Ant International has launched an upgraded artificial intelligence model designed for financial forecasting, signing partnerships with six major global banks as lenders accelerate efforts to deploy specialized AI systems to manage liquidity, foreign exchange and other balance-sheet risks.

The Singapore-based fintech company on Thursday unveiled the Falcon Time-Series Transformer Model 2.0, an upgraded forecasting system aimed specifically at financial applications. Kelvin Li, Ant International’s general manager of platform technology, said the model has been adopted through partnerships with six major banks, including Citi, HSBC, Deutsche Bank, Standard Chartered and Barclays.

The partnerships underline the growing push by financial institutions to move AI beyond customer-service applications and into core banking functions where more accurate forecasts can directly affect trading, treasury management and capital allocation.

Financial institutions manage large and constantly changing pools of cash across currencies, markets and jurisdictions. Errors in forecasting can leave banks holding excess liquidity that earns little return or force them to obtain funding at higher costs. More accurate predictions of cash flows, foreign exchange movements and liquidity requirements can therefore produce substantial savings.

Li said Falcon 2.0 is designed specifically for such financial scenarios and has advantages over general-purpose AI models.

General-purpose large models have “yet to achieve a universal breakthrough in the financial sector,” Li said, arguing that specialized systems can be better suited to highly structured financial data and forecasting requirements.

Ant International said the model’s forecasting capabilities can reduce foreign-exchange hedging and allocation costs by more than 60%. Such savings could be significant for banks and multinational companies with large cross-border exposures, although the actual benefit will depend on the quality of underlying data, the markets covered and how institutions integrate the technology into their existing risk-management systems.

The launch comes as banks globally increase spending on AI amid pressure to improve productivity and automate complex processes. Financial institutions have been among the largest corporate adopters of AI, using the technology for fraud detection, risk assessment, trading, compliance, customer service and software development.

The next phase is focused on specialized systems capable of operating within tightly controlled financial environments. Unlike consumer-facing generative AI, treasury and risk-management applications require reliable numerical forecasting, explainability, data security, and strict controls over how models influence financial decisions.

Time-series models are relevant to these applications because they are designed to identify patterns and relationships in sequential data. In banking, that can include historical cash flows, currency movements, interest rates, transaction volumes, and other market indicators.

Ant International’s strategy puts it in competition with both established financial-technology providers and technology companies seeking to supply AI infrastructure to banks. The company’s focus on specialized financial models could also allow it to target a market where institutions are reluctant to rely entirely on general-purpose AI because of the consequences of inaccurate outputs.

The move is part of Ant International’s broader international expansion. The company, the overseas affiliate of Chinese fintech group Ant Group, raised $1.2 billion in its latest equity fundraising last month as it seeks to expand its business.

The capital raising gives the company additional resources as competition intensifies for enterprise AI contracts and financial institutions become more selective about the systems they adopt.

For global banks, the appeal of specialized AI is increasingly tied to measurable financial outcomes rather than the technology’s novelty. Forecasting improvements that lower hedging costs, optimize liquidity positions, or improve capital allocation can provide a direct return on AI investment.

The challenge is that financial markets are highly dynamic. Models trained on historical patterns can struggle when market conditions change abruptly, meaning banks are likely to require continuous monitoring, human oversight, and safeguards around AI-generated forecasts.

Ant International’s latest model therefore marks a broader shift in financial AI: from general-purpose experimentation toward specialized systems designed to solve specific, high-value problems inside banks.

As banks deepen their use of AI in treasury and risk management, the ability to demonstrate measurable improvements in forecasting accuracy and operating costs is expected to become a key factor in determining which AI platforms gain widespread adoption across the financial sector.

U.S. Treasury yields fell sharply on Wednesday after the Treasury Department moved to increase its purchases of longer-dated government bonds, offering investors some relief after a sustained selloff pushed long-term borrowing costs to levels that are challenging for the U.S. government and private sector.

The yield on the 30-year Treasury fell nearly 10 basis points to 5.188% before recovering to around 5.208%. The move followed the Treasury’s announcement that it would double the size of liquidity-support buyback operations for longer-dated securities from $2 billion to at least $4 billion per operation. The programme will cover the 10-to-20-year and 20-to-30-year segments of the Treasury curve from September 9 through November 4.

