Can We Close the Sim2Real Gap?

​As embodied systems move from demos to deployment, the central question is becoming more technical and more controversial: can simulation, synthetic data, and world models replace enough real-world data to make Physical AI scale?

​We’ll discuss what works, what breaks, and what it actually takes to turn simulation into real-world performance.

​Join Savant and Nebius for an evening with these guest speakers:

• ​Jonathan Lwowski, Field CTO at Nebius

• ​Harry Mellsop, Founder, Antioch

• ​Evan Morikawa, VP, Engineering, Generalist AI

​Ashis Ghosh, Founding Robotics Engineer at The Bot Company and Venture Partner at Savant, will moderate the panel discussion. We will explore what it actually takes to move Physical AI systems from simulated environments into real-world deployment.

​Topics will include:

• ​The emerging simulation + data + VLA + RL training paradigm for Physical AI

• ​Whether world models can learn physics, contact, causality, and proprioception well enough to transfer into the real world

• ​The role of teleoperation, synthetic data, autonomous data collection, and fleet learning in closing the Sim2Real gap

• ​How teams should think about evaluation, failure modes, deployment loops, and real-world robustness

• ​Where enduring value may accrue across data, evals, simulation, deployment tooling, compute, and infrastructure

​This event is designed for robotics researchers, embodied AI founders, frontier lab builders, infrastructure teams, and investors actively thinking about the next generation of Physical AI.

​Agenda:

• ​5:30 PM — Doors Open + Light Bites

• ​6:15 PM — Welcome Remarks

• ​6:30 PM — Fireside Chat

• ​7:00 PM — Audience Q&A

• ​7:25 PM — Closing Remarks

• ​7:30 PM — Networking

• ​9:00 PM — End of event

​About the Host:
Savant is a community of technologists embarking on good quests, ideas that are hard, consequential and civilizational-defining. We help top founders explore & start new companies in Physical AI through a powerful network of peers, a dedicated hardware lab, and founder-led support – often before we ever write a check. Together, let's move civilization forward!

​Nebius is a full-stack AI cloud built for teams training, deploying, and scaling advanced AI systems. Its infrastructure supports large-scale GPU workloads across training, inference, synthetic data, and Physical AI — making it especially relevant for builders working on simulation, robotics, and real-world deployment.