Summary: Nillion is building what it calls a “blind computer”: a privacy-focused network for storing and processing sensitive data without exposing it to application backends or infrastructure operators in plaintext. The current primary-source surface is broader than a single protocol or product: the docs separate private compute (nilCC), private storage (nilDB), private LLM inference (nilAI), a coordination chain (nilchain), and Blacklight, an Ethereum-L2-based verification layer for TEE workloads. Taken together, the official materials make Nillion look less like a single MPC or TEE point product and more like a multi-layer confidential-compute stack with onchain coordination and verifier incentives.
What it does:
Provides private compute through nilCC, which runs developer Docker workloads inside TEEs with attestation, workload hashing, and workload-manager tooling
Provides private storage through nilDB, which splits encrypted data into shares distributed across multiple nodes and exposes higher-level SDKs for querying and permissions
Provides private LLM inference through nilAI, using OpenAI-compatible APIs that run inside TEEs and return signed responses plus attestation surfaces
Operates Blacklight, a community-run verification layer that challenges and verifies TEE workloads via Nillion’s Ethereum L2, staking, and reward distribution contracts
Maintains nilchain, a Cosmos-SDK coordination layer for payments related to blind computations and storage operations
Key claims:
Nillion docs describe the blind computer as a decentralized network of nodes that stores and processes sensitive data while keeping it encrypted and private from both app backends and infrastructure operators
The private-storage docs say sensitive data is stored as encrypted shares across clusters of nilDB nodes so no single node can reveal the original value
The private-LLM docs say nilAI exposes OpenAI-compatible APIs, runs models inside TEEs, offers attestation endpoints, and signs responses so developers can verify provenance
The nilCC docs describe a confidential-compute stack where developers can run Docker Compose workloads inside AMD SEV-SNP or NVIDIA confidential-computing environments with attestation and workload-integrity verification
Blacklight docs say the verification layer uses Nillion’s Ethereum L2 plus permissionless verifier nodes to continuously challenge and verify TEE workloads, while the nilchain repo describes a separate Cosmos-SDK coordination layer for payments around blind computation and storage
Whitepaper: No canonical standalone whitepaper or litepaper was found in the official materials reviewed during this pass. The strongest primary sources were the official site, docs for the blind-computer modules and Blacklight, the GitHub organization, and the nilchain repo; see ../whitepapers/nillion-primary-sources-2026-04-27.md.