Sui wins the Move race through object dominance and TVL growth

Sui leads the Move ecosystem with $2.6 billion in total value locked as of October 2026, leaving Aptos with $58.62 million. This massive gap in liquidity follows a 59% monthly growth in Sui’s DeFi ecosystem throughout 2025. Sui handles over 800 real-world transactions per second using its Mysticeti consensus engine. This engine facilitates sub-second finality, which supports the network’s object-centric architecture.

Sui stores assets as objects with unique 256-bit identifiers and version numbers. Address-owned objects skip consensus, so transactions against them have the lowest latency. Shared objects require consensus and allow multiple users to access the same data. This design allows Sui to process transactions in parallel if they touch disjoint sets of state. In contrast, Aptos stores assets as resources inside accounts and uses Block-STM to compute execution ordering at runtime. Aptos handles assets via a 32-byte address that controls on-chain resources, though it later added Move Objects as an extension of the account model.

The Sui DeFi stack includes DeepBook, a native central limit order book. DeepBook v3 provides margin trading and referral-based income sharing. While DeepBook dominates contention by volume, it does not impose disproportionate sequential bottlenecks. Sui’s stablecoin market cap reached $921 million on October 7, 2026. This figure exceeds the stablecoin holdings of TON, Mantle, and Optimism.

Aptos targets sub-second latency with BFT optimizations

Aptos achieves sub-second end-to-end latency for all transaction types by reducing consensus latency from 5 to 3 message delays. Its AptosBFT v4 protocol relies on Jolteon and remains resilient to a third of validator stake failing. The network uses adaptive block proposals to include inline batches that have not yet received quorum certificates. This optimization speeds the time transactions enter proposals. Aptos also uses latency-aware peer selection, where fullnodes ping upstream validators to select nodes with the lowest latency and proximity to the validator set.

Aptos execution relies on Block-STM, a multi-threaded, in-memory engine that detects dependencies during execution. This allows Aptos to handle high throughput without developers declaring conflicts ahead of time. In a previewnet test, the platform reached a peak throughput of 30,000 transactions per second and processed 2 billion transactions in one day. Aptos provides irrevocable finality, meaning once a block is finalized, no party can revert it as long as BFT assumptions hold. This differs from Solana, where dishonest validators could cause an optimistically confirmed transaction to roll back.

Structural differences define the Move competition

Sui and Aptos share a common origin from the Diem project but maintain separate validator sets and token supplies. Sui uses a dialect of Move where transactions name the objects they touch. This requires the transaction author to declare the conflict upfront. Aptos uses a version of Move closer to the original iteration and infers parallelism at runtime. If a user writes to a popular resource on Aptos, that transaction will serialize inside Block-STM.

Sui’s Programmable Transaction Blocks allow up to 1,024 actions in a single transaction. A user can swap tokens, provide liquidity, and stake LP tokens in one atomic action. This reduces gas costs and improves the user experience. Sui also provides zkLogin, which lets users onboard using Google, Twitch, or Apple accounts. Aptos relies on sequence numbers to prevent replay attacks, while Sui uses object versions to ensure a user cannot spend the same owned object twice in conflicting ways.

The Sui network experienced a technical glitch that caused outages starting January 14, 2026. Developers attributed the incident to an error in the transaction scheduling logic rather than a DDoS attack. This follows a previous outage in November 2024 where the blockchain remained down for two hours. Will the object-centric model eventually encounter the same scaling bottlenecks as account-based systems?