BNB Chain builds new Layer 1 for agentic trading, targets 2027 mainnet

Quick Take

• BNB Chain is building a new Layer 1 blockchain designed for agentic trading, targeting sub-50-millisecond transaction preconfirmation and no public mempool to help mitigate front-running.
• The chain will run alongside the existing BNB stack, with a testnet planned for the end of 2026 and mainnet deployment in early 2027, according to the project's H2 roadmap.

BNB Chain is building a new Layer 1 blockchain designed for agentic trading, offering its first detailed look at an architecture that has been in development for months.

The chain will sit alongside the existing BNB Chain stack and target sub-50-millisecond transaction preconfirmation, according to the project's H2 2026 technical roadmap shared exclusively with The Block ahead of publication.

Its stated design goal is to approach the execution experience of a centralized exchange while preserving the self-custody and transparency of running onchain.

The honest gap on CEX-grade execution

Asked whether an onchain venue can match a centralized exchange, BNB Chain conceded the limit of its own pitch.

A centralized matching engine executes in microseconds, but the team argued that no trader experiences microseconds in practice, with real-world round-trip times running single-digit milliseconds for co-located market makers and higher for everyone else. Sub-50-millisecond preconfirmation puts the chain in the range a typical exchange user actually feels, David Z, BNB Chain chief technology officer, told The Block.

The exception he named pointed to co-located high-frequency trading, where a centralized exchange still wins today. What the chain offers instead is settlement finality over assets users custody themselves, rather than a balance in a company database that can take hours to withdraw.

"For co-located HFT, a CEX still wins today. For everyone else, this is the CEX experience without the custodial risk," Z said.

How the chain is built

The proposed architecture removes the public mempool through a feature called TxStream, routing transactions directly to the block leader to cut latency and make common forms of front-running, such as sandwich attacks, more difficult.

With no public mempool, Z said, there is no window in which an attacker can see a pending trade and build a sandwich attack around it, because ordering is committed before anyone learns a trade exists. The design raises an obvious question of whether the block leader itself becomes the single point of extraction or censorship.

BNB Chain's CTO explained leaders rotate every 200 milliseconds, too fast for any validator to build a business on the position, while sub-50-millisecond commitments make ordering behavior publicly auditable and put a misbehaving validator's stake and reputation at risk. "TxStream doesn't eliminate MEV. Nothing does. It makes the dominant attacks impractical by design," Z told The Block.

A second component, PriorityLane, would reserve block space for oracles, liquidations, and bridges, governed onchain.

BNB Chain said it targets throughput above 100,000 transactions per second on the new chain through co-optimized consensus, parallel execution, and LtHash-based storage, alongside sub-one-second block finality. The figures are design targets for a chain that has not yet reached testnet.

The team plans to ship the chain on testnet by the end of 2026, with a mainnet release in early 2027.

The execution-layer bet

BNB Chain's CTO identified the performance gap in the execution layer rather than in consensus or storage. "EVM chains do a huge amount of repeated work because the same popular contracts (a DEX swap, a token transfer) run millions of times," Z said.

The team points to just-in-time compilation, a decades-old technique from mainstream software, and strength reduction as the core optimizations, and calls the work unglamorous compiler engineering.

On the risk of splitting BSC's own liquidity, the CTO said the new chain will have an official native bridge to BSC, with BSC serving as the settlement hub and BNB as the unified asset across every chain. "The goal is growing the total BNB Chain pie, not re-slicing it," Z stated. He framed the new chain as complementing BSC by targeting use cases the older chain was not built for, rather than rehosting its mature DeFi ecosystem.

The new chain would be the fourth in BNB Chain's stack, joining BNB Smart Chain, opBNB, and Greenfield, with the company positioning the expansion as offering developers flexibility rather than forcing migrations.

The roadmap also arrives amid a broader race among high-performance blockchains to break execution-layer bottlenecks, with projects including Solana's Firedancer, Monad, and MegaETH pursuing their own parallelization and throughput gains.

Quantum security and H1 receipts

BNB Chain is separately testing quantum-resistant security, using a hybrid approach that layers post-quantum protection over existing cryptography rather than replacing it.

Developers are researching how account abstraction could let users adopt quantum-safe protection without changing their wallet addresses.

"Nobody in the industry has a complete quantum migration scheme yet, including us," Z said. Where builders control the design, it has already made post-quantum choices, pointing to a lattice-based LtHash state commitment that makes the chain's state integrity post-quantum today.

Account migration is harder, the CTO said. Keeping a wallet address stable while the signature scheme changes means binding the address to an upgradeable authentication policy rather than to a single key, an approach its account abstraction is designed for but has not finalized. Z called harvest-now-decrypt-later a genuinely open problem for the whole industry, since any account whose public key has been exposed is theoretically vulnerable once a powerful enough quantum computer exists, and no protocol design retroactively fixes that.

The roadmap also highlighted BNB Smart Chain's H1 performance gains, including block intervals cut to 450 milliseconds and benchmark throughput nearly doubling to about 5,200 transactions per second. The full roadmap will be published on Wednesday, the team told The Block.