Working papers and technical preprints for the Singularity Cloud Network.
Decentralized AI compute, verifiable inference, confidential hardware, tokenized compute markets, and agent-native cloud infrastructure.
This repository is the focused publication surface for the Singularity Cloud Network paper series. It is separate from the broader EventHorizon Labs research hub, which tracks pillars, experiments, and research-roadmap work across the lab.
Research home: ehlabs.xyz/research
The series studies one question from multiple angles:
How can AI compute become an open, community-owned, verifiable, confidential, and agent-native cloud network?
| # | Paper | Theme | Status | Source | |
|---|---|---|---|---|---|
| SCN-001 | Tokenized Compute Markets | Market mechanism for self-growing AI compute capacity | Working paper v0.1 | TeX | |
| SCN-002 | Singularity Grid | Verifiable, confidential, OpenAI-compatible inference across independent nodes | Forthcoming | - | - |
| SCN-003 | Singularity Hive Mind | Sharded large-model serving across many local/cloud nodes | Forthcoming | - | - |
A Self-Growing Mechanism for Community-Owned, Verifiable, Confidential, and Agentic AI Compute
The first paper introduces Tokenized Compute Markets, a mechanism that couples capital formation, physical infrastructure deployment, and AI inference into a self-growing loop. Trading activity funds real machines through a Hardware Fund, while inference remains priced in a stable unit and settled through cluster tokens.
Key topics:
- mechanism design for community-owned compute
- tokenized settlement and batched operator payouts
- hardware-funded capacity growth
- confidential AI inference and TEE trust tiers
- agentic market creation and autonomous compute participation
- failure modes: velocity, wash markets, demand/supply reflexivity, and liquidity fragmentation
Read:
papers/
2026-tokenized-compute-markets/
README.md
metadata.yml
releases/
tokenized-compute-markets-whitepaper.pdf
source/
paper.tex
docs/
series-roadmap.md
Use the repository-level CITATION.cff or the paper-specific metadata file. Until a DOI is assigned, cite the GitHub release or paper directory.
Unless a paper directory states otherwise, text and papers in this repository are released under the Creative Commons Attribution 4.0 International License. Code snippets, scripts, and machine-readable metadata are released under the MIT License.
EventHorizon Labs
Research toward decentralized, verifiable, confidential, and agentic cloud infrastructure.