October 25, 2024 – Earlier this year, NVIDIA made waves in the tech industry with the announcement of their new high-performance GPU, Blackwell, in March. The company then proceeded to announce sampling in August, followed by large-scale production. It seemed like everything was going according to plan as NVIDIA began delivering the product to customers in batches.
However, rumors started circulating around mid-year that there were architectural design issues with the Blackwell GPU, potentially causing delays in its production and delivery. Despite these whispers, NVIDIA remained silent on the matter, neither confirming nor denying the allegations.
Recently, in an interview, NVIDIA’s CEO, Jen-Hsun Huang, finally addressed the elephant in the room. He confirmed that the Blackwell GPU indeed had a significant design flaw. While this defect did not affect the GPU’s normal functionality, it significantly impacted the yield rate, making it difficult to produce large quantities efficiently.
Huang revealed that to rectify this issue, NVIDIA had to redesign seven chips, a testament to the severity and complexity of the problem. He emphasized that the fault was entirely NVIDIA’s, stemming from their design and not from TSMC’s manufacturing process. In fact, he praised TSMC for their assistance in overcoming the production challenges and their remarkable speed in restoring the production schedule.
Fortunately for gamers and enthusiasts, Huang clarified that this design flaw does not affect the Blackwell architecture-based gaming GPUs, specifically the RTX 50 series, including both desktop and laptop variants. This announcement is sure to bring relief to those eagerly anticipating the release of these highly anticipated graphics cards.
