it becomes increasingly complex to argue that the rights granted to these NFT holders are unique and non-fungible. We have ascertained that the NFT provides homogeneous and proportional rights to every NFT holder. In theory, you own a “small square” that represents a % decision power over the management of the physical asset. Suddenly, the first point does not seem so straightforward: what may represent the physical asset is not an NFT but rather an entire collection. Additionally, even if a holder owns the whole collection of the NFTs, it only allows you to “make decisions.” Still, the non-profit Particle Foundation will hold the art in perpetuity. Therefore, the NFT does not represent any physical asset but rather the right to “make (limited) decisions” in relation to the underlying asset.

Before moving on to the final assessment of the NFT, it is also relevant to mention the third point: fraction parts of a UNFCA should not be considered unique and non-fungible. This is not applicable to this example since the fraction parts do not pertain to any UNFCA but rather to a physical asset that does not have any UNFCA representing it.

Figure 18: Fractionalised scheme

As the diagram above demonstrates, the MiCA preamble's point would only apply in the example provided on the left. However, in this case, as described in the example on the right, there is no UNFCA representing the asset; therefore, it is not applicable.

We did a comprehensive analysis of this collection. However, we still need to address one of the main points that may lead to further discussion: the features of each NFT that may render the characteristics unique and, therefore, non-fungible. As you may have noticed in Figure 16, every NFT points towards a different image and video file that pinpoints what part of the physical art you “own.” Therefore, every single NFT of the collection has visibly different features, so, in theory, they are unique and should be non-fungible.

However, the metadata is stored on a centralised off-chain hosting service, which means the risk of losing access to the metadata is higher than in a decentralised off-chain hosting solution, rendering its features potentially non-relevant.

Figure 19: Particle / Banksy Love is in the Air tokenURI [38]

In this case, even if the metadata is inaccessible, the technical implementation (form) of the NFT would still have relevance due to the utility (substance) associated with the NFTs of the collection: the right to “make (limited) decisions” in relation to the underlying asset. This leads us to the fourth point: Apparent NFTs whose de facto features or whose features are linked to their de facto uses would make them either fungible or not unique. This means that when the characteristics (features) and technical implementation (form) are unique, their utility (substance) may still make them fungible, in line with the regulator’s substance-over-form approach.

Further to the above, and to consolidate the relevance of the substance, the market price perception proves that the NFTs may be fungible to the point where two different NFTs have the same price. The prices of other NFTs from the same collection may differ, but their relative value can be ascertained when compared to an existing market or equivalent asset.

As a result, the characteristics and technical implementation of the Particle | Banksy Love is in the Air NFTs render them unique; however, their utility is homogenous and fungible. Therefore, considering the substance-over-form approach, these NFTs shall not be deemed a UNFCA under MiCA, and thus, MiCA would apply to such a crypto-asset.

4.2.6 ERC1155: A Fungible NFT

In the sixth example, we will assess a token standard that mixes the concepts of ERC20 and ERC721 and enables the issue of multiple different tokens under a single, smart contract. For this article, the only relevant feature that the standard allows is to create a “class” of NFT based on the metadata/token URI of the ERC721 and make it fungible as an ERC20. This concept has its use cases (e.g. the FULL SEND METACARD should issue the NFT as this ERC1155), especially for video games or proof-of-participation (POPs) NFTs. The metadata is embedded in the tokens, but the tokens do not have any token ID. They all point to the same metadata without transferring any individually. [39]

This example is relatively straightforward, based on 1) the arguments provided in the FULL SEND METACARD analysis and 2) the fact that making them fungible from a technical implementation perspective strips them of any potential uniqueness since they are all replicas and merely a “balance.” As a result, these “hybrid” solutions should not be deemed UNFCA under MiCA, and therefore, MiCA would apply to such a crypto-asset.

Interestingly, this solution may apply to the “tokenisation of real-world assets” (RWA), where the RWA data is represented with the ERC721 standard. Still, it is “fractionalised” based on the ERC20 balance implementation. The ownership of a particular real estate may be tokenised and divided amongst 100 different tokens, all representing 1/100 of the ownership of the real estate.[40] In this case, the ERC20s are fungible by default, and no further assessment is required for non-fungibility purposes.

Nevertheless, it is essential to underscore that the ERC1155 allows for multiple crypto-assets. Even though it provides an optimal solution for mixing features of ERC721 and ERC20, it may still allow the issuing of vanilla NFTs and fungible crypto-assets. As in the example provided in OpenZeppelin[41] (below) Gold is a fungible token (even if “hybrid"), while Thor’s Hammer is an NFT as only one is minted.

