State of Nillion Q2 2025

Key Insights

• The Nillion ecosystem has over 60 projects live or in development. Partners include Coinbase Developer Platform, Monad, Deutsche Telekom, Alibaba Cloud, stc Bahrain, and Vodafone’s Pairpoint.
• Nillion released Fission, a research paper with Meta on privacy-preserving LLM inference, and published The Nillion Standard, outlining new foundations for privacy and computation in the AI era.
• Nillion’s circulating market cap fell 33.9% from May to June, ending Q2 2025 at $56.9 million.
• Nillion’s eligible staked supply steadily increased from 3.3% at the end of March to 4.4% by the end of June.
• Nillion launched its Technical Roadmap, Product Roadmap, and Phase 0 upgrade, the network’s most significant release since mainnet.
• Nillion was named a 2025 top AI startup by market intelligence company CB Insights.

Primer

Nillion is a decentralized, privacy-preserving computation network founded in 2021 to address a critical gap in digital infrastructure: how to handle, store, and process sensitive data securely in distributed systems. Nillion leverages Privacy-Enhancing Technologies (PETs) such as Multi-Party Computation (MPC), Fully Homomorphic Encryption (FHE), and Trusted Execution Environments (TEEs), which allow users to conduct computations on encrypted data without exposing the data. This ability is called Blind Compute, and it fundamentally alters the trust assumptions inherent in digital systems, offering a new paradigm for privacy and data security.

The Nillion network is built upon a dual-layer architecture. The coordination and governance layer, nilChain, ensures node integrity through staking, slashing, and community-driven upgrades, while the orchestration layer, Petnet, handles privacy-preserving computation and data storage. This separation helps ensure scalability and security. A suite of application frameworks supports this infrastructure, including nilCC (private computation on sensitive inputs), nilDB (encrypted distributed database), and nilAI (private AI tooling), enabling developers to build a broad spectrum of privacy-first applications.

Website / X (Twitter) / Discord

Key Metrics

Financial Analysis

Market Cap

Nillion’s circulating market cap declined 33.9% from May to June, ending Q2 2025 at $56.9 million. NIL, the network’s native token, mirrored this decline, ending June down 33.9% MoM at $0.29.

Eligible Supply Staked

Throughout Q2, Nillion’s eligible staked supply, the share of staked tokens out of the stakeable token supply, steadily increased from 3.3% at the end of March to 4.4% by the end of June. The most significant growth occurred in April, when the staking ratio rose 0.82 percentage points (pp) to 4.1%, before slowing to smaller gains of 0.25pp in May and 0.11pp in June.

Staking APR

During Q2, Nillion’s staking APR declined, falling from 24.6% at the end of March to 22.6% by June. The largest monthly decrease occurred in April, when APR fell 6.2pp to 24.6% from 30.8% in March, followed by more moderate decreases of 1.4pp in May and 0.6pp in June. Despite this downward trend, the June APR of 22.6% remains higher than many proof-of-stake networks, providing holders with a relatively strong rate of return.

Network Analysis

Petnet Activity

From mainnet launch on March 24, 2025, to Aug. 29, 2025, Nillion’s Petnet activity included 422 million secrets and 200 GB of data stored in nilDB, 1.3 million inference calls to nilAI, and 42 validators on nilChain.

Validators

Since launch, the number of validators on Nillion increased 9.0% from 46 in March to 50 by June. Active validators, those participating in consensus and securing the network, remained stable at 42 throughout the period.

Developer Activity

In Q2, Nillion’s average daily GitHub commits increased from 11.2 in April to 13.8 in June. Development activity reached a monthly average low of 6.5 commits per day in May before rebounding 113.3% in June.

Qualitative Analysis

Notable Developments

Organization

In April, Nillion was named to CB Insights’ annual AI 100 list of the world’s most promising AI startups. Selected from more than 17,000 companies, Nillion was included in the Data Preparation & Curation category within the AI Infrastructure sector.

In July, John Woods joined the Nillion Association as Chief Technology Officer. Woods previously contributed to projects such as Monero and Algorand, and brings over a decade of experience in decentralized technologies. As CTO, he will oversee core technology, developer experience, and productization, guiding Nillion's next stage of growth.

