AMD Stock Is Up 320% Over the Past Year. Should Investors Shift Their Attention Away From Nvidia?

AMD is benefiting from AI-driven demand, with Q1 revenue up 38% to $10.3 billion and data center sales up 57% to $5.8 billion. Management raised its server CPU TAM to $120 billion by 2030, doubled from prior estimates, and lifted Q2 revenue guidance to $11.2 billion, 46% above its previous outlook. The article remains constructive on AMD but notes Nvidia still dominates the GPU market with an estimated 86% share and trades at a far lower P/E than AMD.

Analysis

The market is starting to re-rate the AI stack from a single-bottleneck GPU story into a broader compute budget story. That is positive for AMD because incremental inference and agentic workloads should pull through more socket count, but the second-order effect is that the “AI spend per rack” may rise faster than unit economics for any one vendor, which helps suppliers with a full platform but also preserves pricing power for the incumbent GPU leader. In other words, AMD can take share without necessarily compressing NVDA’s economics meaningfully over the next 6-12 months.

The bigger opportunity may be in CPU attach rather than pure accelerator share. If agentic workloads really move from GPU-heavy to more balanced CPU/GPU architectures, then the beneficiaries extend beyond AMD to networking, memory, and interconnect vendors, because more heterogeneous compute increases system complexity and capex intensity. That dynamic also argues against the market’s tendency to frame AMD as a clean substitute for NVDA; in practice, the pie is expanding, and the fastest path to monetization is owning both layers of the stack rather than betting on displacement.

The main risk is valuation and expectations. AMD has already discounted a lot of the upside from better product mix and TAM expansion, so the stock becomes highly sensitive to any guide-up that is merely “good” instead of “exceptional.” That sets up asymmetric downside over the next 1-2 quarters if enterprise AI deployments pause, if inference optimization reduces hardware intensity, or if hyperscaler capex is reallocated back toward GPUs. NVDA’s lower multiple gives it a cleaner risk-adjusted setup unless there is visible evidence that CPU content is inflecting faster than consensus.

Contrarian take: the market may be underestimating how much this expands the addressable market for the entire datacenter complex, not just AMD. If CPU demand is the leading indicator for agentic AI scaling, then earnings surprises could show up first in vendors with less headline AI exposure and more hidden leverage to rack-level buildouts. The trade is less “AMD versus NVDA” and more “own the enablers while fading the most expensive narrative exposure.”

Key Decisions for Investors

• Stay long NVDA as the core AI hardware anchor; use a 3-6 month horizon and treat any relative weakness versus AMD as a chance to add, since the multiple is still more defensible than AMD’s and the earnings bar is lower.
• Initiate a pair trade: long AMD / short a basket of over-owned AI beneficiaries with weaker fundamentals only on post-earnings pullbacks; if AMD disappoints on guide quality, the downside is likely sharp because expectations are crowded.
• Prefer a barbell: long AMD and NVDA together, sized 60/40 in favor of NVDA, to capture stack-wide AI spending while reducing single-name valuation risk over the next 6-12 months.
• Avoid chasing AMD after large upside gaps; wait for a 10-15% consolidation or for a quarter with confirmation that CPU attach rates and guidance are still accelerating before adding risk.
• For more convex exposure, consider AMD call spreads into the next earnings window rather than outright equity, because the stock’s premium valuation makes downside re-rating more likely than sustained multiple expansion.