Nillion Completes $25 Million in Financing. What is its main feature of blind computing? What is the difference between it and ZKP and FHE?

Many friends who saw the news that Nillion raised $25M in financing were curious about what "blind computing" is. Just when we have some understanding of the unfamiliar concepts of MPC, ZKP, FHE, and TEE, a new concept has emerged. So, what is the general workflow of blind computing? What is the blind computing solution provided by Nillion? Next, let me talk about my understanding:

1) What is Blind Compute? Simply put, blind computing is a secure computing method that allows the server (node) to perform computing tasks on a piece of encrypted data, ultimately achieving privacy protection.

The goals of enhanced encryption algorithms such as ZKP, TEE, MPC, and FHE are consistent, but the differences are: ZKP zero-knowledge proof generation requires huge overhead, and is suitable for off-chain storage + computing, and on-chain only verification scenarios, such as Rollup Layer2; TEE trusted execution environment is a method that relies on hardware manufacturers to perform calculations in an isolated environment; although FHE fully homomorphic encryption can perform calculations directly on encrypted data, it currently only supports specific operations;

"Blind computing" is a more general computing framework, because encryption technologies such as ZKP, TEE, and FHE may be used as part of its technical framework.

As we all know, ZKP, TEE, FHE, etc. are currently in the stage of exploring and optimizing the application of Crypto technology. Blind computing has the potential to aggregate and apply these core encryption technologies, thereby exploring an integrated engineering practice solution for privacy protection.

2) The core logic of blind computing is to enhance distributed nodes, so that a single node has the ability of segmented storage + computing at the same time, and add a verifiable open governance network, so as to achieve the result of effective work without knowing the "complete" data. How to understand it?

Normally, to protect data privacy, data needs to be stored at Node A, then encrypted and computed by Node B, and then decrypted and verified by Node C to complete the data storage + computing work. This process has a huge cost loss for data transmission, and the data is exposed in the process of repeated Encrypt-->Decrypt. The cost of mutual trust between nodes is also high, making it difficult to ensure privacy.

The business logic built by Nillion just makes up for this shortcoming. Its general workflow is as follows (for understanding only):

Nillion has built a distributed node network, each of which has enhanced storage + computing capabilities. When the Nillion network receives data transmission and processing requirements, it first performs compilation preprocessing through the Nada specific language, so that the original data is split into many fragments, and all of them are in an encrypted state.

The AIVM virtual machine will then schedule and allocate the data, and its distributed nodes will randomly store and calculate these data fragments, and finally complete the aggregation and unified verification. During the whole process, a single node cannot know the entire data content, but when pieced together, it can complete the encrypted transmission and calculation of the overall data.

Why blind computing can aggregate and apply technologies such as ZKP, TEE, and FHE? The logic is very simple. FHE homomorphic encryption technology can be fully applied in data preprocessing, that is, in the data encryption stage, and node storage and computing data can be performed in the TEE trusted execution environment. When aggregating and verifying the work results of the nodes, ZKP can be used to improve the verification aggregation efficiency.

3) In my opinion, technologies such as ZKP, TEE, FHE, and MPC all have some engineering implementation defects to a greater or lesser extent. Currently, almost every track in the Crypto field is crowded with projects, but most of them are working on cost and efficiency optimization, and are all focused on specific application scenarios of Crypto.

Although the blind computing framework proposed by Nillion has not yet been widely used, its integrated encryption solution is likely to be widely adopted in broader data protection fields such as AI verifiable computing and machine learning.