Nvidia Delivers Blackwell Ultra Throughput And Cost Gains
Nvidia's February 16, 2026 benchmark data shows GB300 NVL72 systems with Blackwell Ultra GPUs deliver up to 50x higher throughput per megawatt and 35x lower cost per token versus the Hopper platform for low-latency AI workloads. Nvidia attributes gains to architectural changes and TensorRT-LLM optimizations that accelerate attention-layer processing, enabling cloud providers and inference services to cut operational costs and scale agentic AI and coding-assistant deployments.
Key Points
- 1Shows up to 50x throughput per megawatt and 35x lower cost per token versus Hopper
- 2Addresses transformer attention bottlenecks via accelerated softmax and doubled attention-layer processing
- 3Enables cloud providers to reduce inference costs and scale long‑context agentic AI deployments
Scoring Rationale
Strong vendor-validated performance and broad cloud adoption drive high impact; limited independent peer-reviewed benchmarking constrains certainty.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
