Detailed Analysis of the Current Status and Solutions for NFT Storage

Source: Memo Labs Blog Compiled by: Hu Tao, Chain Catcher

Apart from currency, most objects in the real world are unique, with distinct characteristics and values. The features of NFTs can perfectly adapt to this situation, enabling the tokenization of different objects in the digital world. NFTs leverage the tamper-proof, publicly transparent, and traceable characteristics of blockchain technology to map the real world into the digital realm, constructing the long-awaited utopia—the metaverse.

However, due to existing technological limitations, NFTs have not yet been able to fully realize the aforementioned ideas. Due to communication congestion, limited scalability, high gas fees, and other reasons, the metadata and media data of NFTs are not fully stored on the blockchain but rather off-chain. Without the protection of blockchain technology, this part of the data is not completely secure and reliable, meaning that current NFTs do not yet achieve the same level of trustworthiness as Bitcoin. However, this flaw is often overshadowed by the enthusiasm for NFTs and receives little attention.

Although NFT transactions are also completed on the blockchain, their storage is typically done off-chain—on centralized servers of projects, third-party cloud servers, IPFS, or other NFT storage projects. This is because the metadata involved is often complex, and media data requires larger storage space.

NFT storage has become the weakest link in the current NFT systems, but as a key infrastructure of the metaverse, it is bound to play an important role in the new social ecology in the future. This article will explain the underlying architecture of NFTs, summarize the current state of NFT storage, and delve into the opportunities and challenges faced by NFT storage.

1. Current State of NFT Storage

1.1 Product Types

With its unique attributes, NFTs have brought about a certain degree of change in various fields such as the metaverse, digital art, collectibles, gaming, DeFi, public utilities, and sports. This article summarizes the highest market cap NFT products in each category as the target group for studying the current state of NFT storage.

1) Decentraland

Decentraland is a virtual reality platform based on Ethereum. Users can create content and dApps and monetize them. The created content can be made available for interaction with other users. Land in Decentraland is marked using a Cartesian coordinate system, with the community holding permanent ownership, and creators having complete control over their works.

Decentraland stores ownership of digital assets and other tradable information on the Ethereum blockchain, while other information requiring real-time interaction, such as scene states and user locations, is stored on users' computers or the private servers of scene owners. Developers need to choose which information is worth storing on the blockchain, as it incurs higher costs.

2) The Sandbox

The Sandbox is a community-driven UGC platform where users can gain ownership of digital land and creative content. Their works can be freely traded and become components of this user-generated metaverse. All elements in the metaverse are driven by the community.

The token on Sandbox, SAND, is an ERC-20 token, while the certification and trading of digital assets on the platform follow the ERC-1155 and ERC-721 standards. This information is stored on the Ethereum blockchain. On the other hand, the actual media data of digital assets is stored on IPFS and utilizes Amazon's S3 cloud service to support the web frontend. Unminted digital assets are stored on S3 cloud servers, requiring decentralized storage solutions to further protect data privacy.

3) CryptoVoxels

CryptoVoxels is an Ethereum-based metaverse. Users can build, develop, and sell digital assets on the streets, with ownership permanently recorded on the blockchain. Styled similarly to Minecraft, users can use custom pixel blocks to construct their land. The platform also provides users with a native COLR token for coloring land. Currently, CryptoVoxels stores the media information corresponding to the digital works created by users on land in servers operated by the company. It has indicated on social media that it will consider transferring data to a decentralized storage system.

4) CryptoPunks

CryptoPunks is a set of pixel-style avatars with different characteristics, totaling 10,000. These characters were initially distributed for free through Ethereum wallets and now need to be purchased through secondary trading platforms. Initially, CryptoPunks aggregated the 10,000 characters into a single image to save gas fees and stored the hash of this image in a smart contract on the blockchain, but did not disclose the storage location of the original media data. As concerns about NFT storage risks have grown, CryptoPunks spent 75 million gas fees to store all avatars on the Ethereum blockchain.

5) Bored Ape Yacht Club

Bored Ape Yacht Club consists of 10,000 uniquely characterized ape avatars. All avatars have been minted and can be purchased on secondary trading platforms. BAYC has published the TokenID, SHA-256 hash, and IPFS hash for each avatar on its official website. It has also backed up the media data of each avatar in a decentralized storage system and released backup information.

6) Gods Unchained

Gods Unchained is an Ethereum-based NFT card game similar to Hearthstone. Players can build their own decks and play in different game modes, such as ranked matches and arena adventures. Cards can be freely traded on the market and are owned by players. Currently, the ownership of NFTs in the game is stored on the blockchain, while the metadata and media data of NFT cards are stored on the company's servers and integrated with smart contracts through API interfaces.

