The AI market has become a 'rubber band' - the question now is how far it can stretch, says Goldman strategist

By Nora Redmond

The AI market is currently priced for increasing capital expenditure as demand for its execution grows, per Goldman Sachs.

The AI market has become a rubber band, with a growing divergence between so-called hyperscalers and the companies selling semiconductor chips as software becomes cheaper to develop outside the West, according to Goldman Sachs.

"Whatever the proximate reason, it increasingly feels like the market has spent the past few weeks ignoring almost every negative development for the AI capex [artificial intelligence capital expenditure] trade," Rich Privorotsky, strategist at Goldman's global banking and markets division, wrote in a note on Tuesday.

AI hyperscalers, which includes the likes of Amazon (AMZN), Alphabet (GOOGL) and Meta (META), have continuously underperformed while their spending commitments have increased, he said. At the same time, AI hardware stocks, or the companies responsible for the manufacturing of semiconductor chips, computer servers and data centers, such as Nvidia (NVDA) and Taiwan Semiconductor Manufacturing Company (TW:2330), instead rose.

However, developments in regions like Japan and China are making the software cheaper than ever to run - something which currently isn't being reflected in spending forecasts among hyperscalers, he added. China's GLM-5.2 large language model, for example, was trained on 100,000 Huawei processors, and no involvement from Nvidia, per trade publications.

"The question is how far that rubber band can stretch ... If frontier intelligence can increasingly be developed in the East at a fraction of the cost incurred in the West ... then the largest capital allocators are also the ones most exposed to over-investment risk," Privorotsky said, outlining that the "breaking point" was likely to be one of the key spenders realizing that value for shareholders would be better improved by allocating less money to AI.

"The problem is that 'slightly less' is not embedded in anyone's assumptions," he noted. "The entire AI complex is priced for ever rising capex as inference demand grows."

Federico Manicardi and Victoria Campos, of JPMorgan's international market intelligence, said a number of factors are weighing on tech: "elevated expectations, euphoric sentiment, lack of free cash flow, push back from CTOs on surging token costs, a pivot to lower-priced models, draconian restrictions from Washington, higher debt/equity supply, etc. I think progress on orchestration, model fusion, quantization etc. are also things that should be monitored as they point at rising efficiency and headwinds to pricing power."

Shares in technology have faced a sell-off since Monday's session, with South Korea's Kospi index KR:180721 down 10% after reaching a new record closing price the day beforehand. The dip was driven by losses in Samsung Electronics (KR:005930) and semiconductor memory chip manufacturer SK Hynix (KR:000660). Futures linked to the Nasdaq (NQ00) declined by about 2.5% as constituents Micron Technology (MU) fell more than 7% and Intel (INTC) shed 6.5%.

-Nora Redmond

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

06-23-26 0550ET