Figure 20: OpenZeppelin ERC1155 code example [42]

Therefore, every crypto-asset minted under the ERC1155 standard may require a case-by-case analysis to assess the influence of its characteristics, features, and technical implementation on its fungibility and uniqueness.

4.2.7 ERC404: A New Concept of Apparent Fractions of NFT: PANDORA

The seventh example to be assessed is the Pandora NFT collection. As described above, the standard may seem similar to ERC1155 initially; however, it may have a different approach to issuing crypto-assets. Although this may seem trivial at first glance, the ERC404 token standard is unique in facilitating representative fractional ownership of an NFT. The ERC404 token is similar to an ERC20 that may be divisible, and when held as a unit, it may give the token holder the right to mint a random NFT available on the smart contract mapping. The ability or the right provided to mint an NFT is restricted and limited to the ERC404 token per se.

While the ERC1155 would define an NFT and make it fungible by mixing the ERC20 standard, the ERC404 does precisely the opposite. When a holder owns a unit of the ERC20, it gives you the right to mint an NFT. However, the assessment should be similar in both cases. The reason is that even though it appears that the ERC404 token may represent a “fraction” of an NFT, it may not be the truth. A fraction of a unit of the ERC404 token may represent a “fraction” of the right to mint and hold a random NFT. Therefore, the assessment of the tokens should be comparable to regular ERC20 and not fall under the same category as the ERC721 Fractionalised NFTs: Particle | Banksy Love is in the Air described above.

Regarding the NFT, depending on the technical implementation, the assessment should be similar to the other examples provided in this section, as it should follow similar patterns.

Therefore, the ERC404 tokens should be classified as fungible crypto-assets, and the NFTs may depend on their interpretation, as discussed above.

5. Conclusion

This article aimed to set up the foundations for the reader to understand what NFTs are, create discussion on the concepts of fungibility and uniqueness, how such concepts may be different between tangible and digital assets and finally provide a framework to assess NFTs under MiCA.

The framework takes into consideration the different components of the NFTs, such as the form, features and substance and how they can potentially have an impact on the characteristics, utility and technical implementation of the NFT to classify them UNFCAs or non-UNFCAs.

However, even with the framework provided, it is difficult to provide a one-size-fits-all answer generally applicable to NFT collections which have their unique characteristics and uniqueness based on cross-chain or distributed off-chain solutions: depending on the risk assessment of the relevant metadata availability of the NFT, a considerable amount of collections may be at risk of not being considered UNFCAs under MiCA. Further guidelines on this framework are ideally published by the competent authorities to provide clarity to the NFT market, especially since if such are deemed to qualify as crypto-assets falling within MiCA’s scope, then certain provisions such as the market abuse ones can be applicable in a manner which juxtaposes the realities of NFT trading today.

---

[1] Source: https://docs.openzeppelin.com/contracts/5.x/tokens

[2] Source: https://ethereum.org/en/developers/docs/standards/

[3] Source: https://etherscan.io/address/0xbc4ca0eda7647a8ab7c2061c2e118a18a936f13d#code#L1480

[4] Source: https://github.com/Pandora-Labs-Org/erc404-legacy

[5] Pandora tokenURI source: https://etherscan.io/token/0x9e9fbde7c7a83c43913bddc8779158f1368f0413#readContract#F13 (Query function 13, TokenURI, introducing ‘3’ as the ID)

[6] DeFrogs tokenURI source: https://etherscan.io/token/0xd555498a524612c67f286df0e0a9a64a73a7cdc7#readContract#F17 (Query function 17, TokenURI, introducing ‘2’ as the ID)

[7] A formal definition is as follows: “A URI (Uniform Resource Identifier) is a string containing characters that identify a physical or logical resource. URI follows syntax rules to ensure uniformity. Moreover, it also maintains extensibility via a hierarchical naming scheme.”

[8] Example provided by the Cyfrin Updraft course, Advanced Foundry, Section 1: Develop an NFTs Collection: https://updraft.cyfrin.io/courses/advanced-foundry/how-to-create-an-NFT-collection/upload-data-on-IPFS?lesson_format=video

[9] Example provided by the Cyfrin Updraft course, Advanced Foundry, Section 1: Develop an NFTs Collection: https://updraft.cyfrin.io/courses/advanced-foundry/how-to-create-an-NFT-collection/upload-data-on-IPFS?lesson_format=video

[10] Source: https://github.com/metagood/OnChainMonkeyData/blob/main/svgs/2/2012.svg?short_path=42f3d28

[11] Source: https://github.com/metagood/OnChainMonkeyData/blob/main/svgs/2/2012.svg

[12] The discussion between ownership and possession is outside the scope of this article.

[13] The data availability risks and concerns are outside of the scope of this article and should be provided in a subsequent article.

