AMD Revenue Beats, But Earnings Miss as AI Margins Shrink

AMD delivered a strong second quarter in 2025, recording $7.69 billion in revenue, a 32% year‑over‑year rise and topping Wall Street projections of approximately $7.42 billion. This growth was driven by surging AI and computing demand as data centre revenue climbed 14% to $3.2 billion, PC client chip sales skyrocketed 57% to $2.5 billion, and gaming revenue jumped 73% to $1.1 billion, buoyed by new Radeon GPUs and Zen 5‑based Ryzen AI product launches.

Adjusted earnings per share came in at $0.48, in line with consensus estimates but the standout figure was the hit to gross margin. Adjusted non‑GAAP margin fell to 43%, versus a potential 54% absent an $800 million charge tied to U.S. export restrictions on MI308 AI GPUs destined for China. This export constraint echoes across sectors, costing AMD an estimated $1.5 billion in revenue across Q2 and Q3.

Investors initially cheered but pushed AMD’s stock down around 4–6% immediately after hours, reflecting concern over narrower margins and ongoing regulatory uncertainty despite the revenue beat and optimistic outlook.

Looking ahead, AMD forecasted Q3 revenue near $8.7 billion, well above consensus, and anticipated gross margins rebounding to approximately 54% once export-related drag subsides. This outlook underscores confidence in the new Instinct MI350 series, which uses CDNA 4 architecture built on TSMC’s advanced 3 nm process and supports low‑precision formats like FP4 and FP8, delivering significant gains in AI inference performance.

CEO Lisa Su emphasised that robust demand for AMD’s broad AI and computing portfolio positions the company well for the rest of 2025, supported by momentum from MI350 deployments, Ryzen and EPYC share gains, and systems integration through the ZT Systems acquisition.

Technically, AMD’s competitive strength lies in offering inference-focused AI acceleration with the MI350 line, which some analysts expect could generate $10 billion in annualized AI chip revenue by year‑end challenging Nvidia’s dominance in data‑centre GPU markets. UBS analyst Timothy Arcuri raised AMD’s price target to $210, citing growing confidence in AMD’s AI roadmap.

Despite being behind Nvidia, AMD differentiates through a full-stack strategy combining Zen 5 CPU cores, CDNA 4 GPUs, Infinity Fabric interconnect, and its ROCm software stack into rack-scale systems like the upcoming “Helios” MI400 architecture (due in 2026) and future MI500 series in 2027.

On balance, AMD’s Q2 report highlights the company in transition: scaling AI momentum and broad-based computing growth, while grappling with export-driven margin impact. With strong free cash flow and resolute investment in AI acceleration, AMD appears focused on closing the gap and competing aggressively in the global AI chip race.