ARM Trades at 100x Forward P/E vs NVDA at 33x on Vera CPU Royalties of $40M-$50M

Arm Holdings has become a case study in how valuation consensus can diverge sharply from fundamentals when a story captures market imagination. The semiconductor licensing giant is trading at approximately 100x forward P/E, a premium that appears anchored to assumptions around Nvidia's planned Vera CPU and the potential royalty stream flowing to ARM through cross-licensing. By contrast, Nvidia, which is actually capturing the bulk of hyperscaler AI capex spend, trades at 33x forward P/E with $91 billion in validated guidanceCompany-issued forecasts of future financial performance..

The Vera narrative is compelling in theory. Nvidia has disclosed plans for a standalone Vera CPU architecture aimed at data-center inference workloads, and this chip would theoretically require ARM instruction-set licensing. Market estimates for Vera's revenue run from $400 million to $1 billion annually in the out-years. At these penetration rates, ARM would capture 2 to 5 percent of that revenue through royalties, roughly $40 million to $50 million per year at the high end of estimates.

Yet ARM's valuation is implying far more aggressive capture rates and a much larger overall serviceable market. The 100x multiple suggests that traders are pricing in either a massive expansion of ARM-based CPUs in data centers (a market Nvidia currently dominates), or a belief that ARM's total licensing opportunity extends well beyond Vera into other secular growth areas like automotive, IoT, and edge AI. This extrapolation may be justified long-term, but it is worth noting the dramatic gap versus Nvidia's near-term certainties.

ARM's stock has surged to $315, and short positions have reportedly been squeezed as momentumThe empirical fact that winners keep winning over the medium term. accelerates. The company reported strong Q1 earnings and has signaled confidence in the licensing cycle, but the valuation premium to Nvidia is becoming increasingly difficult to justify on a P/E basis alone. Skeptics worry that the market is front-loading a decade of licensing upside into a single name, and that any disappointment in Vera adoption or hyperscaler CPU diversification would trigger a sharp repricing.

A realistic bull case for ARM might be a 40x-50x forward multiple, still well above Nvidia but acknowledging real licensing optionality. At 100x, ARM is pricing in perfection, and historically when perfection is priced in, the market tends to find imperfection.