Forward the original title: Delve into Mind Network ‘深入 Mind Network ：当全同态加密遇见 Restaking，加密 AI 项目的共识安全触手可及’

AI and Restaking are widely recognized as the leading narratives accompanying this bull market cycle.

The former has already produced various star AI projects, while the latter, with EigenLayer at its core, has given rise to multiple LRT projects, with various points-earning gameplay constantly emerging.

However, there is a very clear feeling that these two narratives seem to have entered a halftime rest phase. Although the number of projects in the track has increased, they have become increasingly homogeneous, and innovative stories from 0 to 1 are becoming harder to find.

At the same time, when AI and Restaking become a “correct narrative,” this “correctness” does not mean “perfection”:

Have a large number of AI/Depin projects truly decentralized? Recent data also shows that the TVL of Eigenlayer is declining. Can Restaking only be used to ensure the security of Ethereum’s ecological AVS?

Therefore, in the second half of the hot narrative, the projects that solve key common problems are the treasures that need to be unearthed.

Starting from this thought, the Mind Network on the current market has caught our attention. It can both solve the problem of many AI/Depin projects not being decentralized enough and make Restaking more useful and valuable.

If EigenLayer is seen as the solution for Ethereum’s ecosystem re-staking, Mind is the solution for re-staking in the AI field:

Through more flexible use of re-staking and the consensus security solution of fully homomorphic encryption, it ensures the token economy security and data security of decentralized AI networks.

What’s more important is that the project has already completed a $2.5 million seed round of financing participated by well-known institutions like Binance in 2023. Currently, it also has deep cooperation with hot new AI/Depin projects like io.net and Myshell. The anticipation of mainnet launch and incentive activities also bring great expectations.

However, for most readers seeing this project for the first time, on one side is the hard-to-understand fully homomorphic encryption, and on the other side is the profit-chasing Restaking. How can these two be combined to solve key issues for AI projects?

In this article, let’s delve into Mind Network, and understand this potential project that combines AI, Restaking, and fully homomorphic encryption, all hot narratives in one.

AI Projects Eagerly Slay Dragons, Yet Become Dragons Themselves Due to Lack of “Zero Trust”

To understand what Mind Network does specifically, we first need to understand the issues current AI projects are facing.

Perhaps, the dragon slayers are gradually becoming dragons themselves, which has become the best footnote to describe the current crypto AI projects.

From the perspective of slaying dragons, the core narrative of crypto AI (or DePIN) projects is decentralization. They use more decentralized computing power, algorithms (models) and data to challenge the monopoly of large companies on all elements of AI, breaking the trust in these companies’ authority.

Although this narrative is correct and naturally popular, AI, after becoming decentralized, seems to have a greater potential to become a dragon:

It is unable to achieve “zero trust” towards validators in a decentralized environment.

Sounds a bit difficult to understand? Let’s look at a specific example.

For instance, in a common crypto AI project, everyone needs to decentralize the validation/voting of AI models to select the best model.

However, in practice, the business model often has the validators (nodes) within the project selecting the best-performing AI model. How can you ensure that the one they select is indeed the best?

Following their selection under the PoS mechanism does not equate to “the best selection, fair selection.”

Similarly, in the business of AI agents, when ranking services based on performance, how can you ensure the top-ranking services are truly the best?

In DePIN scenarios, when a task is assigned to a node in DePIN for computation, how can you ensure the validator fairly assigns this task to the appropriate node, instead of cheating and giving it to a familiar node?

These examples actually reflect a key common problem — in each decentralized AI network, the decision of the validator becomes the center you must trust.

So, you must trust the decision of the validators or key participants in the network, hoping that they won’t act maliciously or make the right decisions.

Projects loudly proclaiming decentralization are, in fact, constrained by internal trust within the network. Zero trust has not yet been achieved, the current AI narrative is not perfect.

What is needed to face these problems?

Clearly, we need to resolve the reliance on trust in validators or key participants for validation/voting/decision-making within the current AI projects network through some technical mechanism and economic design.

This is also Mind Network’s niche and battleground.

The Holy Grail of Fully Homomorphic Encryption: Mind Network Places It in the Perfect Position

Mind Network excels in what is considered the Holy Grail of cryptography: Fully Homomorphic Encryption (FHE).

But what do the problems exposed in AI and Depin projects have to do with FHE?

