Nvidia (NVDA) has entered into a multiyear partnership with Hark to advance personalized agentic AI, including a technical collaboration and gigawatt-scale compute capacity on the Vera Rubin platform, Hark said Thursday.
Hark said Nvidia's AI infrastructure supported the development of its Hark Handoff computer use model, the distributed inference behind Hark's platform upon release, and training of its foundation models into the future.
Hark said it is using other Nvidia technologies, including the Megatron stack for model training, Dynamo for inference, and NVSentinel for GPU cluster resiliency.
Shares of Nvidia were up about 8% in Thursday trading.
Price: $226.37, Change: $+16.71, Percent Change: +7.97%