AMD’s EPYC CPUs have pushed the company’s data center revenue to new highs, exceeding Intel’s for the first time. In the competition for data center CPU revenue, there can only be one leader. While both Intel and AMD are enjoying renewed demand thanks to “Agentic AI,” AMD has managed to capture the larger share, outperforming its rival for the first time in the first quarter of the year.
According to DigiTimes, AMD’s data center revenue surpassed Intel’s for the first time during the first quarter. AMD had already been doing better than Intel in the data center market since the third quarter of 2025, but this first quarter marks the first time the company achieved higher revenue in data centers.
Data centers are quickly shifting more computing tasks to CPUs as AI inferencing has created a huge demand for processing power. Before, GPUs handled most computing, but now things have changed. The ratio of CPUs to GPUs has quickly dropped from 1:8 to 1:4, and is now moving towards 1:1. This means that for every four GPUs, there needs to be one CPU, and in the future, each GPU will have a CPU as demands grow.
We have already seen several companies asking for more CPU capacity or even building their own CPUs. AI firms are currently either talking about or developing their own custom designs to meet their needs, as chipmakers can’t keep up with their computing demands. This isn’t just a problem for Intel or AMD; it’s a large-scale supply chain issue, as TSMC, the world’s leading semiconductor maker, faces its own limitations. Companies like AMD rely heavily on TSMC to generate their revenue. That’s why AMD is also looking at Samsung to get additional chips.
The main obstacle to this revenue growth is the supply chain. With an unlimited supply, AMD and Intel could be making much more money than they currently are, just based on the demand that is needed and already booked for this decade alone.
While x86 companies continue to grow, Arm-based competitors from NVIDIA (Vera) and Arm (AGI) are also setting record figures for the future. Vera will be a key part of the Rubin platform. NVIDIA’s Rubin GPUs, along with its Vera, which is further boosted by the Groq 3 LPX (LPU), are seeing massive adoption. Meanwhile, Arm doubled its revenue estimates for AGI CPUs, showing huge interest in that chip too.
Looking ahead, Intel and AMD will also start shipping newer platforms. AMD’s Venice and AI-optimized Verano will be in the spotlight, while Intel works to release its 18A “Diamond Rapids” and its SMT-enabled follow-up, Coral Rapids, in the coming years.