1.2 Trading Platforms

Opensea is the earliest and currently the largest NFT trading platform, accounting for over 90% of the trading market. Initially, Opensea also used centralized servers to store NFT metadata and media data. As the value of individual NFTs skyrocketed and data loss occasionally affected the platform, Opensea now also offers users decentralized storage options. Currently, creators can choose to use IPFS for decentralized NFT metadata and media data, but they need to pay for it.

Rarible is currently the second-largest NFT trading platform, supporting both ERC-721 and ERC-1155 protocols. The project stores the metadata and media data of NFTs minted by creators on the website's backend, which is a centralized server. New buyers can call them on the blockchain as needed.

SuperRare is an online art gallery that can also be used for trading. It has also issued its own trading token, RARE. NFTs auctioned on SuperRare do not display very detailed technical information to users, such as smart contracts, token IDs, and metadata, which may be a reason for the platform's consistently low market share. According to our research, SuperRare uses IPFS for metadata and media data storage.

1.3 On-Chain Storage

The blockchains currently used for NFTs mainly include public chains like Ethereum, Flow, BSC, and side chains like Polygon and Ronin.

Due to high gas fees and communication congestion on the blockchain, most NFT projects choose to store only NFT ownership data on the blockchain to ensure that ownership is immutable, traceable, and undeniable. Transactions do not require centralized credit institutions as intermediaries and can be completed directly on the blockchain through smart contracts. This method expands the circulation of NFTs and uses credit intermediaries not controlled by any third party.

The media data representing the actual form of NFTs is stored off-chain, and in some cases, more complex metadata information is also stored. Separating them from the ownership storage system casts a shadow over the ownership that is strictly protected by blockchain technology.

1.4 Off-Chain Storage

The mainstream off-chain storage methods currently used for NFTs include: centralized, centrally verifiable, decentralized, and decentralized repairable storage.

1) Centralized Storage

Most NFT projects do not have as large a market share as Opensea, and many are still in their infancy, paying little attention to the security of off-chain data storage. Specific identifiers in smart contracts can be used to return relevant metadata and media data. They typically use URLs on web servers as identifiers. These servers are operated by companies or provided by cloud service providers like Amazon. The risk of centralized storage is tampering and denial of service.

2) Centrally Verifiable Storage

Taking CryptoPunks as an example—initially, it stored the integrated image of its products on a centralized server and then stored the encrypted hash of that image in a smart contract for verification. The benefit of this approach is that the image can be verified through the hash, ensuring that no modifications have been made, thus protecting the NFT media data from tampering. However, the media data itself is stored on a central server, rather than being backed up across the network's nodes like NFT ownership stored on the blockchain, leading to potential risks of data loss and denial of service.

Centrally verifiable off-chain storage is an optimization of centralized methods but still carries many risks and cannot adequately meet the high reliability storage requirements for authenticated data in NFTs and the metaverse.

3) Decentralized Storage

IPFS, as a representative of decentralized storage, is gradually being accepted by the NFT industry. IPFS aims to provide decentralized addressing as a complement to traditional centralized HTTP. Taking Bored Ape Yacht Club as an example—its metadata and media data are stored on IPFS; IPFS provides redundant backups and stable content addressing. As an addressing network running on multiple nodes, it addresses the pain points of ineffective URL addresses in previous centralized storage methods, avoiding reliance on centralized service providers.

The decentralized addressing method of IPFS further improves the storage method of NFT metadata and media data, but as an addressing system, it cannot provide sufficiently secure and reliable storage services. Although the CID address will always exist in the system, the corresponding specific data does not have such stability. The reason is that the network nodes in IPFS are self-driven when backing up content—if only a single node or a few nodes back up the corresponding content, when these nodes fail or go offline, the stored data may disappear, leaving only an invalid message with the CID.

4) Decentralized Repairable Storage

As a new possibility for solving off-chain storage issues for NFTs, decentralized repairable storage systems have attracted widespread attention both within and outside the industry. Decentralized distributed cloud storage projects like Filecoin, Memo, and Arweave are actively exploring better storage solutions for NFT followers, with Filecoin and Memo launching NFT storage projects based on their respective storage ecosystems.

NFT.Storage is an NFT storage project based on the Filecoin ecosystem launched by Protocol Labs. NFTs stored through this project will be stored in IPFS or Filecoin. Currently, the capacity limit for a single piece of stored data is under 100MB. Its repair function is built on the incentive mechanism of Filecoin. Through a scoring and verification system for storage nodes, damaged or lost data can be promptly identified and repaired. However, the storage in IPFS is provided by Protocol Labs and requires more network nodes to participate, necessitating further decentralization. Storage on Filecoin has not yet connected to the main network and is provided by test network nodes, thus posing a risk of data loss due to network resets.

