Nvidia, already the market leader in providing high-end processors for generative AI use, will release an even more powerful chip as the demand to run large AI models continues.
The company announced the availability of the GH200 super chip, which Nvidia said can handle “the most complex generative AI workloads, spanning large language models, recommender systems and vector databases.”
The GH200 will have the same GPU as the H100, currently Nvidia’s most powerful and popular AI offering, but triple the memory capacity. The company said systems running on GH200 will start in the second quarter of 2024.
Nvidia did not reveal the price for the GH200; the H100 line currently sells for roughly $40,000.
Complex AI models require powerful GPUs so the system can make the computations necessary to generate text or a photo of a horse in the style of Banksy. Running these models requires a ton of processing power, and even with Nvidia’s H100 chips, some have to “break up” the models among other GPUs just to run.
Nvidia has a near monopoly in generative AI-capable GPUs. Cloud providers like AWS, Azure, and Google all use Nvidia’s H100 Tensor Core GPUs and tack on services to help clients get projects using large language models up and running to differentiate themselves.
Nvidia also faces competition from AMD, which wants to ramp up production for its own AI GPU in the fourth quarter of this year.