Editor's note: Fully Homomorphic Encryption (FHE) is a technology that processes data without decrypting it. This means that companies can provide services without viewing user data, and users will not notice any difference in functionality. Since data is encrypted during transmission and processing, network behavior can be encrypted end-to-end. In other words, FHE allows zero trust to be better implemented and can be shared on untrusted domains, and the person performing the calculation cannot read the data.
Industry Vision
Zama, a leader in the FHE industry, recently published a post about its “Master Plan.” The post announced that the company had raised $73 million at an undisclosed valuation and outlined its vision for creating an end-to-end encrypted network called HTTPZ (“Z” for “Zero Trust”).
Zama was founded four years ago and has achieved the advancement of FHE from theoretical mathematics to practical code, thereby improving the accessibility of developers and expanding the scope of FHE applications. Currently, Zama's FHE library suite supports end-to-end encryption applications in various industries and has greatly improved the speed of FHE schemes. Its fhEVM (a confidential smart contract solution) solves the privacy problem in blockchain transactions. Zama believes that FHE has a lot of potential in blockchain applications, including privacy tokens and decentralized identities (DIDs), and emphasizes the application of FHE in artificial intelligence, which will have a wider impact in the future.
Several FHE builders in Web3 share Zama’s goals and are pushing to make it a reality.
This article will share the views of the founders of Mind Network, Fhenix and Inco, three popular projects in the FHE track, and explain how they realize end-to-end encrypted networks in Web3, why these projects will fundamentally change the way users interact with the network, and why they believe that the application scenarios of FHE are promising.
Mind Network
Mind Network is the first general-purpose FHE-based Restaking Rollup solution, providing secure computation and consensus for EigenLayer and the Ethereum ecosystem.
Crypto AI and DePIN still have some challenges to solve to beat their Web2 competitors. In crypto AI, if other validators can replicate the predictions, the system is tempted to reduce the amount of computation but still receive token rewards through verification, thus reducing network security. Therefore, encrypted output is key.
Another challenge facing crypto AI is how to launch a decentralized verification network. EigenLayer provides the market with a service for this problem, which allows security to be shared through ETH and liquidity staking tokens. But at the same time, because AI calculations are more complex, AI requires more sophisticated operations than ordinary cryptocurrency transactions. This is another key challenge that AI systems need to solve.
In the case of DePIN, users earn token rewards by contributing specific data, but they also inadvertently expose important data such as devices, geographic locations, and income to the network. If DePIN becomes the industry standard for today's IoT, Web3 users will have worse confidentiality than users in the Web2 model. This is a key challenge that DePIN aims to solve.
Mind Network provides solutions to the above problems. Mind Network uses Zama's FHE library to implement verifiable decentralized computing on encrypted data, thereby solving the first problem mentioned above. Secondly, Mind Network expands EigenLayer's consensus service to meet the needs of artificial intelligence computing, thereby realizing the key to artificial intelligence networks - probabilistic consensus.
Currently, Mind Network's artificial intelligence solutions have achieved initial product-market fit with projects such as IO.Net, AIOZ, Chainlink, Connext and Akash.
Fhenix
Since its inception, Ethereum has chosen to trade data integrity for confidentiality. Users can trust Ethereum to follow the rules of the system, such as keeping honest financial records, but users cannot maintain the same trust when it comes to sensitive information.
This opposition greatly limits the types of use cases that Ethereum can handle. In fact, for Ethereum to truly develop into Web3, users need to ensure that Ethereum can not only do what the current network can do, but also do it better. Take the "poker game" as an example - although Ethereum can be trusted not to cheat, it cannot make each player hide the cards from each other. If this cannot be done, the game cannot be played at all.
Such applications can only be achieved if the problem of on-chain confidentiality is solved, which is where FHE comes in. Fhenix uses and extends Zama's cryptographic library to build an FHE coprocessor. FHE coprocessors are extensions of Ethereum (L1, L2, or L3) that allow applications to outsource specific computations that require processing sensitive data. For example, a DAO governance mechanism could run a private voting mechanism, allowing people to encrypt their votes, and then have the coprocessor (on encrypted data) tally them while only revealing the final result.
Fhenix's FHE coprocessor technology is based on a lightweight FHE Rollup architecture, which greatly improves scalability. Assuming that each chain is equipped with such a coprocessor, it can promote the emergence of countless new applications. Fhenix believes that this will become a catalyst for more than one billion users to flock to cryptocurrency.
Inco
Inco is an EVM-based Layer 1 blockchain, secured by Ethereum through EigenLayer, and simplifies the complexity of FHE, enabling developers to build confidential DApps in 20 minutes using the most commonly used smart contract language Solidity and tools in the Ethereum ecosystem such as Metamask, Remix, and Hardhat.
Additionally, similar to how Celestia provides Data Availability (DA) to Ethereum and other blockchains, Inco, as a modular confidential computing network, extends confidentiality to Ethereum and other public L1 and L2 by providing confidential storage, computation, and access control.
For example, an untrusted on-chain game can be developed on Arbitrum with most of its core logic hosted on Arbitrum, while Inco is specifically used to store hidden information (such as cards, player status, or resources) or perform private computations (such as payments, voting, or hidden attacks). Inco's goal is to bring confidentiality to the value layer of the Internet and drive the next stage of large-scale applications.
Summarize
The founders believe that an end-to-end encrypted network is the only potential solution to the most critical problems of the network, and it may take four years or eight years to achieve this goal. But the zero-trust infrastructure enabled by FHE brings reasonable and mandatory privacy protection for transactions and data, which will help bring DePIN to the masses and help decentralized artificial intelligence defeat centralized artificial intelligence.
Looking ahead: The significance of fully homomorphic encryption
Fully homomorphic encryption (FHE) is the holy grail of cryptography and the key to contemporary privacy and security needs. Its origins can be traced back to 1978 when it was first proposed by Rivest, Adleman, and Dertouzos. However, it was not until 2009 that Stanford PhD candidate Craig Gentry realized this vision with a groundbreaking paper that provided the first working FHE scheme.
This technology enables complex computations to be performed on encrypted data without the need for decryption, providing a solution where data remains secure and private even during analysis, a process known as “creating a shared private state.” In just the past few years, advances in FHE have greatly improved efficiency and usability, transforming it from a theoretical concept to a practical tool for secure data processing.
Today, FHE is at the forefront of Web2 network security, with widespread use in cloud computing and data analytics. In these areas, sensitive information must be protected without compromising the ability to extract valuable insights. Web2 already has strict privacy protections in place, but despite being centralized, it is still vulnerable. Web3 was originally built for public data, and this is a key challenge that the Web3 ecosystem needs to solve. If Web2 became Web3 tomorrow, our grocery bills, app subscriptions, phone bills, etc. would all become public information. Solving the confidentiality problem in Web3 is critical. FHE may be a powerful solution for users to achieve enhanced privacy and security in the future, while allowing operations on encrypted transactions, data, and smart contracts while maintaining confidentiality.
Among the three methods of Zero Knowledge Proofs, Multi-Party Computation and Fully Homomorphic Encryption (FHE), FHE is the cornerstone. These three methods constitute a new vertical field in Web3: Decentralized Confidential Computation (DeCC). DeCC will greatly expand the use cases of Web3 and make Web3 widely used.