The training layer for Physical AI.
SIM XR builds the data infrastructure for humanoid robotics. Robotics teams need millions of teleoperation demonstrations to train manipulation policies — data today is captured by in-house teams in lab studios at roughly $50K–$100K per fully-loaded operator per year. It does not scale to the millions-of-demos requirement modern Vision-Language-Action models need.
Our bet: the workforce that records this data is already deployed. Tens of millions of Quest 3, Vision Pro, and Pico 4 Ultra headsets sit idle in living rooms. An operator at home wears their own headset and teleoperates a simulated humanoid robot inside a photoreal 3D-Gaussian-Splat reconstruction of the customer's real environment. The hand-motion data is real. The action stream is real. The scene is rendered. The robot is simulated. We ship customer-shaped demonstration datasets in days, not months.
How SIM XR makes money.
Per-episode pricing on customer-shaped demonstration datasets. A target customer is a humanoid robotics team training manipulation policies — they pay for completed, quality-validated episodes recorded against their robot and their environment. The two structural advantages we sell against in-house teleop programs are cost per episode (consumer hardware + at-home operators) and scenario coverage (one captured environment multiplied into many randomized training scenes via object, task, and layout variation).
Who pays
VLA fine-tuners, behavior-cloning teams, simulation-to-real research groups.
What we sell
Recorded in the customer's 3DGS-reconstructed environment with their robot's embodiment and the operator network's hand-motion data.
How it scales
Operators connect from existing consumer headsets. One captured scene yields many randomized training variants.
What's live today
One-click in-browser WebXR connect to live VR scenes on any headset. 9-scene catalog. simxr.app
Georgy Molodtsov
Molodtsov
Produced and curated XR experiences across Europe and CIS. Director of JMJ OXYMORE (400K+ visits) and producer on Notre‑Dame VR (75M+ viewers). Founder of Film XR (Estonia/France).
Hands-on research in 3D Gaussian Splatting and Volumetric Video for XR / Film / Animation. The core technology behind SIM XR's customer-environment-as-a-scene capability.
Member of ATAS Emerging Media Peer Group. Mentor, Venice Biennale College Cinema — Immersive (2025). Fulbright Graduate Fellow.
Founding team.
SIM XR was founded by three operators across XR, consumer product, and 3D computer vision. Direct LinkedIn profiles below.
Validated by NVIDIA and AWS.
SIM XR is a member of the NVIDIA Inception Program (since April 2026) — the technical-validation startup network for the broader NVIDIA ecosystem, aligned with our Isaac Sim 6 + CloudXR stack. We are also an AWS Activate Member via NVIDIA Inception Global, running the production simulation stack on AWS EC2 GPU instances in eu-north-1.
Get in touch.
Robotics teams that need demonstration data, engineers and researchers who want to collaborate, and program partners: request early access via the main page, or email gm@simxr.tech.