AMD Q3 EPS Beats by $0.04; Data Center Sales Hit $4.3B
AMD beat Q3 forecasts with EPS of $1.20 (+$0.04) and revenue of $9.25B, driven by record $4.3B data center sales (+22% YoY) and $1.3B gaming revenue (+181% YoY). It acquired ZT Systems for $4.9B to bolster its AI strategy, with MI300 GPU and Ryzen expansion into automotive, IoT and cloud.
1. Record Q3 Earnings and Market Share Gains
In the company’s fiscal third quarter ended October 31, 2025, AMD delivered EPS of $1.20, surpassing consensus by 3.4%, and revenue of $9.25 billion, beating analyst estimates by $500 million. Data center sales hit a record $4.3 billion, up 22% year-over-year, while client processor revenue rose 46% to $2.9 billion and gaming segment revenue surged 181% to $1.3 billion. Over the past month, shares climbed 10.83% following a 12.69% decline, extending a one-year gain to 95.74%. In Japan, the company’s gaming chips have captured a 45% market share, on track toward a 70% goal.
2. Strategic AI Push with MI300 GPUs and ZT Systems Acquisition
AMD’s MI300 GPU family continues to challenge its leading competitor in AI graphics, offering comparable performance at roughly one-quarter of the price. To support end-to-end AI deployments, the company acquired ZT Systems in a $4.9 billion transaction; integration of ZT’s cloud-architecture expertise aims to position AMD as a turnkey AI solutions provider. Partnerships are already in place with major cloud providers, and management projects a sustained double-digit CAGR in data center revenue through 2030, underpinned by expanding AI workloads in enterprise and hyperscale environments.
3. Supply Chain Risks and Competitive Pressures
Despite strong growth, AMD faces notable headwinds. As a fabless designer, the company relies on a single external manufacturing partner in Taiwan, exposing it to geopolitical tensions in the Taiwan Strait. On the competitive front, margin compression looms as MI300 chips are priced at a steep discount to rival products, and next-generation offerings from competitors promise faster AI training platforms. Additionally, U.S. export restrictions mandate capped performance versions of top products for certain regions, complicating engineering road maps and potentially delaying overseas deployments.