Overview

Since the publication of our prior report on Aptos in September 2024, Aptos has demonstrated notable growth in ecosystem traction in combination with continued improvements on its technical infrastructure. Aptos has continued to innovate on the frontiers of its technical stack with research on Raptr, a theoretically optimal consensus mechanism, and Move 2, an improved rewrite of the original Aptos Move programming language to support higher expressivity of smart contract logic and operations with high volume data sets. Additionally, the introduction of Zaptos and Shardines offer continued architectural improvements through parallelization and pipelining, allowing Aptos to maximize resource utilization, improve bandwidth, and reduce end-to-end latency. These technical improvements position the chain to offer best-in-class infrastructure for high-volume financial activity, forming the foundation for a global trading engine. 

Evidence to this positioning, the ecosystem has maintained impressive growth. For June, key metrics like application revenue, DEX volumes, RWAs and stablecoin supplies, and TVL all sit at or near all time highs. With continuing improvements to the technical architecture and development functionality offered by Raptr, Shardines, Zaptos, and Move 2, Aptos seeks to separate itself from the pack of high performance L1s. As technical upgrades will continue to roll out on mainnet before year end, the network will further position itself as a leading venue to settle new, unique, and high-volume use cases and activity in an ultra-low cost environment.

Innovations on Aptos’ Architecture

Over the past year, Aptos has introduced a number of innovations on its technical architecture to further optimize for performance, latency, and bandwidth. These upgrades, including Raptr, Shardines, Zaptos, Move 2, and Shelby, further position Aptos as leading infrastructure for high-volume financial use cases and activity.  

Raptr

Aptos continues to iterate at the frontiers of consensus mechanism design with their research on Raptr. While yet to be integrated on mainnet, this new state machine replication protocol seeks near-optimal latency and throughput by utilizing 100% of the network’s resources, with the capacity to settle 260K transactions per second with latencies less than 800ms. 

Within this design, there are three core components to building a block on Aptos; data dissemination, consensus, and execution. Each of these functions are decoupled and performed by distinct processes; Quorum Store for data dissemination, Raptr for consensus, and BlockSTM v2 for transaction execution. Each of these processes are purpose built to be optimized for their intended functionality, and each process can run in parallel with one another on separate channels to prevent interference. 

Raptr’s design eliminates the data dissemination bottleneck of traditional leader-based designs. Instead of one leader broadcasting a fully constructed block to the network for consensus verification, which creates the bottleneck, data dissemination is decoupled from consensus, allowing the two processes to run concurrently and in parallel. 

When transactions are originated, validators broadcast batches of transactions through Quorum Store, the data dissemination layer. All validators effectively stream batches of transactions across the network asynchronously and in parallel, coupled with a proof of availability for the underlying data. As batches are propagated across the network, the leader can then propose a block that references the underlying batches. 

In traditional consensus models, a new round of consensus can only start once the previous round has completed. Instead, Raptr introduces *pipelining, *which allows multiple consensus rounds to run simultaneously as distinct instances. With partial block commits, validators vote on batches that are components of a block, without the need to vote on the fully constructed block. Leaders can propose a block or its component batches optimistically, prior to receipt of all of the batches, while other validators can vote on the proposed block if the relevant batches are held locally, or fetch the data from Quorum Store. A minimum batch age is implemented to ensure that a batch has adequately propagated across the network before a leader can include it. Raptr also employs multi-signature schemes for data signatures on consensus messages. Collected signatures are aggregated into smaller proofs for more efficient verification, removing the signature verification bottleneck. With the design enabled by Raptr, multiple instances of consensus verification can run concurrently and asynchronously on different batches as components of a block.  

As opposed to running data dissemination and consensus as discrete steps in a process, Raptr can run these processes concurrently. With the ability to execute data dissemination and consensus as distinct processes, latency is reduced.  

Once transaction batches are confirmed through consensus, they can be passed to the BlockSTM execution environment while consensus runs concurrently on the continuing stream of batches. With this, execution is decoupled from consensus, eliminating bottlenecks exhibited in legacy consensus designs.

