NVIDIA’s upcoming earnings report is already generating buzz due to issues with the company’s new Blackwell AI GPUs. Analyst firm KeyBanc has reported that design flaws in the B100 AI GPU have led to a significant setback, causing NVIDIA to effectively cancel the chip. The company is facing delays in shipping the B100 due to the need to redesign the Blackwell tile, a process that will take an estimated 3 months.
KeyBanc explains that NVIDIA will prioritize the ramp of the B200 for hyperscalers (large-scale cloud providers) and has decided to replace the B100 with a lower-cost/performance GPU, dubbed the B200A, which will be targeted at enterprise customers. The B200A, born from the necessity to address the issues with the B100, is expected to be released in the second half of 2025, a significant delay from the original Q4 2024 shipping plan.
The delay in the Blackwell architecture, coupled with the B100’s cancellation, highlights the challenges NVIDIA is facing in bringing its next generation of AI GPUs to market. The company is currently working to ramp up production of the B200 for hyperscalers, but limited supply is expected in Q4 2024 due to production constraints. As a result, the B200A will be introduced in the second half of 2025 to address the needs of enterprise customers.
This news sheds light on the competitive landscape in the rapidly growing AI hardware market. NVIDIA’s rivals, such as Intel and AMD, are also developing their own AI chipsets, creating a fierce race for market share. NVIDIA’s delays in releasing its Blackwell-based GPUs present an opportunity for competitors to gain ground. It remains to be seen how this situation will play out in the long run and what impact it will have on NVIDIA’s overall dominance in the AI chip market.
