DECENTRALIZED AI INFRASTRUCTURE · BUILT ON SOLANA

IOCHAIN

DISTRIBUTEDCOMPUTENETWORK

Turn idle consumer GPUs into a global AI inference network. Browser-native nodes. FHE privacy. ZK proofs. On-chain revenue sharing. No PhD required.

EXPLORE DEMO READ WHITEPAPER

SCROLL

┌─ IOChain v0.1.0

│ STATUS: BUILDING

└─ NODE.COUNT: 0

CHAIN: SOLANA ─┐

PRIVACY: FHE │

PROOF: ZK ┘

0M+

Consumer GPUs Worldwide

~80% idle

USD/month saved vs cloud

per developer

0.00%

Privacy guarantee

FHE-protected

Ecosystem participants

complete flywheel

THE FLYWHEEL

Four Participants.
One Ecosystem.

Investors, developers, deployers, and AI agents form a self-reinforcing flywheel. Each participant captures value; the network grows stronger.

INVESTOR ─────── funds ──────► DEVELOPER ─── deploys model ──► MODEL REGISTRY
    ▲                                                                    │
    │                                                                    │
  earns                                                              distributes
    │                                                                    │
    │                                                                    ▼
AI AGENT ◄────── calls ───────  ROUTER ◄──── serves ──────── DEPLOYER NODE
    │                                                              (GPU/CPU/NPU)
    └────────── pays IOC ──────────────────────────────────────────────► ┘

01INVESTOR

Fund Models. Earn Per Request.

◈

Purchase semi-fungible tokens representing fractional ownership in AI models. Every inference request triggers atomic on-chain revenue distribution to your wallet.

─Fractional model ownership via SFT
─Atomic per-request revenue split
─On-chain transparent ledger
─Diversify across multiple models

02DEVELOPER

Deploy Models. Monetize Intelligence.

◇

Upload your model weights, issue your own tokens, set your revenue split, and let the network handle deployment, scaling, and payments automatically.

─One-command model registration
─Encrypted weight storage
─Automatic token issuance
─Royalty on every inference call

03DEPLOYER

Contribute Compute. Earn Passively.

○

Open a browser tab or install a lightweight node to contribute your GPU, CPU, or NPU. No Linux required. Earn IOC tokens proportional to your compute contribution.

─Browser-native node (WebAssembly)
─One-click desktop installer
─CPU / GPU / NPU supported
─Privacy-safe sandbox execution

04AI AGENT / USER

Access AI. Pay Per Request.

△

Call any model through the IOChain API with the same interface you use for OpenAI. Privacy-preserving, semantically routed, and priced at a fraction of cloud costs.

─OpenAI-compatible SDK
─FHE-encrypted inputs
─Semantic request routing
─Pay-per-request pricing

EXPLORE THE ECOSYSTEM→

PATENT PENDING

Eight Technology
Pillars.

Privacy, integrity, and efficiency — across every phase of the inference lifecycle. Hardware-free. Software-defined. Mathematically guaranteed.

T1EXECUTION

WebAssembly Sandbox

Models compile to WASM binaries. Memory isolation enforced at the VM level. No privileged system access. Runs anywhere — including your browser tab.

wasm_exec.linear_memory = isolated

T2PRIVACY — WEIGHTS

Sliding-Window Streaming

Model weights stream layer-by-layer. Only one layer decrypted at a time. Previous layer overwritten with random noise before the next loads. Complete weights never reside in accessible memory.

plaintext_layers_in_memory = 1

T3PRIVACY — INPUTS

Fully Homomorphic Encryption

User inputs encrypted under FHE. The node computes on ciphertexts — it sees neither your data nor the result. Plaintext-ciphertext operations cut computational cost by 100×+ vs full FHE.

bootstraps ÷ 100 vs ciphertext-ciphertext

T4INTEGRITY

Layer-Wise ZK Proofs

Each compute node commits layer activations to a Merkle tree on-chain. Distributed proving nodes generate ZK sub-proofs per layer. Aggregate proof verified in constant time.

proof_time ÷ N_layers per node

T5VERIFICATION

FHE Canary Injection

Known test inputs, encrypted under FHE, are indistinguishable from real user requests. Injected probabilistically. If a node cheats, it gets caught — and slashed.

P(detection) = 1 - (1-p)^n

T6ROUTING

Private Semantic Router

Your intent stays private. Solver nodes compute homomorphic dot products on encrypted intent vectors to find the best model — without learning what you're asking.

dot(enc(intent), capability) = enc(score)

T7DEPLOYMENT

Browser-Native Nodes

Open a URL and start earning compute rewards. WebAssembly + WebGPU. No installation, no Linux, no config. Or install our native node software for GPU and CLI power.

node_start = open_browser_tab()

T8ECONOMICS

On-Chain Revenue Split

Every inference event triggers atomic Solana token distribution to all stakeholders. Investor, developer, and deployer shares execute in a single transaction. Fully transparent.

atomic_distribution(investor + dev + node)

DEEP DIVE: FULL TECHNICAL STACK→

VIRTUAL DEMO

As Simple As
Opening a Tab.

No DevOps. No credit cards up front. No cloud contracts. Just a browser, your hardware, and a few lines of code.

STEP 01

Install & Connect

One npm install. Works alongside your existing OpenAI or Anthropic SDK calls. Drop-in compatible API surface.

STEP 02

Discover Models

Browse community-deployed models. Filter by cost, latency, capability, and privacy level.

STEP 03

Run with FHE Privacy

Pass privacy: 'fhe' to encrypt your inputs before they leave your device. The compute node never sees your data.

SEE FULL INTERACTIVE DEMO→

EARLY ACCESS

Join the Network.

Testnet launching soon. Join the waitlist to be among the first investors, developers, and node operators on IOChain.