Source: Aptos Labs

Raptr’s dynamic leader election mechanism reacts in real time to the network’s demands and performance metrics to most efficiently allocate node resources. This allows Raptr to maintain high performance under unstable or adversarial network conditions, such as spam or validator churn. In testing, Raptr outperforms the benchmarks measured against the current Aptos implementation, Shoal++, and Mysticeti. Given the performance degradation exhibited by Shoal++ under loads greater than 180K TPS, Aptos chose to not move forward with Shoal++, and instead continue researching the improvements presented within Raptr. Baby Raptr was rolled out on Aptos mainnet at the end of June, improving latency on validator finality by 20%. 

Shardines

Shardines is a parallel execution engine designed to significantly scale blockchain throughput by implementing sharded execution. Instead of processing transactions sequentially, Shardines divides the network’s workload across multiple independent execution threads, or “shards,” which run in parallel. This approach allows non-conflicting transactions to be executed simultaneously, rather than sequentially, which significantly increases throughput while maintaining deterministic outcomes. Shardines leverage deterministic parallelism, allowing it to guarantee the same final result regardless of the order in which different shards process transactions.

This results in a highly scalable architecture capable of settling one million transactions per second (TPS) under benchmark conditions and non-conflicting transactions. In tests involving 30 validator nodes, Aptos achieved approximately 1.1M TPS with near-linear performance scaling as more shards were added. 

Source: Aptos Labs

In testing a more adversarial environment with a higher share of conflicting transactions, such as users trying to access the resources of a small set of applications to execute a particular arbitrage or liquidation opportunity, the Shardines testing still demonstrated TPS in excess of 500k.

Source: Aptos Labs

This level of throughput positions Aptos as one of the most performant blockchains to date and opens the door for more demanding, high-volume use cases. By decoupling execution from bottlenecks traditionally seen in monolithic chains, Shardines enables Aptos to maintain low latency and high performance even under heavy load.

Zaptos

Zaptos is a low-latency execution and consensus architecture that minimizes a transaction’s time to finality. Zaptos addresses latency by optimizing how Aptos handles the distinct workflows of consensus, execution, and storage by running these processes concurrently instead of processing them sequentially. This pipelined and parallelized approach reduces confirmation time to sub-second levels, achieving finality significantly faster than traditional blockchain architectures.

Source: Aptos Labs

Testing results show that with Zaptos, the Aptos network can achieve sub-second finality while maintaining throughput of up to 20,000 TPS.

Source: Aptos Labs

Near-instant finality is particularly valuable for any latency-sensitive application, such as trading, liquidations, gaming, and payments. Raptr, Shardines, and Zaptos each optimize distinct components of the consensus and execution workflows to deliver an optimal implementation for a global trading engine.

Move 2

Move 2 iterates upon the first implementation of the Move language pioneered by the Diem project to offer new capabilities with its smart contract development. The language rewrite supports higher expressivity in smart contract logic, such as the ability to choose, pass, or execute particular logic at runtime.

The Dynamic Script Composer adds support for complex multi-call transactions to string together multiple concurrent or sequential contract calls. Interactions like swap, deposit, borrow, lend, and transfer can be bundled and executed in a singular atomic transaction. The script composer allows for the logic to be written by the developer in TypeScript, which is then compiled into the Move language. The produced Move payload is then executed on chain, ensuring that all or none of the intended actions are performed. The Dynamic Script Composer can support increasingly sophisticated and iterated transactions, unlocking new atomic transaction types, use cases, and strategies within DeFi.

Move 2 offers new standard libraries for data management, allowing developers to utilize large datasets onchain without necessitating complex data structures. The additions of the OrderedMap and BigOrderedMap data structures lower costs and latencies associated with building and interacting with high volume datasets. These new libraries allow different components of a large dataset to be updated in parallel without contention. Designed for high-throughput and high-volume use cases, storage read/write costs are reduced.

Below is a table showing the expected latency (in microseconds) of interacting with an OrderedMap dataset, when compared to the legacy SimpleMap data structure. As demonstrated, the OrderedMap structure offers a 100x improvement in latency on data sets with a higher number of elements.  

Source: Aptos Labs

BigOrderedMap is designed to scale arbitrarily with utilization of additional resources. For smaller data sets, the BigOrderedMap functions very similarly to OrderedMap, where the dataset is stored in a single resource, but scales resource utilization as data volume is added. This allows for scalable performance on operations with a higher volume dataset, whereas the legacy SmartTable structure would exhibit performance degradation as the volume of data scaled. For both OrderedMap and BigOrderedmap, operations to update different parts of the key-value map can be executed in parallel without contention, whereas the legacy data structures would present conflicts and require sequential processing. Coupled with this, the new data structures support added native functions to improve the manipulation of and operations with values in the data set.