Technical

nilGPT

In August, Nillion launched nilGPT, a privacy-preserving AI chatbot built on its infrastructure. Unlike conventional chatbots that store prompts in plaintext on centralized servers, nilGPT processes all interactions in secure enclaves and stores them as encrypted data across decentralized nodes, ensuring no single operator can access user information. The application introduced three modes: personal assistant, companion, and wellness advisor. This launch marked one of the first production deployments of the Blind Computer for end user AI services, and nilGPT has since expanded to support image uploads, mobile access, open-source development, and attestations that make all code changes publicly verifiable.

Blind Provable Interfaces

Also in August, Nillion introduced Blind Provable Interfaces (BPIs) as an extension of traditional APIs. BPIs allow applications to request and process user data through blind computation while ensuring provenance, authenticity, and integrity are verifiable. This framework enables users to retain control over their data, minimizes unnecessary data exposure, and incorporates mechanisms for user compensation. Early implementations include integrations supporting Tickr, where trading data from sources such as Hyperliquid and Binance can be retrieved and verified privately using BPIs.

Partnerships and Integrations

Coinbase Developer Platform (CDP)

In April, Nillion integrated its Private Storage with Coinbase Developer Platform’s Agent Kit, enabling AI agents to securely create, store, and retrieve sensitive data through natural language prompts. A demo highlighted how encrypted medical data could be stored in Private Storage and accessed by private models like Llama and DeepSeek to generate diagnostic outputs, without exposing the underlying information.

Primus Labs

To advance Private-to-Private oracles, Nillion linked its Private Storage with Primus Labs’ zero-knowledge Transport Layer Security (zkTLS) framework. The integration enables private Web2 data, such as account balances, to be imported into blind computation on Web3. Primus Labs’ browser extension retrieves TLS responses from sources like TikTok or Binance, while SecretVault stores the information as encrypted shares across MPC nodes. This allows sensitive records to remain confidential while still supporting secure aggregate computations.

Monad

In May, Nillion announced a partnership with Monad, an EVM-compatible Layer-1 blockchain optimized for high-throughput and low-cost settlement. The collaboration combines Nillion’s privacy-preserving compute with Monad’s execution environment to support decentralized science (DeSci) applications that require both confidentiality and performance.

The integration enables sensitive health and research data to be processed offchain through Nillion’s Blind Compute while governance votes, token transactions, and intellectual property trades are finalized on Monad. This dual approach is designed to meet the requirements of regulated industries such as pharmaceuticals and biotechnology, where compliance and efficiency are critical.

vVv

Nillion formed a strategic partnership with vVv, a fundraising platform offering early access to pre-token sales and token launches. The collaboration extends beyond capital investment, offering Nillion ecosystem builders access to tailored go-to-market support, strategic funding, and operational expertise.

Enterprise Cluster

In June, Nillion launched its Enterprise Cluster on the Petnet, partnering with stc Bahrain, Deutsche Telekom, Alibaba Cloud, and Pairpoint by Vodafone. The cluster introduces enterprise-grade node operators to Nillion’s privacy-preserving infrastructure, designed to secure sensitive data through blind computation rather than transaction validation. By running nodes on Petnet, these enterprises place trust in Nillion’s Blind architecture. Beyond node operations, Nillion is also exploring real-world use cases with each partner, making the collaboration a step toward deploying privacy-preserving solutions across sectors such as telecom, finance, and healthcare.

Research and Publications

Fission

In April, Nillion, in collaboration with Meta, released Fission: Distributed Privacy-Preserving Large Language Model Inference, a research paper addressing the challenge of running large language models (LLMs) both privately and efficiently. Traditional privacy-preserving machine learning systems often face tradeoffs between security and speed, with MPC approaches introducing significant computational overhead.

Fission introduces a hybrid framework that evaluates linear layers under MPC while delegating nonlinear functions to a separate evaluator network. This design preserves privacy while improving performance. In tests, Fission achieved three to eight times faster inference on models such as GPT and BERT compared to CrypTen, a traditional framework for secure machine learning. The framework enables practical use cases in which sensitive inputs such as medical records, financial information, or personal identifiers can be processed without being exposed.

