AMD has launched the Radeon AI Pro R9700, a specialized graphics card designed to tackle professional visualization and Artificial Intelligence inferencing applications. Featuring the fully-enabled ´Navi 48´ RDNA 4 silicon, the card is equipped with 32 GB of memory and is validated for enterprise settings. Its architecture supports deployment of up to four cards in a single workstation, collectively utilizing a 128 GB memory pool via the high-speed PCI-Express Gen 5 host interface. In a notable shift from competitors like NVIDIA, AMD allows board partners such as Gigabyte to create custom versions, giving manufacturers flexibility in design and integration.
Gigabyte´s Radeon AI Pro R9700 integrates seamlessly into its AI TOP product line and is engineered for easy scalability in multi-GPU workstation configurations. The card features a two-slot lateral airflow cooling system, which draws air from both sides to enable dense packing of up to four cards side by side on server or workstation motherboards. A recessed obverse side further reduces airflow interference when multiple cards are installed, and the 300 W-rated 16-pin 12V-2×6 power connector is conveniently placed at the card’s tail end, simplifying cable management for system builders.
Alongside the graphics card, Gigabyte displayed the X870E AORUS Master X3D Ice at its booth—an award-winning motherboard that complements high-performance professional systems. The coordinated launch highlights growing demand for robust hardware solutions tailored to the surging needs of the Artificial Intelligence, machine learning, and professional visualization sectors. By leaning into a board partner-driven approach, AMD and partners like Gigabyte target enterprise customers seeking custom, scalable solutions for intensive computational workloads.