Move 2 also improved account abstraction functionality with custom authentication, allowing for custom authorization logic for accounts. Now, an account can be authorized by a Move module with custom logic for multi-signatures, social recovery, oAuth, biometrics, and more.

For developers, the Move Prover supports formal specification and verification of contract builds. This allows builders to test for bugs and edge cases, and prove formally the absence of any bug. With formal verification of smart contracts, Aptos supports a more secure and deterministic environment for applications to avoid hacks and exploits. At large, the Move 2 rewrite improves developer experience and unlocks new use cases with smart contract logic and high volume dataset operations such as those associated with Dispatachable Fungible Assets, where tokens on Aptos can embed custom logic directly into transfers. This enables automated features like yield accrual, fee collection, and compliance checks without requiring separate smart contract interactions

Shelby

On June 24th, the Aptos Foundation unveiled Shelby, a decentralized hot-storage protocol for data availability, applications, and monetization. Shelby is designed with a mesh of nodes connected by a dedicated fiber-optic network provided by DoubleZero to support real-time, data-intensive applications. With configurable access controls, monetization logic, and sub-second latency, Shelby will service a multitude of use cases involving content creation, streaming, AI inference, data marketplaces, and more. Storage nodes are compensated for providing data, while creators retain full control and customization regarding the economics and utilization of their data.

While interactions with Shelby are coordinated through the Aptos blockchain, it is built to be chain agnostic and extensible to support use cases for applications across Ethereum, Solana, and more. The devnet for Shelby is expected to go live in Q4 2025, with a public testnet to be released thereafter. 

Aptos Ecosystem Activity

Year to date, the Aptos ecosystem has demonstrated strong growth. To better understand the applications at play within Aptos, we recommend reviewing our prior report here. Application revenue serves as an indicator of success for businesses within an ecosystem. Applications on Aptos earned ~$835k in revenue in June, up 15.7x from the start of the year. Application revenue has grown in excess of 30% consecutively month over month since September, 2024.

Within these applications, Thala and Merkle were historically the top earning, combined accounting for over 70% of application revenue. However, in June, the rise of the Hyperion DEX and the KGeN gaming ecosystem have taken material share in Aptos’ application revenues, with 10% and 26% share, respectively.

Despite the lackluster onchain activity across all networks exhibited in Q1 2025, Aptos has maintained the notable growth in volumes highlighted in our September report. Network activity has reached new heights with monthly spot DEX volumes reaching over $5.4B in June, a new all time high and up 280% from this time last year. The network achieved its all-time peak of 19.2K TPS, with infrastructure supporting 500 total nodes and 150 validator nodes. The Aptos Foundation has committed over $200 million in grants and investments to fuel growth within its DeFi sector.

DEX volumes on Aptos have gained significant share against those on Sui. While Sui historically demonstrated ~10x the DEX volumes of Aptos, this lead has closed dramatically in recent months. In June, DEX volumes on Aptos were 67% that of Sui’s $8B.

Since our prior report, Thala and Hyperion have risen to the top DEXs by volume on Aptos, flipping Cellana. Thala launched a second version of their DEX, ThalaSwap v2, towards the end of 2024, offering a number of improvements. V2 added support for 4 fee tiers with one as low as .01%, allowing the DEX to capture a large share of stable swap volumes. Additionally, v2 supports Aptos’s upgraded Fungible Asset Standard, offering improvements upon the legacy coin type and enabling them to take advantage of the Dispatchable Fungible Asset model in their new product, staked liquidity pool tokens (xLPTs). xLPTs internalize the accounting of the LP position’s yields and rewards, and are transferable and composable to be traded on DEXs or used as collateral on money markets. V2 also enabled flashswaps for MEV opportunities and multi-token pools for stablecoins. The v2 design now accounts for 80% of Thala’s DEX liquidity and 97% of Thala’s volumes. Thala now accounts for 25% market share of Aptos spot DEX volume.

Hyperion, launched in February 2025, has quickly risen to the top DEX on Aptos, settling $3.3B in volume in June with nearly 60% market share of DEX volumes. Hyperion functions as a hybrid orderbook-AMM, offering concentrated liquidity, limit orders, and single sided liquidity provision.

