Trivariate: AI Stock Valuations May Have Peaked, but Revenue Growth Could Offset Compression
nashnova research
Trivariate Research founder Adam Parker argues AI stock multiples may have peaked for good, yet revenue growth can still offset the compression — investors need not bail out entirely.
Why might AI valuations have already hit their ceiling?
Parker's core call: rising rates + tightening AI regulation point the same way — AI multiples are likely headed lower and may have peaked permanently.
This means → the old playbook of riding multiple expansion is breaking down. From here, AI stocks need real revenue growth to move higher.
He is not calling for a full exit. His words: "We still want exposure to the space." Put simply = the ceiling is in, but that does not mean sell everything — the question is which names can keep pushing higher on earnings.
How exactly are rates and regulation applying pressure?
Rates: the Fed hiked overnight rates to 3.75%–4%, the first increase in over three years, after August payrolls surprised to the upside and inflation re-accelerated.
This means → compute-infrastructure companies carry heavy debt loads; higher rates eat directly into margins.
Policy: several U.S. states are considering restrictions on data-center construction. A May Gallup poll found 71% of Americans oppose building such facilities near their homes.
Trivariate also expects AI-driven job losses next year to trigger stricter regulation. This reflects a risk beyond financing costs — social backlash is becoming a real headwind.
Which names does the report say can withstand the squeeze?
Trivariate's filter: revenue expansion fast enough to outrun multiple compression — earnings growth fills the hole that shrinking valuations dig.
The list: data-center operator Iren, cloud providers Nebius and CoreWeave, chipmakers AMD and Marvell Technology, and database giant Oracle.
In plain terms = these companies share one trait — revenue is growing fast enough that even if the market pays a lower price per dollar of earnings, the stock still has fundamental support.
How have these names been trading lately?
CoreWeave jumped over 6% on Monday, yet remains down more than 26% over three months — a short-term bounce inside a deeper slide.
Oracle edged up on the day but has fallen over 19% in six months. Its forward P/E of 16.4× sits below the S&P 500's 19.3×, implying a modest discount.
Marvell has climbed steadily since late August, up roughly 200% year-to-date; AMD surged 9% on the day, with a year-to-date gain of 184%.
This means → even within the same "can withstand it" list, dispersion is wide — some names are recovering, others are still digesting earlier losses.
What is the biggest risk to this thesis?
Parker's optimism rests on one core premise: the earnings-revision narrative is still intact — consensus revenue estimates for AI companies are still being revised upward.
This means → if that revision cycle reverses — say, major customers cut compute spending or revenue growth disappoints — there is no valuation cushion left. Downside risk gets amplified.
In plain terms = the safety net under AI stocks right now is not cheap valuations; it is rising revenue. The moment revenue stalls, the net disappears.
市场有风险，内容仅供研究参考，不构成投资建议。