Institutions pile into AI winners
- Goldman Sachs said AI-linked companies now account for nearly 45% of the S&P 500, showing institutional money is clustering in a smaller set of winners. - The bank said that share was about 25% when ChatGPT launched in late 2022, a 20-point jump driven by chips and infrastructure. - The concentration leaves U.S. equities more exposed to a handful of AI names and their spending cycle. (goldmansachs.com)
AI-linked companies now make up nearly 45% of the S&P 500, according to Goldman Sachs, a sign that big investors have crowded into a narrow AI trade. (msn.com) (goldmansachs.com) Goldman’s comparison point is late 2022, when ChatGPT launched and AI-linked companies were about 25% of the index. The share has since climbed roughly 20 percentage points. (msn.com) (markets.financialcontent.com) The move tracks the market’s shift toward companies that sell the picks and shovels of the boom: chips, servers, cloud capacity, networking gear and power equipment. Goldman has argued that the AI trade moved beyond Nvidia into infrastructure suppliers needed to build data centers. (goldmansachs.com 1) (goldmansachs.com 2) That buildout is expensive. Goldman said AI companies may invest more than $500 billion in 2026, while Amazon, Microsoft, Meta and Oracle are on track to spend about $1.5 trillion on AI infrastructure from 2023 through 2026. (goldmansachs.com) (stocktwits.com) The same concentration that lifted the index also raises the market’s dependence on a small group of companies. Goldman has previously warned that U.S. stocks become more vulnerable when returns are concentrated in large technology names. (goldmansachs.com 1) (goldmansachs.com 2) The pressure point is not only valuation. Goldman said investors have already started to distinguish between AI companies with strong earnings support and those funding capital spending with debt. (goldmansachs.com) The physical demands behind the trade keep getting larger. Goldman forecasts global data-center power demand will rise 50% by 2027 and as much as 165% by 2030 from 2023 levels. (goldmansachs.com) That helps explain why investors keep rewarding chipmakers and electrical equipment suppliers when earnings confirm AI demand. Reuters reported on April 24 that U.S. chip stocks hit record highs after Intel gave a stronger revenue forecast. (usnews.com) (msn.com) For now, the AI rally is still broad enough to pull the index higher, but narrow enough that a stumble in a few heavyweights would matter quickly. Goldman’s 45% figure puts a number on that dependence. (msn.com) (goldmansachs.com)
