Sources indicate that AMD is accelerating its workstation GPU product line, with rumors of a new Radeon PRO GPU codenamed “Navi 48 XTW” soon to be unveiled. This GPU, based on the RDNA 4 architecture and equipped with 32GB of high-capacity onboard memory, will be a core component of AMD’s next-generation Radeon PRO W9000 series, bringing powerful performance to professional users.
Navi 48 XTW represents a new endeavor for AMD in the workstation sector. Unlike the consumer-grade RDNA 4 GPUs focused on the mid-range market, this workstation GPU is primarily designed to meet high-performance computing demands. It is reported that Navi 48 XTW will feature 32GB of GDDR6 memory, likely the highest memory capacity model in the Navi 48 series. GDDR6 memory is known for its high bandwidth and stability, making it suitable for handling complex 3D modeling, real-time rendering, and AI training tasks. Although specific technical details have not been fully disclosed, industry speculation suggests that this GPU may utilize TSMC’s 4nm process technology. Compared to the 5nm process of the previous generation RDNA 3 architecture, this could offer an approximately 6% increase in transistor density and improved energy efficiency.
AMD has been steadily expanding its presence in the workstation GPU market in recent years. Its RDNA 3-based Radeon PRO W7900 and W7800, with 48GB and 32GB of memory respectively, have already demonstrated competitiveness in AI inference and content creation. The emergence of Navi 48 XTW signifies AMD’s introduction of RDNA 4 technology into the professional domain, further challenging NVIDIA’s RTX PRO series. Contrary to its strategy of abandoning high-end GPU competition in the consumer market, AMD is choosing to deeply cultivate high-performance products in the workstation sector, aiming to meet the stringent demands of industries such as CAD design, post-production, virtual reality, and data science.
Information regarding the performance of Navi 48 XTW is currently limited, but we can refer to the performance of the RDNA 4 architecture in consumer-grade GPUs. The RDNA 4 architecture optimizes ray tracing engines and AI acceleration units, offering an approximately 15% improvement in rasterization performance compared to RDNA 3, with even more significant gains in ray tracing performance, potentially approaching the level of the NVIDIA RTX 4070 Ti. As a professional-grade product, Navi 48 XTW is expected to further enhance the number of compute units and memory bandwidth to handle parallel multitasking and high-resolution rendering. Additionally, AMD may continue to utilize its signature Infinity Cache technology, using a large-capacity cache to alleviate memory bandwidth pressure and improve data access efficiency.
Navi 48 XTW is expected to be officially unveiled at the Computex trade show in June 2025 or at AMD’s Advancing AI event. The Advancing AI event, which focuses on professional and data center solutions, might be a more suitable platform for the launch of a workstation GPU. Simultaneously, AMD is also advancing two new products in the consumer market, the Radeon RX 9060 XT and RX 9070 GRE, which are speculated to be announced during Computex. This indicates that AMD is employing a differentiated strategy, simultaneously targeting both the consumer and professional markets.
In terms of pricing, the specific cost of Navi 48 XTW has not been disclosed, but referring to the approximately $3999 price tag of the Radeon PRO W7900 48GB version, Navi 48 XTW might be positioned in the $3000 to $4000 range. Compared to NVIDIA’s RTX 6000 Ada (around $6800), AMD may hold an advantage in terms of price-performance ratio. Furthermore, Navi 48 XTW is likely to support the PCIe 5.0 interface, offering higher data transfer rates, and will also feature DisplayPort 2.1, supporting multi-monitor configurations with ultra-high resolutions and refresh rates, catering to the needs of professional users.
It is worth noting that AMD is also continuously strengthening its software ecosystem. The Radeon PRO series GPUs support the AMD Software: PRO Edition driver, providing an intuitive user interface and optimized performance. For AI and machine learning tasks, AMD’s ROCm platform does not yet support RDNA 4, but support is expected to be added in future updates, further enhancing the performance of Navi 48 XTW in AI workloads.
Meanwhile, AMD’s competitor in the workstation market, NVIDIA, is also actively making moves. NVIDIA’s Blackwell architecture RTX PRO GPUs are rumored to feature 96GB of GDDR7 memory, leading in both memory bandwidth and capacity. However, with its higher price-performance ratio and open software ecosystem, AMD still has the potential to secure a place in the professional market.
The launch of Navi 48 XTW is not only a significant upgrade for AMD in the workstation GPU market but also the first demonstration of the RDNA 4 architecture in professional applications. With 32GB of memory, an advanced 4nm process, and optimized ray tracing performance, this GPU can provide professional users with an efficient and cost-effective solution. In the coming months, as more specifications and performance data are revealed, the performance of Navi 48 XTW will be even more anticipated.
This move by AMD indicates that its strategy in the GPU market is becoming more diversified. From consumer-grade to professional-grade, the application scenarios for the RDNA 4 architecture are continuously expanding. Whether in gaming, content creation, or AI computing, AMD is attempting to meet the needs of different users through technological innovation and precise positioning. The introduction of Navi 48 XTW may become a crucial step for AMD in the workstation market.