In the third quarter of 2024, AMD made significant progress in the x86-compatible processor market, overtaking Intel in all segments. According to the results of the quarter, the cumulative share of AMD increased to 23.9%, which is 4.5% more compared to the same period last year (19.4%). This does not include processors used in semi-custom solutions such as SoCs for game consoles and IoT devices.

The biggest successes of AMD were recorded in the desktop segment, where thanks to the release of Zen 5 processors and the reduction of prices for chips of previous generations, the company increased its share to 28.7% (in the third quarter of 2023, this figure was 19.2%). In the mobile segment, the share of “red” processors also increased to 22.3%, which is an improvement compared to 19.5% a year earlier.

In the server segment, AMD continues to gradually increase its market share, which at the end of the third quarter was 24.2%. For comparison, a year earlier this indicator was at the level of 23.3%, and in the first quarter of 2020, AMD occupied only 5% of the server x86 processor market.

AMD has introduced its first 400 GbE-capable Pensando Pollara 400 network card – even before the official release of the Ultra Ethernet (UEC) specification version 1.0, which is expected in the first quarter of 2025. planned for the first half of 2025, immediately after the official release of the UEC.

The Pensando Pollara 400 AI NIC is designed for high performance computing (HPC) and artificial intelligence (AI) networks.

• Intelligent multipath distribution that dynamically distributes data packets along optimal routes to prevent network congestion.
• Path-aware congestion control that redirects data from temporarily congested paths for uninterrupted high-speed data flow.
• Fast detection and bypass of network failures, ensuring seamless communication between GPUs to maximize AI cluster performance and minimize latency.

These features make the Pensando Pollara 400 an excellent solution for a scalable and reliable AI infrastructure, optimizing network interaction and increasing the efficiency of data processing in data centers.