Hark Partners with NVIDIA for Gigawatt-Scale Compute to Power Personalized AI
adcock_brett · x · 2026-08-27
AI startup Hark announced a multi-year strategic partnership with NVIDIA to advance personalized agentic AI, including deep technical collaboration and gigawatt-scale compute capacity on the next-generation NVIDIA Vera Rubin platform.
Key Details:
- Compute Capacity: Hark secures gigawatt-scale compute on NVIDIA's upcoming Vera Rubin platform to support frontier model training and distributed inference at scale.
- Tech Stack Integration: Hark will leverage NVIDIA's Megatron stack for training, Dynamo for inference, Nemotron pretraining data, and NVSentinel for GPU cluster resiliency.
- Product Vision: Hark is building multimodal systems that remember user identity and plans. NVIDIA's infrastructure powers its recently released "Hark Handoff" computer use model and will support the training of future foundation models.
Related event: Hark Partners with NVIDIA for Gigawatt-Scale Compute(2 posts)→
More from Companies & People
- CuspAI CEO Chad Edwards lands on TIME100AI list as startup raises $450M for AI materials discovery — jonkhler · 2026-08-28
- OpenClaw maintainers discuss the project's growth and future — film_girl · 2026-08-28
- OpenClaw Community Acts as a Gateway to Open Source and Coding — martinwoodward · 2026-08-28
- Metriport raises $26M to fix healthcare's data problem, deployed at Amazon and GM — ilyasu · 2026-08-28
- Google deserves credit for deploying AI to consumers at scale for free — andreisavu · 2026-08-28
- Alex Rives, pioneer of protein language model ESM, named to TIME100 AI — proteinrosh · 2026-08-28