Market anticipates strong shipments for NVIDIA’s custom RTX PRO 6000

The NVIDIA RTX PRO 6000 has recently generated significant buzz in the market, with expectations running high for strong shipment performance driven by solid demand. However, Avril Wu, Senior VP of Research at TrendForce, points out that uncertainties remain.

Specifically, the custom low-power variant designed for the Chinese market faces challenges related to cost-performance disadvantages and increasing competition from other players. In addition, memory supply constraints could further impact overall shipments.

TrendForce anticipates that the custom-spec NVIDIA RTX PRO 6000 will be introduced in the second half of this year. So far, it has not generated substantial interest among Chinese buyers. One key reason is the more competitively priced Huawei Ascend 910c, which is considered a cost-effective alternative and already sufficient for many use cases. The 910c is expected to ramp up production volumes soon, gradually replacing the existing 910B solution.

Wu notes that NVIDIA has adopted a diversified memory procurement strategy tailored to different product categories. SK hynix serves as the primary supplier of HBM for high-end AI GPUs, with Micron as the secondary source. LPDDR is mainly supplied by Micron, while GDDR memory relies heavily on Samsung.

The RTX PRO 6000 will be equipped with 96 GB of GDDR7 memory and is positioned as a mid-to-entry-level GPU for NVIDIA. It targets applications such as AI inference, edge-based deep learning training, image processing, and simulation. However, the supply chain is under pressure: Samsung remains NVIDIA’s sole supplier of GDDR7, and ongoing tight supply conditions could affect the overall production and delivery capacity for the RTX PRO 6000 series.

Given these factors, actual shipment volumes in the coming quarters remain to be seen.

At COMPUTEX in May, NVIDIA showcased collaboration with several ODM and OEM partners to promote MGX AI server models powered by the RTX PRO 6000. These systems feature PCIe Gen5 interfaces and are aimed at enterprise customers expanding from cloud to edge AI applications. NVIDIA is also expected to extend its modular MGX reference design strategy to the Chinese market going forward.