AI Trade Decouples as $730bn Big Tech Spend Fails to Reassure
The correlation between major tech buyers and semiconductor suppliers has collapsed to a four-year low, signaling growing skepticism that $730 billion in planned 2026 AI spending will generate timely returns.
Alphabet and Meta have recently escalated their 2026 capital expenditure forecasts, pushing the combined AI infrastructure spending of the four largest hyperscalers to roughly $730 billion. This unprecedented capital deployment encompasses massive data centers, custom silicon, AI accelerators, networking equipment, and the extensive power grids required to run them.
Despite these escalating commitments, the equity market is no longer rewarding the companies writing the checks. The 30-day correlation between major AI spenders and semiconductor stocks has plummeted from +0.78 to near zero. This represents the lowest reading in four and a half years, marking a decisive break from the synchronized rally that defined the sector over the past year.
The divergence highlights a fundamental shift in investor psychology. For much of the previous year, the investment thesis was straightforward: buy the hyperscalers building the infrastructure and the chipmakers supplying them. As the price tag for these projects climbs into the hundreds of billions annually, that simple dynamic has fractured.
Investors are now explicitly focused on the timeline for returns. The four hyperscalers—Alphabet, Meta, Microsoft, and Amazon—carry an estimated $1.65 trillion in off-balance-sheet AI obligations. This staggering liability has shifted the market's priority. The emphasis is no longer on the sheer scale of capital deployment, but rather on identifying which of these tech giants can actually convert these investments into durable cash flow.
Hardware suppliers continue to reap the immediate benefits of this spending cycle. Companies like Nvidia, Broadcom, Micron Technology, and Taiwan Semiconductor Manufacturing are still seeing robust demand fueled by hyperscaler capital expenditures. Furthermore, Wall Street analysts are already anticipating another round of spending increases heading into 2027.
However, the equity market is currently treating the hardware providers and the infrastructure builders as distinctly different investments. The practical question dominating institutional discourse has shifted from how much these giants will spend to who ultimately pays for it and when it will generate acceptable returns. Until the hyperscalers can demonstrate that their massive AI outlays will drive meaningful revenue and margin expansion, the once-reliable trade of buying both sides of the AI supply chain will likely remain under pressure.
