UBS Says AI Stocks May Have to Play by Normal Valuation Rules as Rates Stay High
UBS says AI stocks have not been priced by normal valuation rules, and may now have to be, because interest rates are higher. The view appears in a MarketWatch report published on 27 October 2026.
The argument rests on duration, a measure of how far away a company's cash flows sit and how sensitive their present value is to the discount rate, the annual return investors demand for waiting. A growth stock's value depends heavily on earnings expected far in the future. A higher discount rate cuts the present value of those distant cash flows more than it cuts near-term ones. When investors were willing to look past the next few years, that sensitivity mattered less. If rates stay elevated, it matters a great deal more for companies whose case rests on profits that have not yet arrived.
The bank's caution is not new. In August 2025, Michel Lerner, head of the HOLT analytical service at UBS, warned that the valuation of AI stocks was closing in on dotcom levels. Lerner said AI stocks leave little room for cash flow disappointments, given uncertainty about the return on massive capital expenditure. Capex, the spending on data centres, chips and other long-lived assets, has to earn a return before those cash flows can justify the price. The question is therefore not only whether AI spending produces revenue, but how much of that revenue arrives in time, and at what margin, to support price-to-earnings multiples set in a lower-rate environment.
Market behaviour has tracked those doubts. Reuters reported that the S&P 500 and Nasdaq ended sharply lower amid revived AI doubts, in a session covered in its market report. Breadth data from the same session tells more than the index moves alone. Breadth measures how many individual stocks are moving in the same direction. The S&P 500 recorded 40 new 52-week highs against four new lows. The Nasdaq Composite recorded 82 new highs and 342 new lows. In other words, the S&P 500 saw roughly ten stocks reach new highs for every one hitting a new low, while the Nasdaq Composite saw more than four stocks hit new lows for every one reaching a new high.
In my reading, a narrow index can carry a broad market higher, but the Nasdaq figures point the other way. When the tech-heavy benchmark shows this kind of internal deterioration, the weakness is spread across many names rather than concentrated in a few.
That distinction matters for anyone holding index funds. Passive exposure to a cap-weighted benchmark means exposure to whichever sectors have grown largest, and AI-linked companies carry a heavy weight in several of them. A valuation reset in that group reaches savers who never chose an AI position, through the index they did choose.
UBS's response is more measured than a call to exit the theme. A UBS strategist pointed to stocks relatively untouched by AI, including the U.K. That is a relative-value argument: if AI-exposed names are priced on expectations that may not be met, companies with less exposure to that narrative may be priced on more conventional fundamentals. Whether that holds depends on what those companies earn, which is a separate question from what a strategist prefers.
In my view, the bank's framing is also a comparison, and comparisons have limits. Calling a valuation level "closing in on" an earlier peak says nothing precise about timing, and the gap between a warning and a decline can be wide and unpredictable. UBS has argued for more than a year that AI valuations leave little margin for error. Its latest message is that higher rates shrink that margin further, and that market pricing has not yet fully reflected it.
For readers who want a test, the useful question is whether AI capex converts into cash flow at the rate current multiples assume, and whether the discount rate used to value that cash flow stays where it is. Neither question is settled. Until one of them is, the gap between index-level resilience and the breadth readings from the latest session is worth watching.