The Nillion Standard

In May, Nillion published The Nillion Standard: The Intelligent Internet, a paper proposing new foundations for privacy and computation in the age of AI. It argues that the current internet enables data transfer but fails to protect user intent or provide secure infrastructure for intelligent applications, a gap that becomes critical as AI systems process increasingly sensitive data.

The paper introduces blind computation as a core primitive and outlines three guiding principles: data sovereignty, blind computation, and modular privacy. Together, these form the basis of an “internet for intelligence” where privacy is built into the infrastructure. Nillion has begun working toward this vision by launching applications such as Tickr, a marketplace for verified trading insights, and nilGPT, a personal chatbot that keeps all conversations private. The Nillion Standard presents this as a blueprint for developing applications that preserve user control while enabling collaboration and innovation across domains such as healthcare, finance, and AI.

Ecosystem Overview

The Nillion ecosystem has expanded rapidly, with over 60 projects and more than 75 native applications live or in development. Its modular framework supports a diverse range of use cases, from AI and decentralized science (DeSci) to DePIN (Decentralized Physical Infrastructure Networks) and beyond. Nillion’s Nucleus builder program and extensive developer tooling, including nilAI for private AI inference, nilDB for encrypted database management, and nilCC for privacy-preserving computation, have catalyzed this growth.

Projects

Fulcra

Fulcra introduced a privacy-preserving sleep study on Nillion. The application encrypts user data to generate personalized insights without invasive tracking or third-party data sharing.

Stadium Science

Stadium Science launched a “Sleep-to-Earn” campaign that drew more than 1,200 participants contributing encrypted sleep data. Built on Nillion’s privacy-preserving infrastructure, the program allows users to earn rewards while ensuring raw data remains private.

Skillful AI

Skillful AI launched Planning Strategy, a multi-agent system that plans, executes, and self-corrects with human-like adaptability, and NilRAG, a privacy layer on Nillion that enables agents to run private inference, store encrypted memory, and operate without data leakage. Together, they advance Skillful AI’s mission to build personalized, privacy-first AI assistants that think, adapt, and protect user data.

Healthblocks

Healthblocks officially went live on May 7, 2025, and later introduced token rewards for daily steps. The fitness application leverages Nillion to preserve user ownership and control of health data while enabling third parties to generate insights without accessing individual records. In July, Healthblocks launched a Walk-to-Earn campaign, challenging participants to collectively log six million steps in seven days, with rewards distributed from a $1,500 USDC prize pool.

Nubila

Nubila integrated Nillion to secure its decentralized network of weather stations, enabling privacy-preserving use of hyperlocal climate data. The integration supports applications in insurance, risk management, and disaster forecasting without exposing raw records. In July, Nubila launched a Snap-to-Earn campaign, rewarding participants from a $350 prize pool for submitting real sky photos.

MonadicDNA

MonadicDNA launched its iPhone alpha test in early June and delivered DNA results to participants who had provided saliva samples at ETHDenver. The genomics platform uses Nillion to keep genetic data encrypted throughout its lifecycle, providing a privacy-preserving alternative to centralized providers like 23andMe.

Tickr

Tickr debuted as one of the first native applications on Nillion, providing a marketplace for verified trading insights. The platform privately verifies trading history to generate trust scores, allowing creators to monetize their strategies while keeping wallet addresses and trade logs confidential.

Events and Hackathons

Ukraine Community Meetup

On May 4, Nillion’s local community hosted an in-person meetup in Kyiv, Ukraine. The event focused on introducing builders and enthusiasts to the power of privacy with Nillion.

Nillion Hackathon

In May, Nillion hosted a week-long internal hackathon to prototype applications leveraging its privacy-preserving infrastructure. Projects included Blind Reflections, an LLM-powered private diary built on Private Storage and Private LLMs; Blind Lawyer, a tool for contract analysis with enhanced inference capabilities; Bookie, a reading companion that summarizes saved content; Get Feedback, a secure widget for collecting user input; and nilShare, an application for private file sharing.

Nigeria Community Meetup

In June, Nillion’s local community hosted an in-person meetup in Abuja, Nigeria. The event brought together builders and enthusiasts to discuss privacy technologies and applications of blind computation. It marked Nillion’s first-ever community gathering in Nigeria.

