Nillion

• Name: Nillion
• URL: https://nillion.com/
• Category: confidential-compute network / privacy-preserving data-and-AI infrastructure / verification-layer stack
• Tags: ethereum-ecosystem cosmos-ecosystem
• Summary: Nillion is a layered confidential-compute stack, not one clean product. The useful decomposition is private storage, TEE compute, private LLM inference, a Cosmos coordination chain, and an Ethereum-side verification layer — enough moving parts that the note should stay focused on where trust and control actually sit.
• 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.
• Sources:
  • https://nillion.com/
  • https://docs.nillion.com/
  • https://docs.nillion.com/blind-computer/learn/overview
  • https://docs.nillion.com/blind-computer/build/compute/overview
  • https://docs.nillion.com/blind-computer/build/compute/architecture
  • https://docs.nillion.com/blind-computer/build/storage/overview
  • https://docs.nillion.com/blind-computer/build/llms/overview
  • https://docs.nillion.com/blacklight/learn/overview
  • https://docs.nillion.com/blacklight/learn/network
  • https://github.com/NillionNetwork
  • https://raw.githubusercontent.com/NillionNetwork/nilchain/main/README.md
  • https://raw.githubusercontent.com/NillionNetwork/nillion-docs/main/README.md

Internal linkages

• Best TEE-heavy confidential-compute contrast: eigencloud.
• Best AI-verification and settlement contrast: opengradient.

Control surface

• The leverage sits in TEE attestation policy, share and storage design, verifier incentives around Blacklight, nilchain coordination rules, and the boundary between hosted operator convenience and genuinely decentralized execution.

• Treat Nillion as a layered confidential-compute stack, not as one monolithic blind computer answer.

• Last reviewed: 2026-06-15 UTC