NVIDIA’s much-anticipated Blackwell AI GPUs are experiencing a significant setback, potentially impacting the production of its upcoming GeForce RTX 50 series gaming GPUs. According to a report by The Information, picked up by Business Korea, NVIDIA engineers have encountered challenges with the Blackwell chips during testing. These chips are reportedly failing in high-voltage environments, common in data centers. This revelation has sparked a blame game between NVIDIA and its fabrication partner, TSMC.
TSMC claims that NVIDIA’s rushed production of the Blackwell GPUs is the root cause of the issues. In contrast, NVIDIA maintains that the defects stem from TSMC’s advanced packaging technology. The Information’s report speculates that NVIDIA could potentially shift the production of its Blackwell-based GeForce RTX 50 series gaming GPUs to Samsung. This move could offer NVIDIA a 20-30% cost reduction, but it raises concerns considering the underwhelming performance of the Samsung-manufactured GeForce RTX 30 series GPUs.
While NVIDIA might use this news as leverage to negotiate better terms with TSMC, the possibility of a full shift to Samsung for the RTX 50 series remains uncertain. Furthermore, NVIDIA’s heavy reliance on TSMC’s most advanced silicon wafers for the Blackwell AI GPU family could lead to potential stock shortages in the coming months, especially if NVIDIA chooses to pursue a Samsung partnership.
With CES 2025 approaching in less than 90 days, NVIDIA appears to be nearing the launch of its GeForce RTX 50 series GPUs. However, the current production hurdles add another layer of complexity. While a sudden switch to Samsung might not be on the table, the report from The Information, a reputable source, highlights the precarious situation surrounding the RTX 50 series and NVIDIA’s future GPU strategy. This ongoing saga is sure to keep industry watchers glued to their screens, awaiting the resolution of this critical production dilemma.