Sunday AGI Meetup: Scaling Law in Robotics
“In the next 2–5 years, robotics will uncover its own scaling laws – similar to those seen in LLMs – by analyzing how model size, real‑world data, simulation data, and compute affect performance.” - Jim Fan, NVIDIA Director of Robotics
Data scaling is always the most discussed challenge when it comes to building the robotics foundation model.
Where are we, really, in building the next frontier of AI - general-purpose robots?
What can we learn from the experience scaling LLM?
We're hosting this invite-only roundtable for the researchers and builders in robot learning and LLM to move past the hype and trade notes on the hard, open questions.
We’ll dig into:
The Scaling Bottleneck: Is the key constraint data quality (diverse, multi-modal experiences) or data quantity?
The Right "Model": Is VLA the answer? Is World Model the next bet? Is action data overrated? Where does simulation fit?
What can we learn from scaling LLM and video-gen models? Do we need to rethink our approach if we can scalably obtain 1000x robotics data?
What does it take to win this game?
Who’s in the room:
A curated group of 12-15 leading researchers and builders from academia and industry shaping the future of embodied AI.
We welcome researchers who have built LLM: let's see how we spark each other!
Format & vibe:
Quick intros → open roundtable → social
No slides, no press; Chatham House Rule
12–15 people, invite-only. Address shared upon confirmation.
Hosted in San Francisco this Saturday afternoon, co-hosted with Foundation Capital.