Aptos Foundation and Aptos Labs have committed $50 million to develop infrastructure built for autonomous AI agents on the Aptos Layer-1. The initiative targets high-frequency on-chain trading and agentic workloads, funding first-party products, protocol upgrades, research and a partner investment program designed around sub-second execution, encrypted mempools and confidential data storage.
The package is being framed as a full-stack strategy rather than a single product push. Aptos will direct capital toward in-house applications, core protocol improvements and third-party teams, with existing projects such as Decibel and Shelby already positioned inside the broader roadmap.
Aptos Builds for Machine-Speed Markets
Decibel, an on-chain perpetuals exchange, launched on mainnet in February 2026 and has reported more than $1 billion in trading volume. Its inclusion in the funding strategy gives Aptos an early trading venue for automated market activity, especially as the network courts more machine-driven execution.
Shelby, a decentralized data storage protocol built for AI workloads, addresses another part of the stack. Autonomous agents need persistent data storage and reliable state management, making storage infrastructure as important as trading speed for agentic systems.
Aptos described the strategy in direct terms: “Markets are moving onchain. Machines are becoming the primary participants in them. The $50 million committed across the stack is how Aptos gets there.” The statement places machine participants at the center of Aptos’s market thesis, not as a secondary use case.
Confidentiality is also a core design priority. Sherry Xiao, Founding Engineer at Aptos Labs, said Confidential APT could help businesses protect on-chain salary payments, treasury movements and trading strategies from public visibility, highlighting the need to shield sensitive financial activity from open-ledger exposure.
Confidential Execution Targets MEV Risk
Aptos is treating MEV and front-running as structural problems for automated trading. Analysts cited in the company’s materials estimate that roughly $50 million per quarter is extracted from predictable agent-driven flows, a figure used to justify stronger privacy and execution protections.
The proposed mitigations include encrypted mempools, sub-second finality and FIX connectivity. Together, those tools are intended to support reliable high-frequency execution for institutional counterparties while reducing the visibility that allows bot operators to exploit predictable orders.
The risks Aptos is targeting are not limited to trading. Public on-chain activity can reveal salary payments, treasury movements and strategy execution, while agent models also face data availability and persistence risks if storage and state infrastructure are not hardened.
Integration risk is another practical concern. Neobanks, custodians and wallet providers connecting to agent rails will need secure interfaces between traditional accounts and on-chain execution, especially if autonomous systems begin routing higher-value financial activity.
Project leads at Aptos have emphasized continuous autonomous operation as a design objective. Nikki Baird, VP of Strategy and Product, said agents are the next level of AI capability the company is developing for customers, pointing to a product direction that balances openness with confidentiality.
The commitment changes how Aptos should be monitored. If the network attracts agentic volume, it could concentrate predictable transaction flow and the MEV opportunities attached to it, even as encrypted infrastructure aims to reduce extraction.
The key indicators will be on-chain agent volume, encrypted mempool adoption and Decibel’s trade composition. Those metrics will show whether Aptos’s $50 million infrastructure push changes MEV dynamics or simply shifts extractable value toward new market participants.