21X and LayerZero Report 10,000 Matches Per Second in Onchain Trading Pilot

European trading venue 21X and interoperability protocol LayerZero say a private, permissioned deployment of Zero sustained 10,000 order matches per second in a month-long technical pilot. The companies disclosed the result on November 4 in a joint release, reporting a 45-fold performance improvement against the EVM layer-two baseline they used.

The number is unusually specific for a market crowded with throughput claims. It also comes with an important caveat: an internal pilot is not sustained production trading under real market stress, and the partners have not claimed otherwise.

What did the pilot actually test?

The two companies ran a fully onchain central limit order book, with an emphasis on tokenized US equities. That distinction matters. Matching orders onchain is technically different from recording a trade after an offchain matching engine has already determined the outcome, so the reported throughput addresses a genuine obstacle for blockchain-based venues.

The private instance matters too. Network membership and visibility on a permissioned deployment can be controlled differently than on a permissionless public chain. Comparisons with public networks require caution, because validators, transaction patterns, hardware assumptions and network conditions may not be equivalent.

Why institutions want a permissioned stack

Regulated markets need surveillance, market-access controls, records, resilience and procedures for handling trade errors. Some institutions will not, or cannot, expose all transactional details on a public ledger. A permissioned environment can offer more predictable operations while still delivering programmable execution and a shared data layer.

The trade-off is governance. Permissioning changes who defines participation, how upgrades occur and what happens when the network goes down. A fast benchmark does not settle those questions, and it does not prove the system fits every regulated market.

How should readers read the 45x figure?

Performance comparisons are only as good as their baseline. LayerZero says the pilot ran roughly 45 times faster than a referenced layer-two EVM setup — useful for the experiment's stated conditions, but not generalizable to all blockchains or workloads.

Throughput can shift when orders are cancelled, modified or run through complex compliance checks. Latency, finality, settlement integrity and recovery from outages matter as well. Any venue would still need to show that market data and order priority stay fair at high volumes.

Where does this fit in tokenized securities?

Tokenized securities are usually framed as a custody or issuance problem. Active exchange trading introduces harder requirements: price-time priority, matching, market making and operations during volatile conditions. The partnership signals that infrastructure providers are taking on those less visible tasks.

The promise of an onchain order book is more auditable shared state, but that benefit depends on execution design and on who can inspect the records. A private test is a starting point, not proof of an operating securities market.

What comes next?

The credible milestones are independent testing, disclosure of methodology, resilient multi-participant deployment and approval for any regulated activities the venue intends to pursue. Real trading volume and an incident history would say more than repeated peak-performance figures. Until those arrive, the strongest supported conclusion is that 21X and LayerZero reported a promising benchmark under controlled conditions.