AMD Just Joined Nvidia in the $1 Trillion Club. At $614, Wall Street Is Paying Too Much for a Distant No. 2.

AMD reached a $1 trillion market capitalization after its stock gained 187% year-to-date, but the article argues the valuation is difficult to justify given its substantially smaller AI data-center business and premium earnings multiple. Nvidia generated $89 billion in latest quarterly data-center revenue versus AMD's $6.7 billion, while AMD trades at roughly 158x trailing and 81x forward P/E, compared with Nvidia at about 28x and 24x. Although AMD has secured major future AI deployments with OpenAI, Meta, Anthropic, Oracle, and Microsoft, the first major installations are largely scheduled for H2 2026; the article advises investors to trim gains and avoid chasing the stock.

Analysis

AMD’s announced capacity commitments should not be valued as equivalent to contracted revenue: deployment schedules extend into 2H26, are contingent on power, rack integration, software qualification and customer capex, and include warrant structures that economically resemble customer-acquisition spend. The market is likely capitalizing headline gigawatts before recognizing the offset from share dilution, lower initial gross margins and the working-capital burden of scaling a rack-level offering. This creates a vulnerable setup over the next 1-3 quarters if management cannot translate pipeline language into booked MI-series revenue, gross-margin progression and firm delivery milestones.

The more important competitive issue is not whether AMD wins incremental accelerator share, but whether it can monetize each deployment at Nvidia-like system economics. Hyperscalers have a strong incentive to maintain AMD as a credible second source, which supports unit demand but also caps AMD’s pricing power; customers can use AMD capacity to negotiate against NVDA rather than migrate core workloads. NVDA’s installed software, networking and developer ecosystem make switching costs highest in training, while AMD’s nearer-term opportunity is more likely inference and price-sensitive deployments—segments where margin durability is inherently less certain.

A second-order beneficiary is the AI infrastructure supply chain: TSM can benefit regardless of accelerator share, while optical/connectivity vendors such as MRVL, COHR and LITE gain if rack-scale architectures proliferate. The contrarian point is that AMD’s strategic value to customers may justify a premium multiple for longer than fundamentals alone suggest; a credible software-performance breakthrough or a major customer converting commitments into accelerated purchase orders could force a sharp short-covering rally. Thesis falsification for a bearish AMD view would be upward revisions to FY27 revenue/EPS accompanied by sustained data-center gross-margin expansion, not merely additional capacity announcements.

Key Decisions for Investors

• Initiate a 3-6 month relative-value position: short AMD / long NVDA, sized beta-neutral. Target 15-25% relative outperformance as AMD’s valuation converges toward execution evidence; stop if AMD reports two consecutive quarters of data-center revenue and gross margin materially above consensus or confirms accelerated customer deployments.
• For long-only books, trim AMD into strength and retain only a core allocation until the next earnings print provides booked-order, delivery and warrant-dilution disclosure. Re-enter only if management quantifies FY27 accelerator revenue, rack-level gross margin and customer acceptance milestones; headline capacity alone is insufficient.
• Maintain a 6-12 month basket long TSM, MRVL and COHR versus a modest AMD underweight to capture AI system buildout without taking single-vendor accelerator share risk. Review if hyperscaler capex guidance softens, because optics and foundry demand are more cyclically exposed than accelerator vendors’ backlog narratives imply.
• Watch AMD implied volatility ahead of earnings rather than buying outright calls. A defined-risk bearish put spread is appropriate only if options pricing does not already discount a post-earnings move larger than prior reporting reactions; the missing input is current IV/skew and consensus estimates.