The NVIDIA H100 is a data center GPU based on the Hopper architecture, released in 2022. It became the foundational hardware of the generative AI boom due to its fourth-generation Tensor Cores and a new Transformer Engine, which dramatically accelerated the training and inference of large language models. The Blackwell architecture, announced in 2024, is its successor. Blackwell GPUs, such as the B200, feature fifth-generation Tensor Cores, a second-generation Transformer Engine, and a significant increase in memory bandwidth, and are specifically designed to handle the massive computational demands of trillion-parameter AI models.
The H100 is named after Grace Hopper, a pioneer of computer programming. The Blackwell architecture is named after David Blackwell, a mathematician who made significant contributions to game theory and statistics. These naming conventions reflect NVIDIA's practice of honoring influential scientists and mathematicians in their data center GPU architectures.
The H100 GPU created an unprecedented global demand for compute, becoming a scarce and highly sought-after resource for companies and researchers building AI models. The Blackwell platform is expected to continue this trend, enabling the development of even larger and more capable AI systems. These GPUs are the engines driving the rapid advancements in generative AI, from large language models to complex scientific simulations.