Ant’s Falcon FX AI lands Citi, HSBC and StanChart

Artificial intelligence in the sphere of finance is moving from the stage of experimentation into the level of a component of the operation of global currency trade. Ant International announced on Thursday that it upgraded its Falcon forecasting program to version 2.0 and signed contracts with six major banks, which will use its model in their foreign currency trading activities.

According to Kelvin Li, Ant International’s general manager of platform technology, the company has named Citi, HSBC, Deutsche Bank, Standard Chartered, and Barclays as partners. The list is notable in relation to cryptocurrency and payments. Some of these banks are also developing tokenized deposit and digital payment technologies to facilitate round-the-clock cross-border transactions.

A claim of 60% cheaper hedging, and a jab at general AI

Li positioned Falcon as a tool for a job that general-purpose AI isn’t yet capable of doing well. “Precise forecasting can slash foreign exchange hedging and allocation costs by over 60%,” he remarked, while stating that general-purpose large models have “yet to achieve a universal breakthrough in the financial sector.”

Prior uses help back up that assertion. As stated in an announcement by Standard Chartered in August 2025, Falcon has achieved over 90% accuracy in forecasting and is already handling over 60% of its foreign currency conversions.

Falcon is a transformer-based model with nearly two billion parameters. Ant documented the 2.0 release this month in an arXiv technical report and published its code on GitHub, making the underlying approach more open to scrutiny than a typical proprietary bank forecasting system.

What Citi, HSBC and StanChart already built with Ant

The launch of the new model solidifies the partnerships that had been formed for more than a year.

In July 2025, Citi launched its Falcon pilot project alongside the introduction of its Fixed FX Rates product, which allows online retailers to maintain their exchange rates in more than 70 different currencies. According to Citi, combining the two products has helped its airline client save money on hedging, which Li put at approximately 30%.

Standard Chartered combined Falcon with its Aggregated Liquidity Engine, also known as SCALE. According to the bank, this method allows for forecasting of Ant’s currency exposures with over 90% accuracy and decreases the costs required for liquidity management by 50%.

HSBC’s partnership also extends beyond foreign exchange (FX) forecasting into tokenized currency. The bank created the Tokenized Deposit Service with Ant and processed a cross-border payment using the ISO 20022 messaging standard in 2025.

That overlap matters because AI forecasting and tokenized settlement address different ends of the same problem: predicting where liquidity will be needed and moving money there more efficiently.

Why the timing lands as FX hedging costs bite

The rollout comes as currency risk has become harder for global companies and investors to ignore. Global FX turnover averaged $9.5 trillion a day in April 2025, up 27% from three years earlier, according to the Bank for International Settlements. The BIS linked part of the surge to firms rushing to hedge dollar exposure following US tariff announcements.

Higher interest rates since 2022 have also made some hedges more expensive and left investors weighing the cost of protection against currency risk. Against that backdrop, technology that can materially reduce hedging and liquidity costs has an obvious appeal to multinational banks and their clients.

The concentration risk sitting underneath the hype

There is a regulatory catch. Financial Stability Board work summarized by the BIS in June 2025 warned that financial institutions relying on a small group of AI providers — or on similar models trained on similar data — could create system-wide vulnerabilities.

The concern is not simply whether one model gets a forecast wrong. If multiple large institutions rely on the same signals, errors or biases could encourage correlated decisions precisely when markets are under stress.

Falcon’s adoption therefore cuts both ways. Wider use could make FX hedging and liquidity management cheaper and more efficient, while also creating a shared technological dependency that regulators will want to understand.

Ant, meanwhile, is spending to expand that reach. The company raised $1.2 billion in an equity financing round last month, giving it additional capital as it pushes its financial AI deeper into global banking.