Trending:
AI & Machine Learning

AMD posts record quarter, stock drops 8% - investors want pure AI play

AMD delivered record Q4 revenue of $10.27B with datacenter up 39%, but shares fell in after-hours trading. Wall Street's message: seasonal PC/gaming weakness isn't acceptable when you're competing in AI infrastructure.

AMD reported record Q4 2025 results on Tuesday - revenue up 34% to $10.27B, net income up 213% to $1.5B, datacenter segment growing 39%. The stock dropped 8% after hours.

The disconnect: AMD's Q1 2026 guidance of $9.8B (still 32% YoY growth) reflects seasonal decline in PC and gaming divisions. For investors fixated on AI infrastructure, that diversification looks like a liability. If you're not growing datacenter revenue fast enough to offset consumer cyclicality, you're not pure enough.

CEO Lisa Su isn't apologizing. She's prioritizing enterprise and high-end PC segments where margins matter more than volume. On memory shortages - DRAM prices have tripled, expected to double again in Q1 - Su says AMD planned years ahead for HBM supply needed for Instinct GPUs. "We've been planning for a significant ramp in both our CPU as well as our GPU business over the past couple of years."

The real question for enterprise buyers: does AMD's breadth make their infrastructure portfolio stronger or weaker? EPYC processors delivered double-digit datacenter growth. MI300X GPUs are landing deals (Zyphra's ZAYA1 training, France's Alice Recoque exascale system). ROCm software stack is maturing. Recent moves like divesting ZT Systems manufacturing to Sanmina and the Cisco/HUMAIN 1 GW AI joint venture show focus.

For CTOs evaluating EPYC vs Xeon for inference workloads, AMD's challenge isn't technical execution. Q4 proved they can ship. The challenge is investor perception treating anything non-AI as dead weight. That same diversification cushioned AMD through previous cycles. History suggests it will again.

Worth noting: AMD's last five AI-related announcements produced negative 24-hour stock moves four times. Strong results don't guarantee positive reactions when expectations are this skewed.