NVIDIA's latest Blackwell GPUs have set new records in both AI training performance and scalability in the MLPerf 6.0 benchmarks, the company announced. The results solidify the chipmaker's lead in the race to build next-generation AI infrastructure.
Record-breaking performance
MLPerf is the industry-standard benchmark suite for measuring machine learning system performance. In version 6.0, Blackwell-based systems delivered the fastest training times across a range of AI models, from computer vision to natural language processing. The GPUs also showed strong scalability, handling larger workloads without efficiency losses as more units are added.
Dominance in AI infrastructure
The latest scores reinforce NVIDIA's position as the dominant supplier of hardware for training large AI models. Competitors have struggled to match the combination of raw compute power and energy efficiency Blackwell offers. With demand for AI compute continuing to surge, these benchmark results are likely to strengthen NVIDIA's grip on the data center market.
Pricing and availability details for Blackwell GPUs have not been disclosed. NVIDIA is expected to share more specifics at its next major product event.
