Vikash Kumar ✈️GTC

All forms of intelligence co-emerged with a body, except AI We're building a #future where AI evolves as your lifelike digital twin to assist your needs across health, sports, daily life, creativity, & beyond... myolab.ai ➡️ Preview your first #HumanEmbodiedAI

This is who it wants me to see next week: Glad the report was useful! Here are my picks, thinking specifically about what would give you the best signal, content, and connections: TOP 5 COMPANIES TO MEET 1. Wayve (@wayveai) — Booth + Demo Rides CEO Alex Kendall is giving a keynote, and they're offering demo rides of their AV2.0 system. Wayve is doing what nobody else is — end-to-end learned driving, not rules-based like Waymo. This is physical AI applied to the real world at scale. Getting in that car would be incredible content and a window into where autonomous driving actually goes next. Profile: x.com/wayveai 2. Noble Machines (@NobleMachines) — Booths #941, #3303, and more They're LAUNCHING at GTC with 4 booths, 3 robots, and 2 live demonstrations. This is a company at the exact moment you love to find them — right before the world notices. General-purpose robots doing real-world industrial work. Their CEO Wei Ding just did an interview with A3's managing editor. This is the Scoble sweet spot: early, real, and about to matter. Profile: x.com/NobleMachines 3. CoreWeave (@CoreWeave) — Booth 913 + CoreWeave House The most important AI infrastructure company not named NVIDIA. IPO-bound, powering the biggest AI training runs. Their CoreWeave House keynote watch party on March 16 is where the insiders will be. Speakers Chen Goldberg and Corey Sanders are doing a deep dive on full-stack optimizations for large-scale training. If you want to understand the infrastructure layer that makes everything else possible, this is it. Profile: x.com/CoreWeave 4. Modular (@Modular) — Booth #3004 Chris Lattner (@clattnerllvm) — the creator of Swift, LLVM, and MLIR — is attending. They're showing DeepSeek V3.1 running live on Blackwell, plus Mojo GPU programming. Lattner is one of the most important infrastructure architects alive. Modular is building the programming layer that AI will run on for the next decade. The booth will have live Mojo demos — this is deep tech that matters. Profile: x.com/Modular 5. Enchanted Tools (@EnchantedTools) Their Mirokaï humanoid robots will be ROAMING the convention center — not stuck in a booth. This is the most visually striking, shareable thing at GTC. A French robotics company building character-driven humanoid robots that interact with people naturally. Great content opportunity, and a different philosophy from the industrial robots everyone else is showing. Profile: x.com/EnchantedTools TOP 5 PEOPLE TO MEET 1. Vikash Kumar (@Vikashplus) He's flying in specifically for GTC and posted he's "looking forward to meeting founders, researchers, & engineers building in Embodied AI." He's a leading embodied AI researcher (likely from Meta FAIR or similar). The people who work at the intersection of robotics and foundation models are the ones who know where physical AI goes in 5 years. He's open to meeting — take him up on it. Post: x.com/Vikashplus/sta… Profile: x.com/Vikashplus 2. Louis Castricato (@lcastricato) — Overworld AI He's in SF for GTC with "a sneak peek of our upcoming release" and explicitly said he's "happy to compare notes" on world models, interactive AI, and "whatever wild announcements come out this week." World models are one of the most important emerging areas in AI — the idea that AI can simulate and predict the physical world. Overworld is building this. Early look at their release could be a scoop. Post: x.com/lcastricato/st… Profile: x.com/lcastricato 3. Art Sokolov (@ArtSokolov) He's doing a live demo at the Humanoid booth with his KinetIQ system — two robots coordinating together to fulfill requests from humans. This is practical multi-robot coordination, not a canned demo. Built on the NVIDIA Robotics stack. This is the kind of person building the actual software that makes humanoid robots useful. Great demo to see and film. Post: x.com/ArtSokolov/sta… Profile: x.com/ArtSokolov 4. @dr1337 Bringing a CL1 demo to GTC (in SF March 14-22). This post got 32 likes from the Neuroscience list — high engagement for that community. The CL1 appears to be a neural interface / brain-computer interface device. This is the most "future" thing at GTC and exactly the kind of thing nobody else will be covering. He's offering demos — take one. Post: x.com/dr1337/status/… Profile: x.com/dr1337 5. @kimmonismus Traveling to the USA for the first time specifically for GTC — and that post got 49 likes, which means this person has a real following. Sometimes the best perspective comes from someone seeing something with fresh eyes. International AI community members often have insights about what's happening outside the Silicon Valley bubble. Worth a conversation. Post: x.com/kimmonismus/st… Profile: x.com/kimmonismus Honorable mentions: • @TheHumanoidHub — Your guide to the robotics floor. They've mapped every robotics exhibitor. • @sundeep (Groq) — "See you at GTC!" — Groq's inference speed story is important. • Lin Qiao (@lqiao, Fireworks AI CEO) — On the GTC Live pregame panel with Michael Dell and CoreWeave. The "AI is Essential Infrastructure" conversation. • @MidnightCaptl — Had a meeting with NVIDIA and is "so excited" — 77 likes. Clearly knows something. Worth asking what got them so hyped. Have an amazing GTC! 🚀

You might have seen the WuBOT performing at the 2026 Spring Festival Gala; however, most high-dynamic extreme motions you see are executed by overfitted tracking policies. Until now, training a unified policy capable of performing various extreme motions with a high success rate remained an unsolved challenge. We spent an entire year digging into the barrier between general tracking and extreme physical behaviors. After burning through dozens of G1 robots, we finally identified the bottleneck of learning and physical executability. With these discoveries, we developed OmniXtreme: the first general policy that can execute diverse extreme motions, including consecutive flips, extreme balancing, and even breakdancing with rapid contact switches! This capability is achieved by pre-training a flow-based generative control policy and then post-training with actuation-aware residual RL for complex physical dynamics—a step we found critical for successful real-world transfer. This work is a joint collaboration with @UnitreeRobotics. Together, we are pushing the physical limits of humanoid robots. It is incredibly exciting to see a general "robot gymnast" and "robot breakdancer" come to life! It was also our first time publishing a paper with XingXing, which was an enlightening experience. The model checkpoints are now released—we welcome you to play with them! 📦 📄 Paper: arxiv.org/abs/2602.23843 🌐 Project: extreme-humanoid.github.io 💻 Code: github.com/Perkins729/Omn…

❌ NOT TRUE @ChongZitaZhang A finger ≠ a leg In legged locomotion, low gear ratios help with impact tolerance, store kinetic energy, and back-drivability under heavy load. Unlike legs, hand constraints & requirements are different - high torque density - high positional controllability - brutal space constraints The real trade offs for hands are: - Low gear ratio → back drivability, responsiveness, - High gear ratio → torque density, stability, compactness 🟠So is high ratio bad? **Its depends** -- high gear ratios improve static precision & torque density, but reduce dynamic responsiveness & back drivability Infact, biological hands are not low-impedance torque sources either. Like duality of photons, human hands sometime act as precise, while other times acts as force manipulators. Perhaps we need a "Heisenberg Uncertainty Principle" but for Hands.