Goldman Sachs analysts warn that the explosive growth in AI infrastructure spending by major tech companies is set to peak later in 2026, potentially impacting valuations of AI infrastructure stocks. The investment bank estimates that hyperscaler capital expenditures will reach $1.5 trillion between 2023 and 2026, with spending growing to $667 billion in 2026 alone.

This represents a 62% increase from 2025 levels and is projected to exceed 90% of cash flows, surpassing the dotcom boom era. However, analysts led by Ryan Hammond and Ben Snider predict that the quarterly growth rate of AI infrastructure capex will decelerate in late 2026. This slowdown could create vulnerabilities for stocks considered essential providers of AI infrastructure.

The analysts specifically highlighted Nvidia, whose processors power many AI systems, as facing potential challenges. Despite its dominant market position, Nvidia's rapid growth has shown signs of cooling, with shares underperforming this year amid tough prior-year comparisons and increasing competition from rivals like Advanced Micro Devices. The Goldman strategists warned that even a strong market position can eventually give way to fears of over-earning as competition intensifies and demand sustainability becomes uncertain.