Derive V3 is live. We built a rollup around the needs of a derivatives exchange, using Celestia for data availability, Succinct for proofs and Ethereum for settlement.

As trading grows, we process more updates to positions, balances and margin. We need to handle that volume while giving users access to the data they need to verify their holdings outside Derive’s systems.

With V3, we run the exchange and publish the data behind its state changes to Celestia. We use Succinct to generate cryptographic proofs that those updates follow the protocol’s rules, then submit them to Ethereum for verification and settlement.

Larger portfolios without V2’s gas limit

In V2, we work within an OP Stack allowance of 500 million gas every two seconds. A single BS76 option calculation uses about 1 million gas. Within those constraints, we support portfolios of up to 500 assets and settle between 3 and 30 trades per second.

We built V3’s execution environment around options, portfolio margin and risk calculations. We launch with support for 1,000 assets per portfolio and can raise that limit as we add proving capacity.

We can increase capacity without waiting for another chain’s team to raise its execution limits.

Matching trades while proofs run

In V3, traders receive matching responses in under 1 millisecond. We generate proofs outside the matching process, so we can accept the next trade without waiting for the previous trade’s proof.

We estimate that one RTX 5090 GPU can prove a trade in 0.5 to 5 seconds, depending on the computation required. We can distribute proving work across more GPUs as demand grows.

During a sharp market move, we can continue matching orders while processing the proof workload in parallel. Traders receive a matching response before Ethereum settlement; the sub-millisecond figure measures matching latency.

State data outside our database

Users need the underlying state data to check their positions and balances. Keeping that data in our private database would leave them dependent on us for access.

We publish it to Celestia so users and other operators can reconstruct protocol state outside Derive’s systems. We also use that data for our escape hatch: after the protocol’s censorship timeout, another operator can reconstruct the state and run a sequencer. During development, we integrated the Celestia client into our proof program with less work than we expected. That helped us build the full flow from data publication through to Ethereum settlement.

Balance proofs and private positions

We are also exploring how traders could prove their Derive balances on Ethereum. A trader could use that proof to borrow against an options position through a lending protocol that accepts it as collateral.

For institutions, we are working with the Celestia ecosystem on designs for encrypted account data. The aim is to keep positions and trades private while preserving users’ ability to prove their own state and verify data availability.