Persona AI at the NVIDIA GTC 2026 Keynote

How Persona AI’s welding humanoid got from simulation data to Jensen Huang’s keynote screen in about one month.

Client

Persona AI, builders of humanoid robots for heavy industrial work such as shipyard welding.

Cineshare’s Role

Scene composition, animation integration and cinematic visual production. Director and Omniverse Animator: Pekka Varis.

Tools

NVIDIA Omniverse, Isaac Lab motion data, OpenUSD and a Dell Pro Max Tower T2 with an NVIDIA RTX PRO 6000 Blackwell GPU.

Ambassador & member

  • Dell TechnologiesDell Pro Precision
    Ambassador
  • NVIDIA OmniverseOmniverse
    Ambassador
  • NVIDIA Inception ProgramInception Program
    Member
  • MEFI Metaverse FinlandMetaverse Finland
    Member

The Result in Numbers

0
Keynote audience at GTC 2026
0
Keynote impressions
0
Month from motion data to keynote footage

Who Did What

Precise attribution matters. The robot motion came from Persona AI’s own pipeline; our work made it readable on a keynote screen.

Persona AI’s Part

Developed the robotics motion pipeline, training a reinforcement-learning policy in NVIDIA Isaac Lab from Xsens motion-capture data.

Cineshare’s Part

Scene composition, animation integration and cinematic visual production in NVIDIA Omniverse.

What We Did in the Shipyard Scene

Persona AI wanted to show its welding humanoid working inside an HD Hyundai shipyard scene. We turned the motion data into a path-traced sequence that holds up on a keynote screen.

What Followed

We built an Isaac Sim extension that sets up welding scenes, so scene variations can be produced quickly.

Hardware

Rendered on a Dell Pro Max Tower T2 with an NVIDIA RTX PRO 6000 Blackwell GPU, reviewing path-traced frames locally.

Our Steps

NVIDIA Omniverse

Scene composition, lighting and path-traced rendering.

Isaac Lab Data

Persona AI’s trained motion, brought into the scene as USD animation.

OpenUSD

One scene for robots, environment and effects from different tools.

Dell Pro Max

Local RTX PRO 6000 Blackwell power for fast path-traced iteration.

Further reading

Planning a Launch or Keynote?

Tell us the date, the data you have and who needs to see the result.
What the project needs to achieve
Which assets you have: CAD, USD or Isaac Sim output
When it needs to be ready, and who approves it