The tokenized US Treasury market has reached $14.72 billion across 82 assets, accounting for nearly half of all tokenized real-world assets onchain. At the same time, 74% of treasury functions are now actively using or expanding AI, according to PwC's 2025 Global Treasury Survey. Institutional capital is already allocated to onchain yield instruments at scale, and the operational infrastructure to manage that capital autonomously is closer than most businesses realize.
36% of corporate treasuries still manage FX exposure manually. This is just one of many existing operational gaps AI agents are being built to close.
A treasury agent executing yield routing or FX rebalancing across multiple chains cannot function reliably when the same dollar exists as different token variants with different contracts and different bridge exposures on each chain.
USDT0 gives treasury agents one asset with one set of rules across every chain they operate on, with nearly $100 billion in lifetime volume and no bridged variants introducing inconsistency into the execution loop.
More recently, Fireblocks' May 2026 report on agentic finance named autonomous treasury management as one of the defining institutional frontiers of the year. The world is quickly moving towards agents that reason across live market conditions, rebalance positions in real time, manage cross-border settlements outside business hours, and deploy idle capital into yield-bearing instruments without human intervention at each step. USDT0 is the liquidity and settlement infrastructure that makes this AI-enabled treasury model work.
The phrase “thinking machines” often leads us to believe that artificial intelligence is a tool that we can use in partnership or in place of our own ability to strategize and reason. But when Harvard and Boston Consulting Group ran 758 consultants through a task that required genuine strategic judgment, the ones using GPT-4 were 19% less likely to reach the correct answer than colleagues with no AI at all. Aim the same tool at well-scoped work and the numbers inverted to roughly 25% faster and around 40% higher quality. AI earns its keep when it is pointed at a specific outcome. Used in place of judgment, however, it makes the work measurably worse.
Capital management is a forum where this advantage shows up clearly. Moving money is rule-bound, repeatable, and measured against a single correct outcome, the exact profile of work where AI delivers. Among the highest-value early enterprise deployments are agents handling high-frequency, low-discretion tasks, such as vendor payments, inter-entity liquidity transfers, FX rebalancing across subsidiary accounts. Each runs on a repeatable cycle an agent can execute more precisely and more continuously than any treasury team.
The scale of manual overhead still in place makes the opportunity concrete. PwC found that 36% of treasury teams still manage FX exposure manually, and that manual forecasting data collection runs at 38% even among companies with revenues above $10 billion. A Citi and Ant International pilot on AI-enabled FX hedging reduced one client company’s hedging costs by 30% while keeping forecasting accuracy above 90%. With results like these, it’s not surprising that 54% of CFOs plan to deploy AI agents this year, according to Deloitte's Q4 2025 CFO Signals Survey.
The proof of concept for autonomous onchain execution already exists. In December 2025, BMW Group executed the first fully pre-programmed, automated EUR to USD onchain FX transaction through JPMorgan's Kinexys platform, completed outside traditional settlement windows and without manual intervention. That said, any business looking to scale its treasury automations beyond a walled garden requires borderless settlement infrastructure capable of supporting continuous agentic execution.
The IMF's April 2026 note on agentic payments flags the fact that while AI agents operate on probabilistic reasoning, the settlement infrastructure beneath them must be deterministic, with irrevocable legal finality at each execution step. When the settlement asset is fragmented, the agent is forced to apply probabilistic reasoning to something that is supposed to be a confirmed fact, and continuous autonomous execution becomes unreliable.
Simply put, this means a settlement asset has to behave the same way, under the same rules, on every chain, in order for continuous treasury operations to be possible. Most corporate treasury workflows already touch multiple chains across yield venues, payment rails, and counterparty environments. An agent moving capital across those environments cannot function reliably if the dollar it holds looks different depending on where it lands.
The practical benefits of USDT0 for corporate treasury are well-documented. A single unified supply across chains means one reconciliation flow, one set of procedures, and no wrapped variants introducing bridge dependencies or inconsistent contract behavior. For treasury teams managing liquidity across multiple onchain environments, that foundation already changes what is operationally possible.
For treasury agents, that architectural consistency is what makes continuous autonomous execution reliable. Every settlement instruction resolves against the same asset, governed by the same contract logic, across every chain the agent touches. A yield routing decision moving capital from Tempo to Arbitrum settles against the same dollar on both sides. An FX rebalancing instruction touching four chains in sequence requires no separate accounting for each chain's variant of the same asset.
The same infrastructure also gives treasury agents access to programmable gold as a reserve asset. XAUt0, the infrastructure that brings Tether Gold to every network, operates on the same borderless standard as USDT0. This means an agent managing treasury positions across chains can hold and move a gold allocation with the same contract consistency it applies to dollar settlement. For treasury operations requiring reserve diversification outside fiat, that capability sits within the same infrastructure stack rather than requiring a separate system.
That consistency carries into compliance as well. The IMF note flags authorization traceability as a primary risk in agentic financial systems, specifically the ability to reconstruct what an agent did and why. A single canonical supply gives compliance and audit teams one object to track across every chain the agent operates on, rather than reconciling positions fragmented across multiple token variants with different contract addresses and different onchain histories.
Autonomous treasury management is closer than it looks. The onchain yield instruments exist, agent frameworks are in production, and institutional upside is clear. What has lagged is clarity on which infrastructure actually meets the settlement requirements at these systems' scale.
USDT0's architecture was built around exactly the properties autonomous treasury demands: one asset, one supply, consistent behavior across every chain an agent operates on. As treasury teams begin deploying agents for the operational work that never stops, the settlement layer underneath those agents is what enables reliable, multi-jurisdictional treasury automation.