TechFlow News, May 30: Peter Berezin, Chief Global Strategist at BCA Research, stated that the core risk of the current AI rally does not stem from overvaluation but rather from overly optimistic market expectations for future earnings growth—making it more characteristic of an “earnings bubble.”
He noted that, unlike traditional valuation bubbles such as the dot-com bubble, price-to-earnings (P/E) ratios in AI-related sectors—especially semiconductors—have not experienced extreme expansion. However, earnings expectations for companies in these sectors are rising rapidly, and such growth may not be sustainable over the long term.
Citing the real estate and banking sectors before the 2008 financial crisis, Berezin explained that valuations for related companies then appeared reasonable, yet their profit growth rested on unsustainable foundations; once earnings failed to materialize, stock prices collapsed swiftly.
He further observed that global semiconductor sales have already entered a parabolic growth phase. Although there are currently no clear signs of weakening AI-driven demand, history shows that all bubbles ultimately end.
Regarding investment strategy, Berezin advised investors not to rely excessively on Wall Street analysts’ earnings forecasts. Historical data indicates that stock prices often peak before earnings expectations are formally downgraded; thus, waiting until EPS forecasts begin to be revised downward before exiting positions is typically too late.
He emphasized that, at this stage, monitoring demand-side indicators—such as AI infrastructure investment, data center construction, GPU demand, and semiconductor sales—may offer greater forward-looking insight than focusing solely on earnings forecasts. These metrics could serve as critical barometers for gauging where the AI rally stands.
