A Deep Dive into Nillion

PUBLISHED

06.21.2024

AUTHOR

Nillion

CATEGORY

Tech Updates

Earlier this week, we announced the Nillion Community Round on CoinList, live until June 26 at 17:00 UTC.

NIL is the native utility token of the Nillion Network, humanity’s first blind computer.

We recently spoke with the CoinList team to discuss the current state of AI, the NIL token’s utility, how Nillion empowers builders to develop new types of privacy-aware applications, our vision for the future.

Let’s dive in.

What is Nillion?

Nillion is humanity’s first Blind Computer.

Imagine a digital clone of yourself that could act as your assistant — an AI that knows you better than you know yourself.

Sound exciting? Sound Scary? Either way, it’s coming. And it has a big problem: Data.

That’s why builders hailing from Uber, Coinbase, Consensys, Google, Nike, Polygon, Amazon, and Hedera spent 3 years building: Nillion — the internet’s base layer for all private data.

Nillion has entered the data wars and is providing builders access to a powerful suite of Privacy Enhancing Technologies (PETs) to develop innovative applications like Personalized AI. The technology allows for data to stay hidden (encrypted) even while it’s being used — a function that’s critical for the future of AI and beyond. Blind Computation has since attracted a growing army of builders across AI, DeFi, custody, wallets, identity, messaging, and more.

What problems and opportunities does Nillion address?

Nillion’s Blind Computation network unlocks a new paradigm in AI, enabling data and models to be kept private. Nillion has created a decentralized alternative for data that cannot be entrusted to companies such as ChatGPT or fully public Crypto x AI networks.

But Blind Computation is bigger than just AI. Nillion provides programmable private data capabilities not existent in blockchains and other off-chain co-processors, unlocking use cases far beyond AI. More than 35 companies have officially partnered with Nillion to build on the network. Use cases range from private decentralized dark pools, quantum secure messaging, healthcare analytics, crypto custody, identity applications, and much more.

How do we plan to build this vision? The Orchestration Layer

There is no single silver bullet primitive in cryptography that will be the right tool for every job. Different Privacy Enhancing Technologies (PETs), such as the different flavors of Multi-Party Computation (MPC), Fully Homomorphic Encryption (FHE), and Zero-Knowledge Proofs (ZKPs), are meant to solve different use cases. Nillion brings these technologies together to create a foundational layer of cryptography and Blind Computation.

How does Nillion address concerns related to scalability, particularly as the network grows in terms of users, transactions, and data storage?

At the core of Nillion Network’s architecture is the concept of clustering. Every cluster consists of a variable number of nodes based on the different clusters’ security, cost, and performance assumptions. Given that the compute network is not a blockchain underpinned by the concept of a global shared state, it can scale vertically and horizontally — growth in storage or compute demand can be absorbed by whole clusters scaling up their infrastructure or through the onboarding of additional clusters of nodes.

Every cluster can be set up as a specialized cohort of nodes with different hardware to handle diverse storage or compute requests, for example, by provisioning large amounts of storage to handle more secrets or specific hardware to accelerate specific types of computations. Clusters make the network flexible.

What is the role of the NIL token within the network, and how is it utilized for transactions, governance, or other functions

Securing and Coordination Layer: Staking NIL accrues voting power that is used to secure the network and decide who is part of the active set of validators using a Delegated Proof of Stake mechanism.

Managing network resources: Users must pay NIL tokens to use the Coordination Layer or make blind computation requests, making it possible to manage network resources.

Economics of Petnet Clusters: Infrastructure providers join clusters to facilitate Blind Computation, with secure storage being rewarded in NIL for providing resources to the network.

Network Governance: NIL holders can stake to vote on on-chain proposals on the Coordination Layer or delegate their voting power to others.

You note partnerships with projects across a variety of verticals, including AI, DeFi, Healthcare, and Messaging. What do these partnerships unlock and what other use cases are you most excited about?

AI – Nillion unlocks the private training and inference of AI models, creating a decentralized alternative for sensitive data that cannot be entrusted to OpenAI or transparent crypto networks like Bittensor (TAO). Specifically, it unlocks some of the following benefits:

• AI Inference: Nillion enables enterprises to access proprietary cloud solutions, like ChatGPT, without having to sacrifice privacy and security. At the same time, Nillion enables model sharing without leaking IP. The integration collaboration with companies like Ritual or Nesa demonstrates the strong desire of existing crypto AI networks to integrate Nillion into their networks for private AI computation requests, positioning Nillion as the universal backend for private AI computation.

• Data sharing to earn: The privacy nature of data storage in the network enables personal data to be shared and monetized for AI training without sacrificing privacy. As an example, Rainfall AI is a privacy-based Personal AI platform that enables users to earn from the real-time intelligence generated from their data. Such a product enables businesses to better understand, engage, and acquire customers.