If we look at the essence, these problems point unanimously to the distribution, selection, and decision-making of resources - unrelated to technology, but to “human governance.”

All these areas related to human governance, where there is room for malpractice, are based on the premise that network participants can fully understand known information publicly (if I know a major player voted, then I’ll also vote).

The clever ones among you must have sensed where FHE comes into play:

What if the information is no longer known to everyone?

Fully Homomorphic Encryption (FHE), for short, perfectly addresses these issues related to human governance.

FHE is pursued as the Holy Grail of cryptography, and Vitalik Buterin has recently emphasized its role in the Web3 field. We won’t spend a lot of ink explaining the principle of FHE here, you just need to know its function - it allows complex computations to be performed on encrypted data without the need for decryption, providing a solution where data can remain safe and private throughout the analysis process.

But to hold the Holy Grail, one must bear its weight.

Indeed, while FHE’s encrypted computation is beneficial, its resource overhead is significant. Using it for AI model training is cost-prohibitive and not a sensible approach for crypto AI projects.

Mind Network’s use of FHE has a certain finesse, placing the Holy Grail in the most appropriate position.

That is, FHE isn’t used for AI model training and parameter changes, but rather for cross-validation, selection, ranking, voting, and other areas full of “human governance” after AI model training. This approach keeps resource overhead manageable and the problem to be solved very clear:

If participants in AI networks conduct business without knowing each other’s selection/voting results, there won’t be any “follow the big players, trust authoritative nodes” behavior. This eliminates decision bias brought about by identity influence, allowing decentralized decision-making to return to its original state and truly good AI models and AI services to be recognized.

Hence, the road to making FHE do general computation is fraught with challenges. However, using FHE for a specific decentralized aspect—validation—is self-consistent and feasible. Ensuring zero trust in the validation process allows for consensus security and genuine decentralization in crypto AI projects.

On the other end of security, there’s fairness.

We can use a specific case to see how Mind Network’s fairness is reflected in the encrypted execution of validation:

• 1. AI projects access the fully homomorphic encrypted validation service through Mind’s product SDK.
• 1. At the same time, AI projects register on the Mind Network to confirm their identity. Mind will generate a smart contract on the target project network/chain to synchronize subsequent operation changes and execution results.
• 1. AI projects publish validation tasks on the Mind Network that need to be performed using fully homomorphic encryption (like which AI model is better). The FHE voting service comes into play, allowing validation nodes of AI projects to execute the voting process without seeing each other’s plain text voting results.
• 1. The voting results and related data changes are transferred to Mind’s own chain through a smart contract and are synchronized and accounted for promptly.
• 1. In the above steps, AI projects that use Mind’s services will be charged Mind project tokens as gas fees (tokens have not been released yet).

Similarly, if we consider a specific DePIN project, using Mind Network will also achieve a more fair resource allocation effect. We can take IO.net, which collaborates with Mind Network, as an example:

• 1. IO.net accesses the fully homomorphic encrypted validation service through Mind’s product SDK.
• 1. After accessing the service, each node possessing a GPU gains the consensus ability under fully homomorphic encryption. That is, when an AI computing power task arrives, both the request and the data are encrypted, making it possible to fairly allocate the task to the appropriate node.

Wait, But What Does This Have to Do With Restaking?

Everything mentioned above seems to be on the technical level, but what does it have to do with asset-based Restaking?

Mind Network provides a solution based on FHE, promoting validation security on the technical level for AI networks; however, to join the validation and enjoy this security, it is inextricably linked with the economic network structure of most AI/Depin projects.

PoS, or Proof of Stake, is the underlying consensus logic for most crypto projects.

Therefore, if any AI project accepts Mind Network’s fairer FHE technical support for AI model/service selection, sequencing, and validation, since most project nodes represent voting/validation rights through the PoS mechanism, the size of the staked assets under this node is closely related to the right to participate in FHE-guaranteed fair validation.

Mind Network’s key move at the asset level is to expand the range of Staking and Restaking in a public way, coupled with homomorphic encryption to ensure validation consensus in AI networks.

Different roles participating in the network can satisfy their own different interest demands.

For validation nodes of AI projects, increasing the amount of Restaking provides more opportunities and voting rights for carrying out FHE validation tasks in Mind Network.

For general users, they can stake their LST/LRT assets to the above nodes in a delegation manner to obtain APR income.