The announcement also lifted risk assets. The Nasdaq Composite rose 0.4%, while the dollar weakened, with the dollar index falling 0.7% to 98.95. The reaction underscored how closely investors are watching the Treasury market for signs that policymakers are prepared to intervene as borrowing costs rise.

Relief Not for So Long

But the relief may prove temporary. The Treasury’s move can improve liquidity and reduce the immediate supply of some longer-term securities, but it does not address the underlying forces driving yields higher: enormous government borrowing requirements, uneven demand for long-dated debt and growing competition for capital.

Ryan Swift, chief U.S. bond strategist at BCA Research, said the announcement had two important effects. First, it signaled that Treasury officials were increasingly concerned about the rise in long-term yields. Second, it demonstrated the department’s effort to manage the maturity structure of its borrowing by relying more heavily on short-term Treasury bills while keeping longer-term coupon issuance relatively stable.

“The data does not indicate that rising long-maturity yields were driven by a deterioration of liquidity,” Swift said. The move instead showed that the Treasury “is sensitive to the increase in yields and is willing to take steps to try to mitigate it.”

The strategy, however, has a built-in limit. Shifting more borrowing toward short-term bills can reduce pressure on the long end of the curve, but it also increases the supply that investors must absorb at the front end. Swift said the market would ultimately determine how far the Treasury could pursue that approach.

“The market will be the ultimate constraint on how far the Treasury can shift its issuance away from long-dated coupons and into T-bills,” he said.

That means Wednesday’s rally should not be interpreted as evidence that the fundamental bond-market problem has been solved. Swift said the Treasury’s measures were likely to move yields only temporarily because the government’s ability to alter the maturity structure of its debt is limited.

“The Treasury’s toolbox is limited to changing the maturity structure of the debt,” he said. “If it does anything too extreme, then the market will push back and force them to reverse course.”

The bigger problem is the sheer amount of debt that needs to be financed.

Joseph Purtell, senior vice president and portfolio manager at Neuberger Berman, questioned whether an additional $2 billion per buyback was sufficient to counter the broader supply-demand imbalance that had pushed long-term yields higher.

“For us, in thinking about the context around this, is an extra $2 billion per buyback really worth a full 9 basis points, relative to supply/demand mismatch that got us into this mess in the first place? No,” Purtell said.

He said the Treasury’s intervention would probably help in the short term by signaling that officials have a level of yields they are uncomfortable with. But he warned that it would not resolve the fiscal problem.

“It’s going to help today,” Purtell said. “But it doesn’t address the glaring supply issue. Deficits show no reason to go down.”

That is the central issue confronting the bond market. Treasury buybacks can influence liquidity and the distribution of debt across maturities, but they cannot reduce the federal government’s underlying deficit. Unless Washington narrows the gap between spending and revenues, the government will continue to require large amounts of financing, leaving investors to determine what yield they require to absorb that debt.

The implications extend well beyond government finance. Treasury yields serve as a benchmark for borrowing throughout the economy, influencing corporate debt, mortgages and other forms of credit. A sustained rise in long-term yields can therefore tighten financial conditions even if the Federal Reserve lowers short-term interest rates.

The bond market’s recent behavior is seen as an indication that investors are increasingly demanding compensation for holding long-duration U.S. debt. Michael Lorizio, head of U.S. rates and mortgage trading at Manulife Investment Management, said the Treasury’s decision reflected “an inconsistent amount of demand, especially in off-the-run securities in the very back end of the Treasury curve.”

The timing matters because yields have reached levels that can have a material effect on the government’s interest bill. If higher borrowing costs persist, refinancing existing debt becomes more expensive while new deficits must be funded at higher rates.

That creates a potentially damaging feedback mechanism: larger deficits require more borrowing, greater borrowing increases the supply of government securities, weaker demand can push yields higher, and higher yields increase the cost of servicing the debt. Those additional interest expenses can then contribute to still larger deficits.

