AMD Stock Has Blown Past Nvidia This Year. Is It Still the Better Buy?

AMD has gained nearly 190% in 2026 and briefly exceeded a $1 trillion market capitalization following enthusiasm for its Helios rack-scale AI system, while Nvidia has risen 23% and remains worth roughly $5.5 trillion. AMD reported $11.5 billion of revenue for the quarter ended June 27, up 50% year over year, but guides to approximately 41% growth in the current quarter. The article argues Nvidia offers a more attractive risk-adjusted investment: its forward P/E is about 25 versus AMD's 40, while Nvidia generated $192.9 billion of trailing-12-month net income compared with AMD's $6.4 billion.

Analysis

AMD’s rerating leaves the equity unusually dependent on a narrow set of proof points: rack-scale system bookings converting into recognized revenue, gross-margin accretion despite system-level integration costs, and credible follow-on demand from more than one hyperscaler. The key risk is not simply a lower earnings multiple; it is that a modest revenue or margin miss would expose how much of the valuation embeds share gains against NVDA’s software ecosystem and installed-base advantage. Near term, AI-capex commentary from Microsoft, Meta, Amazon and Oracle is more important than broad semiconductor demand, while HBM availability and TSMC advanced-packaging capacity remain practical constraints on both vendors.

The cleaner relative expression is long NVDA versus short AMD rather than an outright semiconductor short. NVDA’s lower earnings multiple may reflect a structurally superior mix, networking attach, CUDA switching costs, and materially greater capacity to defend pricing; AMD’s upside requires customers to validate lower total cost of ownership at scale, not merely qualify an alternative supplier. Consensus may be too quick to assume that competition necessarily compresses NVDA margins: constrained supply and incremental AI-cluster demand can permit both firms to grow, making AMD’s valuation vulnerable chiefly if its share gains fail to arrive at the pace already discounted.

Over the next 1-3 months, track AMD’s disclosed AI revenue backlog, system shipment cadence, gross-margin guide, and named customer deployments against expectations. Over 6-18 months, AMD can earn its premium only if its rack-scale offering expands accelerator content and creates recurring software/networking economics; otherwise it remains a lower-margin component challenger. The relative thesis is falsified by sustained AMD AI revenue growth materially above guidance with gross margins expanding, or by NVDA showing order digestion, declining networking attach, or a meaningful gross-margin reset.

Key Decisions for Investors

• Initiate a market-neutral long NVDA / short AMD pair over the next 1-2 weeks, sized by beta or semiconductor-factor exposure rather than equal dollars. Target 15-20% relative return over 6-12 months; reassess if AMD raises AI revenue expectations materially and guides gross margin higher, or if NVDA cuts forward gross-margin guidance by more than 200 bps.
• Do not chase an outright AMD short solely on forward P/E. Use the next earnings setup only if options imply a move below the stock’s post-results volatility range and channel checks show no incremental hyperscaler deployment; a defined-risk put spread 3-6 months out is preferable to uncovered short exposure.
• Maintain NVDA as the core AI-compute long, but reduce exposure ahead of hyperscaler capex reports if aggregate capex guidance decelerates or lead times normalize. The immediate catalyst is confirmation that inference and networking demand offset any training-cluster digestion.
• Set a research alert on AMD accelerator revenue, HBM supply commitments, and TSMC CoWoS allocation. Evidence of multi-customer rack-scale production volumes—not pilot announcements—would invalidate the bearish AMD leg and warrants closing the relative short.