BNB Chain wants a faster trading network. The hard part is proving it before 2027.
BNB is making a clean bet: the gap with centralized exchanges is mostly execution speed, not consensus theater.
CryptoVibe Desk · bnb-chain · layer-1 · defi
- →BNB Chain told CoinDesk it plans a parallel trading-focused network, with testnet targeted for late 2026 and mainnet in early 2027.
- →The design removes the public mempool and leans on execution tricks like just-in-time compilation, which makes this a compute bet.
- →Watch the end-2026 testnet for open benchmarks, because all current performance numbers come from one CTO interview.
- public mempool → A waiting room where pending transactions sit before a validator includes them in a block.
- preconfirmation → An early signal that a transaction will be included before the block is fully final.
- finality → The point when a transaction is treated as settled and very hard to reverse.
- just-in-time compilation → A runtime technique that turns code into faster machine instructions right before execution.
BNB Chain is building a separate trading network. CoinDesk reported Wednesday that the chain targets traders and AI agents. Public testnet is planned for late 2026, with mainnet in early 2027.
The thesis is simple and unusually direct. BNB is saying the gap between self-custodied trading and centralized exchange speed is a compute problem. Not mainly a consensus problem. Not mainly a settlement problem. Execution is the bottleneck.
The reported targets are aggressive. BNB Chain told CoinDesk the new network is aiming for more than 100,000 TPS, preconfirmations under 50ms, and sub-second finality. Those numbers are not independently verified yet, so treat them as design goals, not shipped performance.
The mechanism matters more than the headline number. The chain removes the public mempool and streams transactions directly to the block leader. That cuts visible waiting time and reduces front-running. It also changes who sees order flow first, which is always the part engineers should inspect.
At the execution layer, BNB is leaning on just-in-time compilation. That means the system tries to run contract code faster at runtime, closer to how high-performance runtimes optimize hot paths. This is a cache invalidation problem with money attached: speed improves, but correctness and reproducibility become harder to reason about.
BNB Smart Chain still stays alive beside it. CoinDesk said it had about $5B in TVL as of July 8. It also reported 450ms block times and benchmark throughput near 5,200 TPS. That matters because BNB is not replacing the production chain. It is building a faster lane next to it.
If you're trading on-chain, this is the real question: who gets exchange-like speed without giving up the parts that made self-custody worth using? Removing the mempool helps. Reserved blockspace for liquidations, price feeds, and bridges could help. Account abstraction, gas sponsorship, passkeys, and scheduled execution could make the UX less painful.
And that's the catch. The roadmap names a lot of primitives, but it does not give timelines for most of them. Quantum-resistant security is still research-stage, not part of the 2026 to 2027 plan.
Rollups have not seriously contested this exact argument. They mostly talk about settlement, data, and sequencer design. BNB is saying the next fight is raw execution. That is coherent. It is also unproven until the testnet is public and the benchmarks are reproducible.
BNB Chain's choice to build a separate execution-first network is the right split, because BNB Smart Chain's roughly $5B production base should not absorb this experiment.
By the end of 2026, the public testnet needs an open benchmark showing more than 100,000 TPS, under-50ms preconfirmations, and no public mempool under load.
Primary links and supporting reads used by the desk for this story.
Forward this.
