Technology · FHE Confidential Compute § 06 of 11
Confidential Compute · L1 Native

Private
by Computation.

Imagine a smart contract that calculates your portfolio balance, checks a compliance rule, or computes a royalty — but never sees a single number. The data stays encrypted the entire time.

That's FHE · Fully Homomorphic Encryption. Smart contracts that process your data without ever revealing it.
The Pipeline · No Plaintext on Node Optimized by Google HEIR
01
Your Data
Numbers, formulas, IDs
02
Encrypt
Client-side · keys local
03
Compute
On ciphertext · on-node
04
Encrypted Result
Never decrypted on-chain
05
You Decrypt
Recipient only
4
FHE Schemes
BGV · BFV · CKKS · TFHE
2↔1
Scheme Switching
CKKS ↔ TFHE on-the-fly
0
Plaintext Exposure
Data never decrypted on-node
L1
Native Integration
Not a sidechain · not a coprocessor
Why This Matters · A Real Example

Two companies collaborate — without exposing data to each other.

Party A · Farmer

Digitized delivery notes

Pushes encrypted data on-chain — prices, volumes, margins. The blockchain stores them as ciphertext. Nobody else can read them.

BL_023 · 240L · €0.42/L · margin %· ENC
Smart Contract · FHE
f(price, volume) = score
runs on ciphertext only
Party B · Advisor

Performance formulas

Writes a smart contract: feed efficiency, milk quality score, cost ratio. Deploys on-chain. Sees only encrypted output — never the raw numbers.

score = 0x8f2a…3c1d · ENC

The farmer decrypts the result and gets his indicator. The advisor never saw the raw numbers. The blockchain guarantees the computation is correct.

Schemes

Four Schemes, One Runtime.

Each scheme optimized for a different workload — and ATSHI can switch between them on the fly.

01

BGV / BFV

Exact Integer

Zero rounding errors. 100 + 50 = exactly 150. For financial calculations, voting, compliance, audit trails.

Fast · Exact
02

CKKS

Approximate Real

SIMD batching processes thousands of encrypted values at once. Ideal for analytics, statistics, and complex formulas on encrypted data.

Fast · Analytics
03

TFHE

Fast Boolean

Millisecond decisions. Encrypted threshold checks, private auctions, access control. Ultra-fast boolean operations on encrypted data.

Very Fast · Logic
04

Scheme Switching

CKKS ↔ TFHE

Use CKKS for heavy calculations, switch to TFHE for exact comparisons — in the same transaction, no re-encryption. Unique to ATSHI.

Best of Both
Build

Two Ways to Build FHE Smart Contracts.

For developers who code — and for business users who don't.

For Developers

WASM Runtime

Write in Rust or AssemblyScript. Call FHE operations via native host functions — encrypt, compute, decrypt — like any API call.

royalty.ts WASM · OpenFHE
// FHE host functions — native in WASM runtime
import { fhe_encrypt, fhe_compute, fhe_decrypt } from "atshi:fhe";

export function compute_royalty(
  encPrice: Ciphertext,
  encRate:  Ciphertext
): Ciphertext {
  // Multiplication happens on ciphertext.
  // The node never sees price or rate.
  return fhe_compute("mul", encPrice, encRate);
}
  • FHE host functions directly in smart-contract code
  • Threshold decryption via ARCH consensus
  • Circuits auto-optimized by Google HEIR
For Business Users

No-Code Builder

Drag & drop your business logic in a visual interface. The platform generates the FHE smart contract automatically — no cryptography expertise required.

Builder
JSON
Python
HEIR
WASM
Pipeline · auto-generated
  • Visual builder → JSON → Python → HEIR → WASM on-chain
  • Pre-built templates: compliance, scoring, indicators
  • Deploy confidential contracts in minutes
Infrastructure

Built on the best open-source FHE stack.

BSD-2-Clause · DARPA origin

OpenFHE

The most comprehensive FHE library — supports all 4 schemes (BGV, BFV, CKKS, TFHE). Open-source, born from DARPA research. Used by ATSHI for on-chain encrypted computation.

openfhe.org →
Apache 2.0 · MLIR-based

Google HEIR

Google’s open-source FHE compiler. Automatically optimizes FHE circuits — selects the right scheme, manages noise budgets, generates WASM code. Makes FHE accessible without cryptography expertise.

github.com/google/heir →
Applications

What FHE enables.

DEFI

Confidential DeFi

Encrypted positions, no front-running, no MEV. Institutional-grade privacy for on-chain trading.

ML

ML Marketplace

Encrypted AI inference with per-call royalties. Data never exposed, model IP protected.

RWA

Hidden-Price NFTs

Royalties computed on encrypted values — sale price never visible on-chain.

HDS

Healthcare

Compute on encrypted patient data — diagnostics without exposure. HDS compliant.

FIN

Financial Services

Encrypted portfolio analysis, credit scoring, AML checks — regulators audit without seeing client data.

SCM

Supply Chain

Encrypted supplier data, auditable compliance. Share proof without exposing business secrets.

The Confidential Network Is Here

Protect privacy by math, not by trust.

Compute on encrypted data. Build on the first L1 with native FHE.