Several analysts quoted by Reuters therefore cautioned against viewing the Treasury intervention as a lasting solution.

Brian Jacobsen, chief economic strategist at Annex Wealth Management, described the market response as “a temporary salve.” He noted that the announcement illustrated a broader shift in which the Treasury is attempting to influence financial conditions through the composition of its borrowing rather than through a reduction in the government’s financing needs.

Peter Cardillo, chief market economist at Spartan Capital Securities, was even more blunt, calling the move “a gimmick” that would probably work for a period before renewed selling pressure returned.

“What this does is it relieves short-term pressures in the long end of the market,” Cardillo said. “It will probably work for a while until the vigilantes are back again.”

The reference to bond “vigilantes” rings a bell. It describes investors who demand higher yields when they believe governments are pursuing unsustainable fiscal or economic policies. If investors become convinced that Washington is unwilling or unable to bring deficits under control, the market can impose discipline through higher borrowing costs.

The Treasury is also operating in an environment where the Federal Reserve is unlikely to provide an easy backstop. Swift noted that the Fed has the capacity to purchase Treasury securities across maturities, but its current direction is toward shrinking its balance sheet rather than expanding it.

This leaves the Treasury with a narrower set of tools. It can alter the maturity of new issuance, adjust auction sizes and conduct buybacks, but it cannot permanently suppress market-determined yields without addressing the fiscal fundamentals.

Thomas Simons, chief U.S. economist at Jefferies, warned that the abrupt announcement could also damage the Treasury’s credibility with investors.

“If the aim of this is to reduce term premium or long-end yields, I think this is an incredibly short-sighted strategy,” Simons said. “I don’t think that they appreciate what kind of premium is built into yields that is related to the idea that we’re not surprised by things.”

His concern goes to the importance of predictable Treasury communication. Investors price government debt partly on expectations about future issuance. Sudden changes can alter those expectations and potentially increase the risk premium investors demand.

Yet the Treasury’s decision also shows that officials are increasingly sensitive to the economic consequences of yields approaching 5% and beyond.

René Albrecht, senior analyst at DZ Bank, said policymakers are concerned about the effect of high long-term yields on both government finances and private-sector borrowing.

“I think they fear the pain of 5% or higher yields on the long-end, not only because it raises the interest rate costs for the government but also for the private sector,” Albrecht said.

The broader backdrop makes the situation more difficult. Investors are confronting persistent fiscal deficits, elevated government debt, inflation uncertainty and the financing demands of the artificial intelligence boom. Technology companies and data-center operators are committing enormous sums to computing infrastructure, increasing competition for capital at the same time governments are issuing large quantities of debt.

Analysts have warned that competition could become more important if AI infrastructure investment remains elevated. Private-sector demand for capital does not necessarily crowd out Treasury demand directly, but it adds another major source of financing needs to an already capital-intensive global economy.

The bond market is therefore confronting a structural question rather than simply a temporary liquidity problem: who will ultimately absorb the enormous volume of U.S. government debt?

Purtell framed the issue directly: “There are structural fiscal issues that haven’t been addressed for a very long time. The more interesting battle will be addressing longer-term issues, and who will be there to underwrite that debt.”

Jeremy Stretch, head of G10 FX strategy at CIBC, said the Treasury’s intervention demonstrated that officials recognized the risk that the bond-market selloff could spill into other asset classes.

“There are still concerns about inflation, the debt profile in the G4, the impact of AI,” Stretch said. But the announcement also showed that “the U.S. Treasury recognizes what is going on the bond market and is prepared to adjust policy in order to limit pressures on the market.”

The immediate focus for investors will be on the Treasury’s intervention’s capacity to keep long-term yields below the 5% threshold and whether demand for U.S. government debt improves. Analysts see a sustained decline easing pressure on mortgages, corporate financing and equity valuations. A renewed rise, however, would signal that investors are demanding a larger premium to finance Washington’s fiscal requirements.

Wednesday’s rally therefore provides breathing room, not a resolution. The Treasury has shown it can influence the bond market, but the market ultimately retains the power to determine the price of government borrowing.