AI Equities in Late Cycle: HTX Research Warns of Speculative Characteristics

HTX Research has released a report on US AI equities, highlighting that while technological diffusion is still in its early stages, capital expenditure and valuations have moved ahead of it. The research arm of HTX found that the variables driving equity prices are shifting from model parameter counts and capital expenditure to token production costs, task-completion reliability, usage intensity, and enterprise-workflow penetration.

The report notes that the magnitude of capital spending is increasing, with J.P. Morgan Asset Management estimating that five US hyperscalers will spend around $697 billion in 2026. This has led to a shift in market attention from revenue growth to return on capital.

While AI technology itself is growing in real terms, capital expenditure and external financing are displaying increasingly speculative characteristics. The report warns that high-multiple second-tier equities may be overvalued, and that investors should look for alignment between normalized valuation, competitive moats, cash flow, and AI optionality.

HTX Research views Alphabet as offering the most compelling overall asymmetry, but notes that other companies such as Microsoft, Meta, TSMC, NVIDIA, Amazon, Oracle, Micron, AMD, Arista, and Vertiv also have significant potential. The report concludes that AI is reshaping how crypto investors allocate capital, with a growing share of users treating crypto and US equities as different allocation directions within a single global risk-asset system.