Nvidia Trades at Half AMD's Multiple — but Here's the Real Story the Market Is Missing

Nvidia trades at roughly 22x-25x forward earnings versus AMD at 35x-45x, even though Nvidia is expected to grow 25%-35% and AMD 30%-40%. Using a PEG framework, Nvidia screens around 0.7-1.0 versus AMD at about 1.0-1.5, suggesting Nvidia may be the better risk-adjusted value despite AMD's higher upside optionality. The article argues the valuation gap reflects proven growth and margin quality for Nvidia versus greater execution risk and embedded optimism for AMD.

Analysis

The setup is less about which chip vendor is “winning” and more about which stream of cash flows the market is willing to capitalize at a premium. NVDA’s multiple compression signals investors are discounting normalization in growth and maybe a peak in hyperscaler spend intensity, while AMD’s richer multiple implies the market is paying up for pathway optionality before the earnings base has fully materialized. That usually favors the incumbent on a risk-adjusted basis because execution disappointments hurt the challenger twice: lower multiple and lower growth conversion.

Second-order effects matter here. If AMD keeps taking share, the clearest loser may be not NVDA but the broader AI infrastructure stack: networking, interconnect, and thermal/power vendors that have been priced as if one dominant supplier keeps pulling capex forward. A more competitive accelerator market can also slow pricing leverage across the ecosystem, which is bearish for any supplier whose margin structure depends on scarcity rents rather than true unit demand growth.

The key catalyst over the next 1-2 quarters is not headline AI demand, but guidance tone from hyperscalers and any evidence that customers are diversifying away from a single-vendor architecture. If capex remains strong and NVDA’s supply chain still constrains volume, the market may have over-penalized it for decelerating growth that is still far above the market’s. Conversely, AMD’s multiple is vulnerable if share gains come with weaker-than-expected margins or if the market realizes that “AI exposure” is not the same as durable AI profitability.

The contrarian miss is that NVDA may be the cleaner long even with slower growth because its earnings quality is better and its downside is more observable. AMD is the higher-beta expression of the same theme, but the premium valuation leaves little room for product slippage, delayed ramps, or any sign that accelerator demand is being met by supply substitution rather than genuine incremental adoption.

Key Decisions for Investors

• Long NVDA / short AMD as a 3-6 month relative-value pair: prefer the lower-PEG, higher-margin incumbent while the market is still paying for AMD’s optionality; use a 1:1 notional hedge and reassess if AMD prints a step-change in gross margin or data-center share.
• Buy NVDA on pullbacks over the next 2-4 weeks with a risk-defined structure (e.g., call spread or financed collar): upside is capped in the near term, but the stock has better risk/reward if hyperscaler capex remains intact and sentiment resets toward durability.
• Avoid initiating fresh unhedged AMD longs at current levels; if you want exposure, use a call spread 3-9 months out to express upside from share gains while limiting multiple-compression risk.
• Watch for a catalyst basket trade around hyperscaler earnings: if capex commentary stays firm, add NVDA and fade AMD strength on the assumption the market is overpaying for optionality.
• If AMD underwhelms on margin conversion or guide quality, short AMD against a basket of AI infrastructure beneficiaries that still have pricing power; the thesis is that valuation compression will move first in the names with the least earnings quality.