UBS has warned that the AI-driven technology rally has pushed excess returns for US momentum and tech stocks above the levels reached at the peak of the dotcom bubble in March 2000, raising the prospect that more hawkish central banks could trigger a repeat of the brutal 1999-2000 correction.
The note, from Michel Lerner, head of HOLT at UBS, stops short of calling the top but frames the question starkly: is the recent tech sell-off another dip to buy, or the beginning of a proper valuation derating back towards historical norms?
The bank's HOLT valuation framework highlights a key difference from the dotcom era: today's tech sector is generating real cash flows, with US tech's economic profit roughly ten times its 2000 level.
On conventional earnings-based multiples, valuations therefore look less extreme than they did during the internet bubble.
However, UBS argues that a different measure paints a more troubling picture.
Using its HOLT % Growth metric, which captures how much of a stock's value depends on future reinvestment rather than existing cash flows, the bank finds that the AI food chain is now priced to generate supernormal profits at levels never previously reached.
This implies that investors expect a disproportionate share of AI-related value creation to accrue to the enablers of the technology, such as chipmakers and infrastructure providers, rather than to the broader economy of companies that adopt it.
That assumption, UBS argues, is inconsistent with the expectation that AI will diffuse widely to drive productivity gains across multiple sectors.
Market leadership has also narrowed to a historically unusual degree.
In the US, only one in three stocks across the market has outperformed the index over the past three months, the lowest share since 1989, while AI-related positions rank among the most crowded long trades in UBS Quant Research's crowding data.
The rate environment adds a further dimension of risk.
Because so much of tech's intrinsic value is tied to future growth, the sector behaves increasingly like a high-duration bond, making it acutely sensitive to changes in discount rates.
UBS Evidence Lab data points to more hawkish central bank sentiment, and the bank warns that a renewed rise in inflation could reprise the 1999-2000 script in which the Federal Reserve raised rates aggressively into the peak of the bubble, triggering the subsequent crash.
Stretched valuations are not confined to tech: US cyclicals, including indirect AI plays, and aerospace and defence stocks on both sides of the Atlantic also screen as demanding on UBS's growth measures.
On the other side of the ledger, UBS identifies energy stocks and European value names as areas where valuation and momentum scores look attractive and crowding is low, with Shell, TotalEnergies, BP and Equinor all featuring in the bank's screen of stocks with strong fundamentals and undemanding implied growth expectations.