Web3Privacy Now Hackathon

Also in June, Nillion served as the main sponsor of the Web3Privacy Now Hackathon in Berlin, a three-day event focused on advancing secure computation and digital sovereignty. The hackathon brought together interdisciplinary teams to develop privacy-preserving applications, highlighting use cases in areas such as decentralized identity, healthcare, and financial data security. Nillion’s Steph Orpilla hosted a workshop on PETs, outlining their capabilities, trade-offs, and how developers can apply them in practice without needing deep cryptographic expertise.

Korea Community Meetup

In August, Nillion’s local community hosted the first-ever in-person meetup in Seoul, Korea. The event featured presentations from Nillion’s Co-Founder, CMO, and CCO, who shared the project’s vision, walked through the blind compute ecosystem, and showcased applications such as nilGPT. Over 300 attendees engaged in discussions with the Nillion team, learning about the role of privacy in decentralized infrastructure.

Roadmap

Nillion’s roadmap is organized across two dimensions: the technical development of the Blind Computer and the product rollout of its components into production. Together, these roadmaps outline the path toward a modular, privacy-preserving compute layer.

Technical Roadmap

The technical roadmap outlines the phased development of the Blind Computer across 2025:

• Phase 0 (Q2 2025): Initial deployment of components, including nilDB for decentralized encrypted storage, nilAI for private LLM queries, and nilCC for secure computation within TEEs.
• Phase 1 (Q3 2025): Introduction of interoperability between components, enabling data stored in nilDB to be incorporated directly into nilAI queries. Auditability features will provide developers with verifiable assurances of system behavior.
• Phase 2 (Q4 2025): Focus on developer usability, with improvements to SDKs, deployment flows, and interfaces aimed at lowering friction for building applications and workflows.

Product Roadmap

The product roadmap outlines the staged release of Blind Computer functionality across 2025-2026:

• Phase 0 – Production-Ready Privacy (July 2025): Launch of a 10-node nilDB network (including six enterprise-operated nodes) and nilCC support for AI inference across models such as Llama 3B, 8B, 70B, DeepSeek 14B, Qwen 14B, and Gemma 27B. Secure retrieval-augmented generation (RAG) and decentralized permissioning via UCAN replaced the prior JWT-based model.
• Phase 1 – General Compute (Late 2025): Expansion to general-purpose computation, allowing workloads to run securely in nilCC while interoperating with nilDB. Privacy enforcement will be supported by nilAuth permissions and nilCC attestation outputs. AI capabilities will broaden to larger models and include features such as private web search and integrated safety tooling.
• Phase 2 – Default Privacy UX (Early 2026): Emphasis on developer experience, with SDK, deployment, and interface improvements intended to make building privacy-preserving applications comparable to conventional development. By this phase, the Blind Computer is expected to support scalable real-world deployments.

Closing Summary

In Q2 2025, Nillion strengthened its infrastructure, ecosystem, and research initiatives while advancing toward production-ready privacy. The circulating market cap declined 33.9% to $56.9 million, staking participation rose to 4.4% of the eligible supply, and validator count grew to 50. Developer activity also rebounded in June, averaging 13.8 daily commits.

The ecosystem expanded with more than 60 projects and 75 native applications in development, supported by integrations with Coinbase Developer Platform, Monad, and enterprise partners such as Deutsche Telekom, Alibaba Cloud, and Vodafone’s Pairpoint. Notable milestones included the launch of the Enterprise Cluster, new research outputs such as Fission and The Nillion Standard, and recognition by CB Insights as one of the top AI startups of 2025. Technical progress featured the release of nilGPT, a privacy-preserving chatbot, and the introduction of Blind Provable Interfaces (BPIs) as a framework for data access and compensation.

Community engagement accelerated through events and hackathons in Ukraine, Nigeria, Korea, and Berlin, alongside ecosystem campaigns such as Healthblocks’ Walk-to-Earn and Nubila’s Snap-to-Earn. Looking ahead, Nillion enters the second half of 2025 with momentum, supported by a clear technical and product roadmap and increasing adoption across AI, DeSci, and DePIN. The project remains focused on building the Blind Computer as the foundation for an “internet for intelligence,” where privacy and usability converge at scale.