Stablecoins

With support for canonical USDT and USDC, coupled with Circle’s CCTP, the supply of stablecoins on Aptos has grown to $1.35B, up from $90M this time a year ago. Within this, USDT accounts for $985M while USDC accounts for $271M. Additionally, Ethena has taken their sUSDe and USDe products to Aptos using the LayerZero Omnichain Fungible Token (OFT) standard, with now $50M in sUSDe circulating on the chain.

BTCFi

Utilization of BTC on Aptos has grown notably over the past year, with a number of players offering bridged or wrapped versions of BTC to be used on the chain, with a total supply eclipsing $450M. Leading this sector is Echo Protocol, which accounts for nearly $300M in aBTC on Aptos. Echo offers a permissionless two-way bridge that locks native Bitcoin on the source chain and mints aBTC on Aptos. Subsequently, users can access yield opportunities on this asset within DeFi, whether it’s providing liquidity on DEXs, lending, or borrowing against the asset on money markets as collateral.

Additionally, leading exchange OKX has launched a wrapped Bitcoin product on Aptos, xBTC, with now over $50M in supply. xBTC offers institutional-grade custody on the underlying BTC deposits, held in cold storage. A growing supply of BTC further supports maturation in the Aptos DeFi ecosystem through deepening DEX liquidity for this asset or offering demand to borrow stables against this asset as collateral. 

DeFi

The growth in stablecoin supply has been a tailwind for DeFi on the chain, supporting growth in deposits and borrows on money markets. Aries Markets remains the largest application for lending and borrowing on Aptos, with $360M in assets supplied for lending and nearly $200M in active loans. 

On Aries, APT and its liquid staking tokens account for 33% of the collateral balance, while stablecoins account for 60%, suggesting growing demand to borrow against the native gas token and its LSTs.

Stablecoins account for about 70% of the assets supplied for lending operations. Depositors of these stablecoins are earning 3.4%-4.0% APY in APT incentives, on top of the ~4.5% lending APY, for a combined APY of ~8%.

After a few months of running a testnet instance, Aave recently passed an AIP to activate a v3 deployment on Aptos mainnet. The initial deployment will go live shortly, hosting markets for APT, USDC, USDT, and sUSDe. This represents Aave’s first non-EVM deployment, and will further grow competition within Aptos’ market for lending and borrowing.  

Transaction Fees

Historical transaction fees evidence Aptos’ efficient utilization of network resources and cheap environment for transactions. Median daily transaction fees are as low as $.00008, while average transaction fees are as high as $.0005. With this, current fees on Aptos are 90% cheaper than fees on Solana, which currently run a daily median of $.001 and an average of $.01. Additionally, Aptos’ environment is cheaper than Sui, where average transaction fees are running at $.007.

Conclusion

Aptos exhibits increasing utilization of its high performance environment, evidenced by DEX volumes and application revenue growing notably month over month and each having reached new all time highs in June. While green shoots abound, utilization on the network remains far below its theoretical capacity. Aptos is further positioning its technical stack as best-in-class amongst high-performance L1s to settle high-volume financial use cases. With a consensus design approaching theoretically optimal, a growing ecosystem of core DeFi primitives with traction, and deepening stablecoin liquidity, the foundation is set to support much higher utilization of the network’s resources. With Raptr coming to mainnet, it remains an open question if the network can separate itself from the pack and garner the levels of traction and utilization it’s perfectly designed for. At large, technical upgrades, growing DEX volumes, and an increasing supply of stablecoins, BTC, and RWAs on the chain further position it to grow into the global trading engine it has set out to become. 

Disclosure: This research report has been funded by the Aptos Foundation. By providing this disclosure, we aim to ensure that the research reported in this document is conducted with objectivity and transparency. Blockworks Research makes the following disclosures: 1) Research Funding: The research reported in this document has been funded by the Aptos Foundation. The sponsor may have input on the content of the report, but Blockworks Research maintains editorial control over the final report to retain data accuracy and objectivity. All published reports by Blockworks Research are reviewed by internal independent parties to prevent bias. 2) Researchers submit financial conflict of interest (FCOI) disclosures on a monthly basis that are reviewed by appropriate internal parties. Readers are advised to conduct their own independent research and seek the advice of a qualified financial advisor before making any investment decisions.