AMD's Data Center Business Is Skyrocketing. The Stock Is Falling. Here's Why.

AMD stock slid after Q2 results despite strong fundamentals, including 107% year-over-year data center revenue growth (vs. 57% last quarter) and margin improvement to a 17% operating margin from a 2% loss. The sell-off is attributed less to business performance and more to valuation/mispricing: AMD trades near ~70x forward earnings and ~37x 2027 earnings versus Nvidia at ~32x trailing earnings. Article notes AMD’s growth advantage appears smaller on a quarter-over-quarter basis (16% QoQ for AMD vs. Nvidia 21% in its latest quarter), keeping the market cautious on AMD overtaking Nvidia.

Analysis

This is primarily a valuation air-pocket, not a fundamental break. AMD is still benefiting from AI infrastructure spend, but the market is now forcing a higher proof standard: sustained share gains and margin expansion must arrive faster than consensus, or the multiple gets cut first. That tends to happen in three stages: immediate post-earnings de-rating, then 1-3 months of analyst estimate grind, then a bigger reset if the next guide fails to show accelerating operating leverage.

Relative value still favors NVDA over AMD despite the premium narrative investors often tell themselves. NVDA’s ecosystem lock-in, software attach, and supply priority convert revenue growth into earnings more reliably, which matters more than headline growth rates when the market is repricing AI beneficiaries. A secondary effect is that any disappointment in AMD pushes incremental AI allocation back toward the large-cap incumbents and away from the “second-source” trade in the semiconductor basket, which can pressure SOXX/SMH breadth even if the AI capex cycle remains intact.

The contrarian point is that the selloff could become too reflexive if AMD’s operating margin improvement proves durable. If the company keeps expanding margins while data center growth stays above peers, the current gap versus NVDA is too wide to sustain indefinitely. The key falsifier is not a single quarter, but whether AMD can sustain low-20s-ish data center growth and keep operating margin moving up on the next two prints; if that happens, the de-rating should reverse and the short relative-value case weakens quickly.

Key Decisions for Investors

• Prefer long NVDA / short AMD as a market-neutral relative-value pair for the next 1-3 months; the spread works if the market continues to reward execution quality and punish any “good but not perfect” AMD print. Falsify if AMD re-accelerates guidance or NVDA guides down meaningfully.
• For accounts that cannot run pairs, reduce AMD exposure on strength rather than weakness: use a 1-2 week window after any bounce to trim. The trade is a multiple reset from ~70x forward earnings toward a more forgiving valuation band; risk/reward is unfavorable until the next guide improves.
• Buy SOXX or SMH only on confirmation that semiconductor breadth is holding, not just AI names. If AMD weakness starts dragging the index while NVDA holds, that is a sign of narrowing leadership and a better entry for a basket than single-name AMD.
• Use AMD put spreads rather than outright puts if you need downside exposure; the stock can remain expensive longer than expected. A 1-3 month put spread is the cleaner way to express a 10-15% further de-rating without overpaying for volatility.