Metastorage is an NFT storage project based on the Memo ecosystem launched by Memo Labs. NFTs stored through this project will be doubly stored in IPFS and MEFS (a storage system independently developed by Memo Labs). Currently, there is no limit on the amount of stored data. Its repair function is based on the MEFS storage system, utilizing multi-replication and erasure coding redundancy mechanisms, while also providing an open verification method. The KEEPER role in the system is responsible for matching users with nodes that have passed verification and challenges, providing ongoing evaluation and maintenance. Although the overall repair mechanism of MEFS is decoupled from the blockchain, it still requires Memo to involve a larger range of nodes to support the MEFS system and build a stable ecosystem.

Decentralized repairable storage has the potential to become a solution for future NFT storage, better matching the storage of metadata and media data with NFT ownership storage. Currently, the product technology and scale are still in their infancy, and the degree of implementation remains to be further observed.

2. Four Opportunities for NFT Storage

2.1 Support for Metaverse Value

Generally speaking, the metaverse refers to a virtual world constructed using a series of technologies such as the internet and VR. This concept was born decades ago but has never been realized. With the rapid development of blockchain, the metaverse sees the prospect of becoming a reality—blockchain provides an ideal decentralized environment for the metaverse, and the emergence of NFTs offers a feasible way to authenticate digital assets. Due to the limitations of current blockchain technology, the actual content of NFTs requires a storage method that matches ownership storage. Driven by demand, the vigorous development of this technology is expected to break through the security bottleneck of decentralized NFT storage, while the decentralized cloud industry focused on solving NFT storage issues has a broader market space ahead.

In this blockchain-driven virtual reality, participants can have a vast and rich imaginative space, such as enjoying games, showcasing homemade artworks, owning and trading virtual properties, and even profiting from unique virtual economic systems. They can purchase land controlled by decentralized organizations, freely build on it in the form of NFTs, rent buildings to others for returns, or breed and sell rare pets for money.

The metaverse ecosystem encompasses all projects discussed in the previous section, most of which are still in the early stages. Blockchain is typically used to record and guarantee the ownership of users' digital assets, while the media data corresponding to ownership is mostly still stored on centralized servers or IPFS, lacking the same level of protection as ownership. This poses a certain risk to the integrity of digital assets. Without a complete and reliable storage loop, using blockchain technology to protect ownership will also lose its significance.

2.2 Infrastructure for the P2E Gaming Industry

The recent boom in P2E games has attracted widespread attention from players and the capital market. With Axie Infinity surpassing NBA Top Shot to become the highest market cap NFT project, it is believed that NFTs have immense potential in the gaming industry. Some existing crypto games include CryptoKitties, Cryptocats, CryptoPunks, Meebits, Axie Infinity, Gods Unchained, and TradeStars.

One captivating feature of this type of game is the "breeding" mechanism. Users can raise pets themselves, spending considerable time cultivating new offspring. They can also purchase limited edition/rare virtual pets and sell them at high prices. Due to the nature of value circulation in P2E games, the current storage methods cannot meet their higher security requirements.

Additional rewards have attracted many investors to join the games, making NFTs even more important. Another exciting feature of NFTs is that they provide a record of ownership for items in the game. Players can own their personal game items, promoting economic recognition within the ecosystem and benefiting both developers and players. Players and game developers, as NFT issuers, can earn royalties each time they sell NFTs on the open market, sending funds back into the ecosystem and forming a virtuous cycle.

The reliability of NFT storage will determine the growth ceiling of the P2E gaming industry. As the industry develops to a certain extent, the hidden dangers in NFT storage will inevitably receive increasing attention, and various game projects will have to invest in improving NFT storage to reduce risks.

2.3 A Huge Capital Market

The existence of NFTs creates a mutually beneficial business model—players and developers profit from the secondary NFT market while the blockchain community significantly expands the coverage of NFTs, including various types of digital assets and thriving virtual economic activities. Traditional online economic activities rely on centralized companies that provide trust and technology. Although blockchain has developed various financing channels such as ICOs, IFOs, and IEOs, the application scenarios remain very limited. NFTs greatly expand the additional attributes of blockchain, such as uniqueness, ownership, and liquidity.

With the help of NFTs, the application scope of blockchain is rapidly expanding. This allows everyone to be associated with specific events, just as in our real lives. The storage method of NFTs plays a crucial role in realizing this vision. Due to the smaller data volume, FT (Fungible Token) can be stored on the blockchain, while NFTs require equally reliable storage methods.

