AMD tops estimates for first quarter as data center revenue jumps 57%

AMD reported Q1 EPS of $1.37 vs. $1.29 expected and revenue of $10.25 billion vs. $9.89 billion expected, with revenue up 38% year over year. Data center sales surged 57% to $5.8 billion, and Q2 revenue guidance of about $11.2 billion also topped consensus at $10.52 billion. Shares rose about 5% after hours as demand for AI chips continues to drive growth.

Analysis

AMD’s print changes the market from “can AMD participate?” to “how fast can it convert AI demand into supply-constrained revenue.” The key second-order effect is that the bottleneck is no longer just silicon design quality; it is now packaging, HBM access, board-level integration, and rack deployment capacity. That shifts the winner set upstream and sideways: memory, advanced packaging, substrate, and foundry ecosystem names should see more durable pricing power than the headline GPU vendors themselves.

The most important competitive implication is that AMD is becoming a credible second source for hyperscalers that do not want single-vendor dependence. Even if AMD never closes the raw performance gap with NVDA, a multi-sourcing regime can still support outsized share gains because large buyers optimize for supply assurance, power efficiency, and deployment cadence. That creates a more interesting threat to Intel than the market may appreciate: CPU share can stabilize or improve just as AI infrastructure spending expands, especially if agentic workloads raise CPU utilization per rack.

The contrarian risk is that the market is extrapolating too smoothly from demand strength to earnings durability. In the next 1-2 quarters, any slip in packaging capacity, HBM allocation, or rack qualification could delay revenue recognition and compress sentiment even if end demand remains strong. Over 6-12 months, the bigger risk is that the AI capex cycle becomes more price-sensitive: hyperscalers may dual-source aggressively, pressuring ASPs once initial scarcity premiums normalize.

For NVDA, this is not a near-term share-loss thesis, but it does justify a relative-value lens rather than outright bearishness. The market may be underpricing the breadth of AI spending beneficiaries outside the dominant platform vendor; AMD’s rally validates the second-choice supplier trade and should keep multiples elevated across the AI supply chain, but also increases the odds of overcrowding in the obvious winners.

Key Decisions for Investors

• Long AMD vs. short NVDA on a 2-4 month horizon via a moderated pair trade, sized for relative multiple compression rather than directional AI weakness; best entry is on any post-earnings fade in AMD or strength in NVDA, with stop on continued share-gain narrative or material guide raise from NVDA.
• Add exposure to the AI supply chain bottlenecks: long MRVL or AVGO against a basket of pure-play AI compute names for 3-6 months, targeting the thesis that packaging/interconnect/software content captures more incremental margin than accelerators once allocation constraints persist.
• Buy INTC as a tactical catch-up trade over 1-3 months, but only as a trading position: the market is beginning to assign value to x86 resilience in agentic workloads, and the asymmetric upside is a short squeeze if enterprise/server demand surprises again.
• Use call spreads on AMD for the next 90 days rather than outright stock if chasing momentum; implied volatility should stay supported, but upside is now more likely to be path-dependent on supply commentary than on another clean beat.
• Fade crowded long-only AI baskets on sharp strength if HBM/packaging lead times do not improve over the next quarter; that is the first place where the market can discover that demand is strong but the monetization curve is lumpy.