Aptos has switched on an Encrypted Mempool feature on its mainnet, letting users keep pending transactions hidden from bots until block ordering is finalized — a direct answer to frontrunning and sandwich attacks that have cost traders tens of millions of dollars.

The Layer-1 blockchain announced the launch on Monday. Users can opt in, and once they do, a compatible wallet or app encrypts the transaction's contents on their device before it reaches the network. Assets, amounts, called functions and interacting contracts all stay under wraps. Sender address and gas fee remain visible.

Aptos said it built the feature for four groups: institutions submitting large transactions, traders trying to dodge MEV risk, AI agents, and builders.

"When you send a transaction on any major chain today, anyone watching the network can see the details before it executes — the function you're calling, the assets and amounts involved, and the contracts you're interacting with," Aptos said. "Bots use that visibility to trade against you or copy your strategy."

How big is the MEV problem?

The numbers explain the demand. Sandwich attacks cost Ethereum traders roughly $60 million over the 12 months through October 2025, according to Cointelegraph's analysis of EigenPhi data. On Solana, the picture is starker: Bitquery counted nearly 6.9 million sandwich attacks over an 11-month period, generating $29.7 million for bots and hitting about 2.2 million wallets.

That's the economic case for hiding transactions in the mempool — the waiting room where pending transactions sit before inclusion in a block.

How does the encryption work?

The mechanics are threshold-based. Validators agree on transaction order while blind to contents. Once ordering is final, they collectively decrypt the transactions, which then execute and record on-chain. No single validator can decrypt a pending transaction alone.

Because bots can't read pending orders, they can't insert trades ahead of or around them to profit at users' expense.

The protection isn't free. Per Aptos' developer documentation, encrypted transactions carry a minimum gas price at twice the network's base minimum. The system also adds around 20 milliseconds per block in tests involving 50 nodes, compared with the existing consensus implementation. For institutions moving size, that latency trade-off is likely acceptable; for high-frequency retail flow, it's a real cost.

How does Aptos compare with existing solutions?

Aptos isn't first here. Gnosis Chain launched an encrypted mempool using Shutter in July 2024. Ethereum's Flashbots Protect takes a different route, sending transactions privately to block builders.

Aptos claims one distinction: its existing validator set handles both ordering and decryption. Competing designs lean on a separate decryption committee or trusted builders — extra trust assumptions Aptos avoids.

The operational consequence matters for institutional adoption. Firms that have stayed away from transparent mempools — where competitor bots can front-run a large swap the moment it's visible — now have an opt-in mechanism with no third-party intermediary to trust beyond the validators themselves.

The move also fits Aptos' broader infrastructure push. The network recently committed $50 million to AI agent infrastructure, and agent-driven transaction flow is one of the explicit target audiences for encrypted submission.

Adoption now depends on wallet and application integration. Until major interfaces support client-side encryption by default, the feature stays a niche tool for users who know to ask for it.