Let’s take ticket purchasing—a common economic activity—as an example. When purchasing tickets in the traditional market, consumers must trust the third party providing the service. Therefore, consumers face the risk of being deceived or purchasing invalid tickets. These tickets may be fake, forged, or cancellable. In extreme cases, the same ticket may be sold multiple times, or non-transferable tickets may be traded in the market.

NFT-based tickets issued by blockchain can prove the right to attend any sports or cultural event. NFTs benefit from the solutions to double spending, tampering, and forgery issues that have already been addressed in the FT phase of blockchain. The uniqueness of distributed ledgers gives NFT tickets a clear advantage over traditional tickets. NFT-based tickets are unique and tamper-proof, meaning that ticket holders cannot resell the tickets after they are sold. NFTs serve as blockchain-based smart contracts, providing a transparent ticket trading platform for consumers, event organizers, and other stakeholders. Consumers can buy and sell NFT tickets through smart contracts without relying on any third party.

2.4 Protecting Digital Intellectual Property

Digital collectibles encompass various categories—trading cards, wine, digital images, videos, virtual real estate, domain names, diamonds, cryptocurrency stamps, intellectual property, and other physical items. Taking the art industry as an example, first, artists following traditional methods have few channels to showcase their works. Entering traditional channels requires funding, networking, and a significant amount of effort. Due to a lack of attention, prices cannot reflect the true value of artworks. Even if works are published on social networks, platforms and advertisers charge intermediary and advertising fees.

NFTs can convert their works into digital formats with comprehensive advantages. Artists do not have to hand over ownership and content to agents, which means a higher potential for profit. Typical examples include Mad Dog Jones' REPLICATOR sold for $4.1 million, Grimes' works totaling around $6 million, and other great digital artists' crypto works, such as the renowned Beeple and Trevor Jones. NFTs perform well in protecting property rights, while the corresponding metadata, media data, and other content have not been well protected by secure and reliable storage protocols.

Moreover, artists cannot earn royalties from future sales of their works under traditional models. In contrast, NFTs can be programmed so that artists receive predetermined royalties every time they trade their digital works in the market. This is an effective way to manage and protect digital masterpieces. Most importantly, platforms like Mintbase and Mintable have even established tools to support ordinary people in easily creating their own NFT works.

The media data of digital collectibles is essentially the NFTs themselves. If the ownership and royalty rights of a work lose value, the NFTs will become meaningless. Traditional collecting activities often come with higher storage costs. Collecting in the digital age clearly requires better storage solutions.

3. Five Challenges of NFT Storage

To realize the development of the aforementioned NFT storage applications, a series of obstacles must be eliminated, just like any other new technology. This article discusses some typical challenges faced by NFTs from the perspectives of usability, security, regulation, and scalability, such as system-related issues caused by Bitcoin-based platforms and human factors like regulation, policy, and social impact.

3.1 Usability

Usability refers to the process of measuring the effectiveness, efficiency, and satisfaction of users when testing a specific product. Most NFT projects are built on Ethereum. Therefore, it is evident that the main drawbacks of Ethereum have been inherited. We will discuss three major challenges that directly affect user experience.

a) Scarcity Redundancy Mechanism
Through the above analysis, NFTs currently use centralized data centers and IPFS for storage. However, the redundancy mechanisms of both methods are not very reliable. Centralized data centers typically create multiple file copies for redundancy, which is costly. IPFS does not have a self-operating redundancy method. Although the CID corresponding to each file is broadcasted across the network, the data of the file itself is stored locally on nodes, relying on other nodes for spontaneous backups. Filecoin, as the incentive layer of IPFS, has also not fulfilled its mission of incentivizing node backups—the majority of data stored in network nodes is merely for incentives, thus ineffective.

b) Slow Confirmation

NFTs typically send transactions to smart contracts to make activities like minting, selling, and exchanging transparent and trustworthy. However, the current NFT systems are tightly coupled with their underlying blockchain platforms, resulting in very poor performance. Bitcoin's speed is only 7 TPS, while Ethereum can only provide 30 TPS, making the confirmation speed of NFTs very slow.

Addressing this issue requires redesigning blockchain topology, optimizing its structure, or improving consensus mechanisms. Existing blockchain systems cannot meet these requirements. This also leads to complex metadata and "massive" media data being stored in off-chain systems.

c) High gas fees are a major issue for NFTs, especially during large-scale minting when metadata needs to be uploaded to the blockchain network. Each NFT-related transaction is more expensive than simple transfers because smart contracts involve computational resources and storage. Complex processes, immense pressure from communication congestion, and expensive fees severely limit the widespread application of NFTs. In most cases, the cost of producing NFTs far exceeds the current value of NFTs. Storing NFT-related data off-chain as much as possible is currently the mainstream solution to address this severe imbalance. However, it also brings various risks.

