Goldman Calls AI Market a Rubber Band
Get the Finance newsletter
Daily finance — markets, central banks, M&A, the prints that move money. Free.
- Goldman Sachs strategist Rich Privorotsky described the AI market as a "rubber band," noting a widening divergence between underperforming hyperscalers (Amazon, Alphabet, Meta) and rising AI hardware stocks (Nvidia, TSMC) despite the hyperscalers' growing capex commitments.
- Privorotsky wrote that the market has spent recent weeks "ignoring almost every negative development for the AI capex trade," and warned that if frontier AI can be developed in the East at a fraction of Western cost, "the largest capital allocators are also the ones most exposed to over-investment risk."
- China's GLM-5.2 large language model was trained on 100,000 Huawei processors without Nvidia involvement, per trade publications cited by Goldman as evidence that software costs are falling outside the West.
- JPMorgan analysts Federico Manicardi and Victoria Campos said tech faces pressure from elevated expectations, euphoric sentiment, lack of free cash flow, CTO pushback on surging token costs, a pivot to lower-priced models, "draconian restrictions from Washington," and rising debt/equity supply.
- South Korea's Kospi fell roughly 10% since Monday's session after a record close the prior day, driven by a 12.31% drop in Samsung Electronics and a 12.47% drop in SK Hynix, while Nasdaq futures declined about 2.5%.
- Privorotsky added that "slightly less" capex is not embedded in any analyst assumptions, noting "the entire AI complex is priced for ever rising capex as inference demand grows."
Why it matters: Privorotsky's warning lands as tech already sells off: South Korea's Kospi fell 10% and Nasdaq futures dropped 2.5%, with Samsung and SK Hynix each losing over 12%. If hyperscalers cut AI capex from record levels, the hardware rally built on 'ever rising' spending loses its foundation.