This seems to have similarities with the Restaking of EigenLayer that we are familiar with, and they are essentially the same:

EigenLayer is using restaking to ensure the security of different AVS in the Ethereum ecosystem; Mind Network is using restaking to ensure consensus security in the crypto ecosystem, for different AI networks.

It’s worth noting that the reason why it is “the entire ecosystem” is inseparable from another key function of Mind Network: Remote Restaking.

Because of Remote Staking, there is no need to cross-chain your LRT tokens on different chains. You can indiscriminately stake your LRT on different chains to a validation node of a certain AI network through remote staking, greatly lowering the entry threshold for users and integrating fragmented liquidity under the multi-chain situation.

Extensive Ecological Construction and Solid Technical Strength

What other catalysts are currently worthy of attention in Mind Network?

Firstly, in terms of products, the testnet has already attracted 650,000 wallets and generated 3.2 million transactions. The full functionality of the mainnet is expected to launch soon.

Secondly, in terms of ecosystem construction, because the product is positioned to empower other AI projects, it is crucial to attract cooperation from top projects.

Currently, Mind Network provides AI network consensus security services for io.net, Singularity, Nimble, Myshell, and AIOZ, offers an FHE Bridge solution for Chainlink CCIP, and provides AI data security storage services for IPFS, Arweave, and Greenfield. The inclusion of top AI, storage, and oracle projects suggests the potential for Mind Network to become a “golden shovel.”

Furthermore, in terms of background, in 2023, the project was selected by the Binance incubator and completed a seed round of financing worth $2.5 million participated by well-known institutions such as Binance. It also received the Ethereum Foundation Fellowship Grant, was selected for the Chainlink Build Program, and became a Channel Partner signed with Chainlink.

In terms of technical strength, apart from the team itself, which includes top-notch professors and Ph.D.s in AI, security, and cryptography from leading universities and companies, an important aspect worth noting is the collaboration with the top fully homomorphic encryption research company in the industry.

In February of this year, Mind Network announced a partnership with ZAMA, a leading open-source encryption company in the field of fully homomorphic encryption research, which has completed an A-round financing of $73 million led by Multicoin and Protocol Labs.

Recently, the cooperation between the two parties has further expanded, jointly launching a new Hybrid FHE (Mixed Fully Homomorphic Encryption) AI network to promote the application of AI algorithms on encrypted data, which undoubtedly adds another layer of technical benefits to the project itself.

According to sources close to the news, Mind Network chose to use ZAMA’s underlying technical library for its own technical research and development in the cooperation with ZAMA. This move effectively shows Mind’s expertise:

Fully homomorphic encryption has a huge resource overhead, and the underlying library ensures maximum performance output without dragging down performance.

In addition to empowering itself with better technology, Mind Network is also outputting its capabilities to help improve the crypto ecosystem.

In May, the project also collaborated with Chainlink to launch the first Fully Homomorphic Encryption (FHE) interface built on the Cross-Chain Interoperability Protocol (CCIP). This not only enhances the security of cross-chain communication and transactions but also realizes a more trustworthy and user-centric Web3 ecosystem.

As of the time of writing, Mind Network has already reached cooperation with several top projects in different ecosystems and tracks. Considering its positioning to empower other projects, we might expect the “golden shovel” effect in the future.

Conclusion

When Fully Homomorphic Encryption (FHE) meets Restaking, Mind Network could potentially become a new driving force in the second half of this year’s mainstream crypto narrative.

With FHE as a medium, it can reach a large number of crypto AI projects for business optimization, providing support for the true “decentralization” and zero trust of decentralized AI projects. With Restaking paving the way, it further absorbs liquidity from different chains, and the rapid increase in project TVL (Total Value Locked) is also foreseeable.

Undeniably, the Holy Grail of Fully Homomorphic Encryption attracts market attention to new stories, while Restaking attracts market liquidity. As the consensus security of AI projects becomes more accessible, the concentration of attention and liquidity will inevitably lead to expectations for the project’s future development.

Like Mind Network, which polishes the correct narratives (AI, Restaking) to perfection through its own technology, is it not a more gentle disruption in the second half of the mainstream narrative?

Disclaimer：

1. This article is reprinted from [TechFlow]. Forward the Original Title‘深入 Mind Network ：当全同态加密遇见 Restaking，加密 AI 项目的共识安全触手可及’. All copyrights belong to the original author [TechFlow]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.