Technology
Privacy Agents
Community
Twitter
In Private Development

Value exchange at the
speed of thought

Next-generation infrastructure to settle value instantly. Built-in privacy. Agent-native. Currently in private development with partners.

<1ms
Edge finality
50ms
Optimistic root finality
Throughput ceiling

Edge goes brrr. Root seals the deal.

Edge Layer <1ms · bilateral · no consensus
Alice Bob
Agent 1 Agent 2
Carol Dave

Both parties sign every state update. Transfers backed by locked funds. Netting compresses thousands of operations into one settlement.

net settlement flows down
Root Layer 50–900ms · DAG consensus · BFT final
Settlement Batch clearing CLOB Disputes Cudo programs

DAG consensus with permissionless validators. Leaderless — no single sequencer. Settlement batches, MEV-free clearing, force-settlement adjudication.

10,000
edge operations
netting
1
settlement

Thousands of operations compress into a single settlement. The rest stays private.

Three problems that can't be patched

Wrong bottleneck

Alice sends Bob $5. 1,000 validators vote on it. Why?

$
$
$

MEV is architectural

Single ordering = extraction. It's structural, not a bug.

$0.50 $50

Token gas = friction brake

Chain succeeds → token rises → gas rises. Users lose.

Built different

Stablecoin Gas

Fixed-price fees in stablecoins. The chain's success never makes it harder to use.

MEV Eliminated

Batch clearing at uniform price. Ordering is irrelevant. Front-running is structurally impossible.

DAG Consensus

Leaderless. No single sequencer. Optimistic in 50ms. BFT final in 900ms. Formal proofs completed.

Native Order Book

Protocol-level CLOB. Price-time priority. Sub-ms order management via edges. Not a smart contract.

Privacy by Architecture

Edge operations are private between two parties — root never sees them. Three tiers: transparent, confidential, and fully shielded with ZK proofs.

Blockchains have ceilings. Ferros doesn't.

Ethereum ~30 TPS
Base ~100 TPS
Solana ~1,500 user TPS
Then there's Ferros
Ferros · 100 edges 100K TPS
Ferros · 10K edges 50M TPS
Ferros · 100K edges 500M TPS
Ferros · 1M edges 1B TPS

Every new edge adds capacity. There is no ceiling.

Private by default. Not by add-on.

Edge operations are invisible to everyone except the two parties — structurally, not cryptographically. For root-layer settlements, three privacy tiers let you control exactly what the world sees.

Structural edge privacy

Thousands of operations between two parties net to one settlement. Root sees a fraction of the gross volume. The data simply doesn't exist anywhere else.

No encryption needed. No ZK overhead. Privacy is a consequence of the architecture.

Three privacy tiers

Transparent for public DeFi. Confidential for hidden amounts via Pedersen commitments. Private for fully shielded operations through ZK proofs and a protocol-native shielded pool.

You choose per edge. The ratchet only tightens — never loosens.

Selective disclosure

Prove specific facts to auditors and regulators — "I paid invoice X" or "my balance exceeds Y" — without revealing anything else.

Cryptographic proofs. Unforgeable. Stronger than bank records.

Explore privacy architecture →

Built for the autonomous economy

AI agents need to transact thousands of times per hour with predictable costs. Ferros gives them sub-millisecond bilateral transfers, scoped delegation, and streaming payments — all enforced by the protocol.

<1ms transfers

Fast enough for real-time multi-agent coordination and x402 payments.

Scoped delegation

$100/day limit enforced by the protocol. Not the software. Not the agent.

Standing orders

Set rules. Go offline. The protocol executes. Your agent doesn't need 100% uptime.

Deterministic costs

Fixed stablecoin fees. No auctions. No surprises. Budget a year in advance.

Explore Agent Infrastructure →

We wrote a language for this

Solidity doesn't know what contention is. Cudo does. Declare how your state is accessed — the compiler verifies it, the infrastructure routes automatically. Reentrancy, double-spends, and resource leaks are killed at compile time.

owned
Edge · <1ms
per<K>
Root · Parallel
global
Root · Sequential

Interested in Ferros?

Ferros is currently in private development with partners. If you'd like to learn more or explore working together, we'd love to hear from you.