[14] Oxford English Dictionary, s.v. “fungible, adj. & n.”, July 2023. https://doi.org/10.1093/OED/6757729486

[15] https://dictionary.cambridge.org/dictionary/english/interchangeable

[16] For this article, we shall not discuss how perfect copies are comparable to the original work to the point of misleading experts when evaluating the pieces. Therefore, they may appear “fungible” since no one would notice the difference.

[17] PSA Card Auctions website: https://www.psacard.com/auctionprices/tcg-cards/1999-pokemon-game/charizard-holo/values/544028#

[18] PSA is an authentication and grading company for trading cards and memorabilia. PSA 10 is the highest quality grading possible based on PSA standards, as described here:: https://www.psacard.com/gradingstandards

[19] This example was not used before.

[20] https://www.auto-data.net/en/bmw-5-series-e60-530d-218hp-9601

[21] Digital Finance: Council adopts new rules on markets in crypto-assets (MiCA), https://www.consilium.europa.eu/en/press/press-releases/2023/05/16/digital-finance-council-adopts-new-rules-on-markets-in-crypto-assets-mica/#:~:text=The%20EU%20brings%20crypto%2Dassets,crypto%2Dassets%20(MiCA)

[22] MiCA Regulation, Article 3, number 1(5)

[23] For the purpose of this article, we shall not discuss dynamic NFTs where the condition may be a relevant variable. For more details about dynamic NFTs: https://chain.link/education-hub/what-is-dynamic-nft

[24] For example, an NFT owned by a celebrity has no increased value when comparable to other NFTs of the same collection: https://www.marca.com/en/lifestyle/celebrities/2023/07/04/64a37b8fca47414e198b45db.html

[25] Source: https://opensea.io/collection/metacard-by-fullsend

[26] Source: https://opensea.io/collection/metacard-by-fullsend

[27] Source: https://opensea.io/assets/ethereum/0x7ecb204fed7e386386cab46a1fcb823ec5067ad5/9130

[28] Collusion of addresses is practically impossible.

[29] The Crypto Punks collection does not follow any token standard. However, due to its significant importance in the NFT market, it is worth the mention.

[30] Source: https://etherscan.io/address/0x16f5a35647d6f03d5d3da7b35409d65ba03af3b2#readContract#F1 (Query function 1, punkAttributes, introducing ‘0’ as the index)

[31] This assessment is based on the description of the project during its minting phase. New rights/utility may have emerged over time and are out of the scope of this assessment.

[32] Source: https://etherscan.io/address/0xbc4ca0eda7647a8ab7c2061c2e118a18a936f13d#readContract#F20 (Query function 20, tokenURI, introducing ‘0’ as the tokenID)

[33] Source: https://alchemy.mypinata.cloud/ipfs/QmeSjSinHpPnmXmspMjwiXyN6zS4E9zccariGR3jxcaWtq/0

[34] Source: https://opensea.io/assets/ethereum/0xbc4ca0eda7647a8ab7c2061c2e118a18a936f13d/0

[35] This is further sustained in point 67 of ESMA’s Consultation paper on the draft Guidelines on the conditions and criteria for the qualification of crypto-assets as financial instruments, where “In assessing the uniqueness and non-fungibility of a crypto-asset, such crypto-asset may be considered as unique and not fungible if its characteristics and/or the rights it provides distinguish it from the other tokens issued by the same (and any other) issuer.” (our emphasis). This must be read in parallel with point 72, where “(…) the utility function of NFTs can also play a role. In some cases, (…) the specific attributes of the NFT become less relevant compared to the utility it provides, making different NFTs functionally interchangeable for practical purposes.” In the example provided, BAYC only gives access to THE BATHROOM; which is a collaborative graffiti board. Therefore, this utility may be deemed irrelevant.

[36] Particle NFT description: https://opensea.io/assets/avalanche/0xf675a87397a6239eaf95ad948670a5b19d076c59/9908

[37] Source: https://opensea.io/collection/particle-banksy-avax

[38] Source: https://snowtrace.io/token/0xF675a87397a6239eAF95Ad948670a5b19D076c59/contract/readContract?chainid=43114#F20 (Query function 20, tokenURI, introducing ‘1’ as the tokenID)

[39] A more technical explanation may be found here: https://www.youtube.com/watch?v=mM77Ta-g7Hs

[40] It’s out of the scope of this article to discuss the merits of RWA or tokenisation in general. It was merely given as an example.

[41] Source: https://docs.openzeppelin.com/contracts/3.x/erc1155

[42] Source: https://docs.openzeppelin.com/contracts/3.x/erc1155#constructing_an_erc1155_token_contract

Topic

Crypto regulation MiCA