Equity markets retreated Wednesday after inflation data came in above forecasts, lifting Treasury yields and deepening unease over the cost of borrowing, while large financial institutions were simultaneously weighing both the potential gains and the pitfalls tied to the expanding wave of AI investment.

The index slipped 1.1%, putting it on course for a fifth down session in its last six trading days — a run that traces back to when it last hit a record high. By 1:14 pm Eastern time, the Dow Jones Industrial Average had shed 171 points — a 0.3% retreat — while the Nasdaq Composite pulled back by 1.6%.

Inflation in the US ticked up to 3.7% on an annual basis last month, outpacing both the 3.4% recorded in August and the 3.6% that analysts had forecast. Figures released on inflation pointed to the possibility that price pressures may not ease soon, with higher energy costs linked to the conflict involving Iran identified as one contributing factor.

Yields on the 10-year US Treasury note settled at 5.23% by the close of trading, having touched an intraday peak of 5.32% at one point during the session. When bond yields climb, the added expense of borrowing and the improved appeal of fixed-income instruments can put downward pressure on equity valuations.

Bank executives at large US institutions pointed to AI infrastructure investment as an area of opportunity, while noting that questions around demand levels, the cost of financing, and the prospect of returns introduce potential exposure. Speaking at a Goldman Sachs event, CEO David Solomon told attendees that how capital requirements and computing needs across the broader AI landscape would develop over time remained an open question, adding that disciplined judgment in both financing choices and risk oversight would be essential as the technology continues to take shape.

Unrelenting pressure on computing capacity, Fraser told investors, continues to push spending across the broader AI ecosystem higher — a trend the Citigroup chief executive attributed to a supply gap that has yet to close. Substantial investment flowing toward data centres and AI computing infrastructure is carving out a distinct lending opportunity for banks prepared to finance large-scale build-outs.

Borrowing costs at higher levels have the potential to push up expenditures on a given project, narrow the margin of returns, and place added strain on debt obligations. Framing the expansion of AI infrastructure as a notable economic driver, Wells Fargo's chief executive Charlie Scharf pointed to the heavy debt financing involved and the downward pricing pressure felt across certain industries and supply chains as factors that introduce both promise and potential strain.

Across the banking sector, AI tools are being woven into day-to-day operations with an eye toward lifting output and trimming expenses, though how much institutions actually gain from those efforts differs from one lender to the next. Citi's product development pace has quickened by around 40% under AI-assisted workflows, while coding tools have trimmed the bank's technology vendor outlay by roughly 30%, according to Fraser.

Whether the productivity gains from AI translate into a return on the substantial capital being deployed is a question shareholders and analysts are actively tracking. Worries about vulnerabilities in digital infrastructure and the potential for AI-driven systems to act without human direction have continued to build.

Stubbornly elevated inflation, bond yields that have stayed high, and unresolved questions about whether AI spending will generate adequate returns have all piled onto the pressures already bearing down on Wall Street, where stretched equity valuations leave little buffer should corporate earnings or productivity improvements fall short of expectations.