AMD Just Crossed $1 Trillion. Is it Still Worth Buying Now?

AMD surged 10% to a record above $615, lifting its market capitalization above $1 trillion for the first time; the shares are up more than 180% year-to-date. Q2 revenue rose 50% year over year to $11.5 billion, while data-center revenue more than doubled and EPYC server-processor sales increased 70%. Management expects data-center revenue to more than double again in 2027, supported by AI-compute demand and hyperscaler deployments, but AMD trades at roughly 24x trailing sales and more than 55x forward earnings—premiums to Nvidia and Broadcom that leave limited room for an AI-capex slowdown.

Analysis

The investable issue is not whether AMD participates in AI compute, but whether it can convert design wins into a durable rack-scale profit pool before expectations become the earnings base case. Complete-system deployments typically carry lower blended gross margin initially than accelerator-only sales and require validated networking, memory, power, and software integration; any qualification slippage can push revenue recognition by quarters. The market is likely underwriting both sustained hyperscaler capex and rapid AMD share gains simultaneously, leaving little room for a normal deployment ramp.

Meta application traction is an unreliable near-term proxy for incremental semiconductor procurement. Inference demand can be absorbed through better model efficiency, existing capacity utilization, or purchases from multiple vendors; the actionable confirmation is capex guidance, disclosed accelerator mix, and AMD backlog/conversion commentary over the next two earnings cycles. A stronger AMD alternative also creates a second-order risk for Nvidia: it may need to use pricing, financing, or bundled networking/software economics to defend strategic accounts, constraining gross-margin upside even if industry demand remains robust.

Near term, momentum and benchmark buying after a major capitalization threshold can keep the stock dislocated for days to weeks. Over 1-3 months, the valuation gap is vulnerable if AMD does not raise its data-center outlook or demonstrate that rack-scale revenue is accretive to gross margin; a broad AI-capex reset would amplify that downside because AMD has less valuation support from mature, diversified cash flows. Over 6-18 months, the contrarian upside is that a credible second supplier may expand total AI infrastructure spending by lowering customer concentration and supply risk, making a simple short structurally dangerous.

Key Decisions for Investors

• Initiate a 3-month relative-value position: short AMD versus long NVDA in equal AI-beta-adjusted dollars after the current momentum impulse stabilizes. Target 15-25% relative downside if AMD’s next results fail to raise data-center guidance; cover if AMD reports material backlog conversion plus gross-margin expansion, or if the pair moves 10% against entry.
• For long-only AI exposure, rotate incremental AMD allocation into NVDA or AVGO over the next 1-3 months rather than chase AMD. The objective is to retain AI-capex upside while owning businesses with less execution embedded in the multiple; reassess if AMD’s forward sales multiple compresses materially without a deterioration in its data-center outlook.
• Buy AMD downside only through defined-risk structures: 3- to 6-month put spreads, funded where feasible by selling farther-out-of-the-money calls, sized small given squeeze risk. The catalyst window is the next hyperscaler capex updates and AMD earnings; avoid a naked short while passive flows and momentum remain supportive.
• Set an alert around Meta’s capex guidance and AMD’s disclosed rack-system mix, backlog, and gross margin. Treat a capex increase without corresponding AMD-specific procurement evidence as sector-positive but not confirmation of AMD revenue; treat two consecutive quarters of data-center outperformance with expanding margin as thesis falsification.