Nvidia expects its new AI models capable of video creation and human-like voice interaction will drive demand for its graphics processors, CEO Jensen Huang said. “There's a lot of information in life that has to be backed up by video and physics, so that's the next big thing,” he told Reuters.
“We have 3D video and there's a ton to learn from that, so these systems are going to be pretty large,” Jensen Huang added.
Nvidia's H200 was first used in OpenAI's GPT-4o, a multimodal model capable of lifelike voice conversations that can interact between text and images. Other Nvidia customers, Google DeepMind and Meta, have also released AI image or video generation platforms.
The announcement came after the chipmaker forecast quarterly earnings that beat expectations by a wide margin after its data-center division's sales grew more than fivefold in the first quarter, sending Nvidia shares up 9 percent.
“Demand is broad-based and large language models are becoming increasingly multi-modal, needing to understand not just video but text, speech, 2D and 3D images,” said Darren Nathan, head of equity analysis at Hargreaves Lansdown.
“(Video generation) is certainly one of the powerful and proven use cases for AI, and it's expanding beyond just content creation,” he said.
Additionally, Nvidia Chief Financial Officer Colette Kress said Tesla has expanded the cluster of processors it uses to train AI for self-driving cars to about 35,000 H100s.
Get updated on India News, Elections 2024, Elections 2024 Date, Latest News & Top Stories from India & Around the World.
