AMD: CapEx Is Fueling The Next Leg

AMD reported data center revenue up 107% to $6.7B, now 58% of total revenue, with a 31% operating margin. Despite Q2 CapEx rising to $808M, the company generated $1.6B in free cash flow. Management said Helios demand beat initial expectations and expects total data center revenue to more than double by 2027, signaling a strong multi-year growth outlook.

Analysis

AMD is migrating from a “share-take story” to a “platform economics story.” Once data center is the majority of revenue, the market should start valuing the name off AI/cloud capex durability and operating leverage, not just unit growth; that typically expands the multiple before the earnings base fully catches up. The key second-order winner is the manufacturing ecosystem tied to advanced nodes and packaging, while Intel’s server franchise is the clearest relative loser as socket share loss compounds into weaker pricing power and weaker software/enterprise pull-through.

The near-term risk is that this becomes a narrative trade if the demand pool is concentrated in a few hyperscalers. In that setup, one procurement pause, a roadmap slip, or a supply constraint can slow sequential growth quickly and compress the stock within 1-2 quarters even if the long-term thesis remains intact. What the market should watch more closely than headline growth is margin trajectory: if operating margin stalls while capex stays elevated, investors will start questioning whether AMD is buying growth rather than monetizing it.

Contrarian view: consensus is likely still too anchored to AMD as a cyclical follower, so the rerating may be underdone if data center keeps compounding and mix stays favorable. But the upside is not linear—if 2027 expectations are being pulled forward, the stock can become sensitive to any guidance moderation. The thesis is falsified if the next two quarters show decelerating data-center sequential growth, a flat-to-down margin path, or weaker hyperscaler spending commentary; that would be the point to fade the momentum rather than chase it.

Key Decisions for Investors

• Long AMD vs. short INTC over the next 3-6 months: the relative call is cleaner than an outright long because AMD’s data-center mix gives it better margin leverage while INTC still carries structural server-share risk.
• Buy AMD on a 3-5% pullback rather than strength; use a 1-2 quarter horizon and require continued data-center acceleration as the stop condition.
• If entering ahead of the next earnings print, use a call spread instead of stock to cap downside from any guidance reset; target a 2:1 to 3:1 payoff if the market starts modeling the 2027 ramp earlier.
• Pair trade: long AMD / short SOXX or SMH only if semis sell off broadly on macro risk while AMD holds relative strength; this isolates single-name re-rating from sector beta.
• Set an alert on the next two quarters of sequential data-center growth and gross margin: if either disappoints, reduce exposure quickly because the stock will likely be priced on forward revisions, not trailing results.