• AI Semantic search: Retrieval Augmented Generation (RAG) is a key part of the AI stack today, enabling users to provide relevant customized content in running LLMs. Today, the process of semantic search on vector data requires trust in a single company in storing and processing those vectors. Skillful is a company building out privacy-preserving RAG to store vectors in the Nillion Network with quantum-security while conducting semantic search on those vectors without having to decrypt the vectors.

• AI Synthetic data: Synthetic data can be generated while unlocking collaborative model training without leaking anything about personal data. This process results in more data with greater accuracy. Mizu aims to utilize the Nillion Network to facilitate private synthetic data generation.

• Agents: Protocols like Virtuals Protocol aim to use the Nillion Network for creating personalized private Agents. With the sensitive, personalized data sitting in Nillion, Agents can be used broadly without having to sacrifice user privacy.

Decentralized Compute – Nillion provides programmable data privacy capabilities not existent in blockchains and other off-chain co-processors. A blind computation network dramatically expands the design space for applications, such as:

• DeFi: A decentralized, fully privacy-preserving order book with use case applications across crypto and TradFi. For instance, Choose K is building a decentralized private order book enabling fully private trade execution without the risk of front-running or adverse MEV capture. Aphotic builds on top of Choose K’s order book to create a dark pool. Those two builders building on top of each other’s solutions exemplify the benefits of creating an open public network.

• Medical: A fully privacy-preserving method to analyze sensitive healthcare data patterns. For instance, patient data such as blood tests can be aggregated and processed in a fully regulatory-compliant manner. Maya Health uses real-world evidence to inform clinical practices, drug development, and regulation to support safe, effective, and accessible psychedelic practices. This company will use the Nillion platform to collaboratively compute on secure private user data.

• DAOs: The blind computation functionality enables DAOs to benefit from pre-vote and post-vote privacy, shielding users’ votes while generating the aggregated vote outcome. DecentDAO is building DAO tooling and implements such a solution.

Secure Storage – As a public, decentralized network for private data, Nillion enables the hidden storage of sensitive data with quantum security. This security introduces new storage use cases for crypto:

• Messaging: A decentralized, quantum-secure messaging system that ensures full user privacy. Mailchain uses Nillion to enable users to store messages securely and privately, while detecting and analyzing potential scams.

• Passwords: NilQuantum is building on Nillion to develop a solution to store user passwords quantum-securely in a decentralized manner.

• Documents: Flux supports group messaging as well as crypto functionality such as token-gated access, in-app payments, and integrations with other web3 services. Flux builds on top of the Nillion Network to store documents privately and securely on the network.

• Custody: By utilizing privacy-enhancing technologies like MPC, the Nillion Network is also uniquely positioned to enable users to store keys and execute transactions in clusters of their choice. Companies like Ambient, Interstellar, or JLabs intend to develop custody solutions on the network. Furthermore, companies like Salt are using the blind computation functionality to create private trading policies, positioning the network as a truly decentralized MPC custody network.

Personalized AI and privacy are all the rage in crypto these days. What is the most interesting trend in personalized AI that you’re seeing?

Since the emergence of generative AI in its current form, the industry has developed more and more towards enabling those systems to benefit from personalization and customization. Major companies like OpenAI introduced custom GPTs while endeavoring further iterations of personalized chatbots, enterprises developed proprietary models in-house, and start-ups formed to bring personalized agents to life.

What they all have in common is that personal, sensitive data is at the center of training and running those personalized models.

For truly personalized AI to exist, it needs ALL your data. Every message you have ever sent. Everything that makes you, you. Ask yourself a question, would you give an AI assistant 10 years of your text conversations knowing OpenAI owned it?
Most wouldn’t. 59% of consumers would NOT feel comfortable using personalized AI because of data privacy concerns. Without private data processing, personalized AI will not gain “the next internet” levels of mass adoption.

Apple’s recent release of Apple Intelligence and its focus on running AI models locally on the device and in a secure hardware enclave for computationally more intense models is a step in the right direction. Their focus on private AI is a huge testament to the Nillion thesis of the need for blind computation.

While Apple makes the first logical step in using Trusted Execution Environments, Nillion is developing the cryptography that eliminates trust by creating a truly decentralized privacy endgame using technologies like Multi-party computation (MPC), and Homomorphic Encryption (HE).

Nillion – The internet’s base layer for all private data.

The world changed when Ethereum decentralized computing with smart contracts. Today, Nillion will ignite the next wave of innovation by decentralizing trust for your most valuable data.

Allowing you, and only you, to own your data – unlocking the mass adoption of personalized AI, and countless other emerging technologies.

Nillion does this through: Blind Computation

What is the best way to get involved in the Nillion community?

Learn more by visiting the Nillion website or following us on Twitter, Telegram or Discord.

 

 

Not available for residents of the United States, China, Canada, and certain other jurisdictions.