3.2 Storage Security and Privacy

User data is the top priority for any system. For off-chain storage data associated with labels on the blockchain, there is a risk of losing connection between the two or being misused by malicious parties.

1) Inaccessibility of NFT Data

In mainstream NFT projects, most encrypted hash values are used as identifiers rather than real media data. They are then recorded on the blockchain to reduce gas consumption. However, the possibility of losing or damaging the original files makes users feel insecure about NFTs. Some NFT projects have begun collaborating with specialized file storage systems, such as IPFS, which allows users to address content through hash values. As long as someone manages the IPFS network, users can always obtain the corresponding content that matches the hash value.

Nevertheless, such a system still has unavoidable flaws. When users upload NFT metadata and media data to IPFS nodes, there is no guarantee that their data will be replicated across all nodes. Data is stored on IPFS and may only be hosted by a single node, with no backup on any other nodes. If the only node storing it becomes disconnected from the network, the data may become unavailable.

Additionally, NFTs may also point to incorrect file addresses. If this is the case, users cannot prove that they indeed own the NFT. In summary, relying on external systems as core components of the NFT system will always have vulnerabilities.

2) Anonymity/Privacy

Most NFT transactions rely on their underlying Ethereum platform, which only provides pseudonymity rather than strict anonymity or privacy. Users can partially hide their identities. If the connection between the real identity and the corresponding address is known to the public, all activities under the infected address can be observed.

Existing privacy protection solutions, such as homomorphic encryption, zero-knowledge proofs, ring signatures, and multi-party computation, have not yet been widely applied to NFT-related solutions due to their complex cryptographic primitives and security assumptions.

3.3 Regulatory Policies

The legal and policy issues faced by NFTs cover a wide range of areas. Potentially relevant areas include commodities, cross-border transactions, KYC (Know Your Customer) data, etc. It is crucial to have a proper understanding of relevant regulatory reviews and litigation before entering the NFT space.

In some countries, the legal requirements for cryptocurrencies are very strict, and the same goes for NFT sales. Regulatory difficulties are unavoidable when minting, trading, selling, or purchasing NFTs. Legally, users can only trade derivatives such as stocks and NFTs on authorized exchanges. Other countries, such as Malta and France, are attempting to implement appropriate laws to regulate digital asset services. They require buyers to follow complex and even contradictory terms. Therefore,

Taxable Property. Products related to intellectual property, including artworks, books, domain names, etc., are considered taxable property under the current legal framework. However, NFT sales have not yet been included. Although a few countries like the United States tax cryptocurrencies as property, most regions of the world have not considered taxing crypto assets. This could significantly increase the number of financial crimes disguised as NFT transactions to evade taxes from the respective governments. Individual participants are taxed based on any capital gains related to NFT property. Furthermore, transactions such as NFT-for-NFT, NFT-for-IP, and Eth-for-NFT should also be taxed. In addition, high-profit properties or collectibles should be subject to higher tax rates.

3.4 Scalability

The scalability of NFT solutions includes two aspects. The first emphasizes whether a system can interact with other ecosystems; the second point is whether the NFT system can be updated when abandoning the current version.

1) NFT Interoperability

Existing NFT ecosystems are isolated from each other. Once users choose a product, they can only trade within the same ecosystem—this is limited by the underlying blockchain platform. Currently, if someone wants to trade across different ecosystems, they need to do so through a third-party trading platform like Opensea. Breaking away from the trust mechanism of the original blockchain platform increases trust costs. Interoperability and cross-chain communication have always been obstacles to the widespread promotion of dApps, and cross-chain communication can only be achieved with the help of external trusted parties. This inevitably compromises the quality of decentralization to some extent.

Fortunately, most NFT-related projects use Ethereum as their underlying platform. This means they share similar data structures and can exchange under the same rules. Different NFT projects have different storage methods. How to establish a unified risk structure while maintaining decentralization is an important issue for the future.

2) Updatable NFTs

Transitional blockchains typically update their protocols through soft forks and hard forks, illustrating the difficulties and trade-offs of updating existing blockchains. Although it is a universal model, new blockchains still have strict requirements, such as tolerating specific adversarial behaviors and remaining online during the update process. NFT programs heavily rely on the underlying platform and must remain consistent with them. While data is typically stored in separate components (such as IPFS and MEFS file systems), the most critical logic and tokens are still recorded on the blockchain and need to be appropriately updated.