AI agents may be ready to trade crypto — but the blockchains underneath them are not. That is the problem BNB Chain is betting its next architecture on solving. The project has unveiled plans for a BNB Chain new blockchain, a dedicated layer-1 network engineered from the ground up to handle the demands of AI-driven finance, high-frequency trading, and a future where quantum computers could render today’s cryptography obsolete.
The new network is not a replacement for the existing BNB Chain infrastructure — it is an addition. Both chains will run in parallel, with the new layer-1 purpose-built for workloads that today’s architecture simply cannot absorb at scale: automated payments, AI agent transactions, and high-frequency crypto trading at speeds that approach traditional financial markets.
The timing is deliberate. The broader crypto industry is currently racing to build payment rails for AI agents — software systems capable of executing financial transactions without waiting for human approval. Stripe-backed Tempo launched a payments-focused layer-1 alongside its Machine Payments Protocol in March. MoonPay released its Open Wallet Standard the same month, with contributors including PayPal, the Ethereum Foundation, the Solana Foundation, and Ripple. In May, Amazon Web Services partnered with Coinbase and Stripe to launch Amazon Bedrock AgentCore Payments, enabling AI agents to spend USDC stablecoins on APIs and data services. Coinbase followed in June with its own agent-facing product. BNB Chain’s roadmap lands squarely inside this wave.
According to its H2 2026 technical roadmap, the testnet is expected to go live by the end of 2026. The mainnet is then planned for early 2027. The developers framed the announcement as both a progress report and a forward declaration: “This roadmap opens with the receipts and closes with what comes next — a second half focused on doubling performance again, and an architecture designed for the decade ahead.”
The performance targets the team has set are aggressive by any standard. The new chain is designed to eventually process more than 100,000 transactions per second by running multiple transactions in parallel and rethinking how data is stored and verified. Confirmation times are targeted at under 50 milliseconds, with block finalization targeted at under one second.
Those numbers matter in context. Traditional financial exchanges operate at microsecond latencies, but most public blockchains still confirm transactions in seconds. Closing that gap is what separates a blockchain that AI agents can actually rely on from one that creates bottlenecks in automated workflows.
TxStream is one of the most structurally significant components of the new design. The system eliminates the public mempool — the waiting area where pending transactions are typically broadcast and visible to anyone before they are confirmed. Instead, TxStream routes transactions directly to block leaders, cutting latency and substantially reducing front-running opportunities, where sophisticated actors exploit advance visibility of pending orders.
Removing the public mempool is a notable design choice. It directly addresses one of decentralized finance’s most persistent problems — maximal extractable value exploitation — but it also changes the transparency dynamics that some users associate with public blockchains. The developers have not detailed how they plan to balance those trade-offs, and the implications for network openness remain to be seen at the testnet stage.
The roadmap does not start from zero. Earlier in 2026, BNB Smart Chain upgrades reduced block intervals from 750 milliseconds to 450 milliseconds and pushed benchmark throughput from roughly 2,800 transactions per second to 5,200. That near-doubling of capacity during the first half of the year provides the empirical foundation the team is using to project further gains in the second half and into 2027.
Speed is only part of the story. BNB Chain is also researching how to protect the network against threats that do not yet fully exist: attacks from quantum computers capable of breaking the cryptographic schemes that secure wallets today.
The approach under investigation uses account abstraction, a mechanism that would allow users to upgrade their wallet’s security model without changing their address. In practical terms, it means a network user could migrate to quantum-safe cryptographic standards as that technology matures, without abandoning their existing on-chain identity.
The developers were measured in their language here. “There’s no finish line here. Quantum computing will keep evolving, and so will our testing and research,” they wrote. “The point is that when it matures, BNB Chain’s infrastructure is already prepared.” That framing is honest about where the work stands: the quantum-resistant features remain firmly in the research phase, with no confirmed deployment timeline attached.
What makes the quantum work strategically interesting is not its immediacy — it has none — but its positioning. A blockchain that can credibly argue its architecture is already thinking about post-quantum security has a meaningful edge in institutional conversations about long-term infrastructure bets, particularly as governments and financial regulators increasingly raise quantum readiness as a compliance consideration.
It is designed to handle AI agent transactions, high-frequency trading, and automated payments at speeds comparable to traditional finance, targeting over 100,000 transactions per second and sub-50 millisecond confirmation times.
The testnet is expected by the end of 2026, and the mainnet is planned for early 2027, as outlined in BNB Chain’s H2 2026 technical roadmap.
TxStream removes the public mempool and sends transactions directly to block leaders, reducing latency and limiting front-running opportunities that typically arise when pending transactions are publicly visible before confirmation.
No. The new layer-1 will run alongside the existing BNB Chain, not replace it. Each network is intended to serve different use cases and transaction profiles.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.


