The stock market is starting to treat AI chipmakers and AI spenders as two separate trades instead of one, The Kobeissi Letter reports. The 30-day correlation between the largest US capital-expenditure spenders on AI infrastructure and the semiconductor index, the SOX, has fallen to near zero, the lowest reading in at least four and a half years. That's down sharply from +0.78 in April, and well below the +0.60 average correlation since the start of 2022. Since early June the two groups have moved in opposite directions: semiconductor stocks have rallied even as the biggest AI infrastructure spenders, the hyperscalers pouring hundreds of billions into data centers, have declined, and the pattern has also run in reverse on other days. Kobeissi reads the split as a sign that investors no longer see AI chip demand and AI infrastructure spending as a single story that rises and falls together. Chipmakers keep benefiting from AI demand regardless of who is paying for it, while hyperscalers face growing questions about whether their spending will generate enough revenue to justify the cost. Kobeissi's conclusion is that the next phase of the AI trade will be judged on profitability rather than on how much companies are willing to spend.
Why it matters · A falling correlation means chipmakers can keep winning even if a hyperscaler's AI spending disappoints, but it also means the market is starting to grade AI spenders on returns, not just on how much they spend.
Worth asking · Is this correlation breakdown the start of investors properly separating AI chip demand from AI capex risk, or just short-term noise from a handful of earnings surprises that will re-correlate once the dust settles?