embodied-ai
(38)
Wayve Raised $2.5 Billion to Teach One AI Driver Every Road
The London company skipped hand-coded rules and high-definition maps for a driving model that learns from experience. Its harder test begins now: turning an elegant research bet into ordinary cars and rides people can actually buy.
Lei Xu Is Teaching Cars to See - Then Asking What Sight Is For
The former Tesla Vision engineer built Nullmax around a stubbornly practical idea: autonomous driving only matters when the software can leave the demo, fit the car and survive the road.
Ananth Kashyap Is Teaching AI to Get Its Hands Dirty
The Sureform founder started with a high-school iOS app and arrived at a harder question: how do you teach AI what real work feels like? His answer begins in the wonderfully untidy places where people already know what to do.
Allen Pan Is Teaching Factories to Change Their Minds
After engineering autonomous systems at General Motors and advanced factory automation at Tesla, Allen Pan is betting that the useful factory of the future will not merely run. It will learn, adjust and run again.
Unitree G1 Makes a Humanoid Buyable. Making It Useful Is the Expensive Part.
The four-foot robot can walk, dance and recover its balance for the price of a used car. Researchers should read the configuration sheet before they read the choreography.
Apurva Shah Builds Worlds Where Robots Can Fail First
After two decades making digital worlds believable, the former Pixar technical leader is using simulation to make machines more dependable in the real one.
Waymo Has Left the Lab. Now It Has to Teach Cities to Trust an Empty Driver’s Seat
Alphabet’s robotaxi company has turned autonomous driving into a service people can summon. Its next test is less cinematic and more difficult: making safety, fleet economics and public trust work in city after city.
Dyna Robotics
Dyna Robotics is a Redwood City, California startup building embodied-AI foundation models that let robots perform dexterous, real-world manipulation tasks around the clock. Founded in 2024 by repeat entrepreneurs Lindon Gao and York Yang (who sold Caper AI for $350M) and former DeepMind researcher Jason Ma, the company's DYNA-1 model runs commercial deployments in laundries, factories, restaurants and gyms. Dyna raised a $23.5M seed and a $120M Series A in 2025, reaching a valuation above $600M with backing from NVIDIA, Amazon, Salesforce Ventures, Samsung Next and CRV.
Robots in the Real World: Four CEOs on the Machina Stage
At the Machina conference in Paris, All-In host Jason Calacanis sits down with the leaders of four of the world's top robotics companies—ANYbotics, 1X, Boston Dynamics, and Agility Robotics. Across four rapid-fire interviews, the CEOs explain why four-legged 'dog' robots conquered industrial inspection first, why humanoids are finally doing useful work, how teleoperation and internet-scale video data are training the next generation of machines, and where the real money and the real dangers lie—from explosive oil rigs to weaponized robots coming out of China. The through-line: robots are no longer research demos, they're deployed assets measured by ROI, and 'hard takeoff' where robots build robots may be as little as three years away.
Zakariea Sharfeddine
Zakariea Sharfeddine is the co-founder and CEO of InLoop Robotics (YC P26), a startup building AI-native warehouse fulfillment powered by robotic arms you hire by the month. A KIT-trained informatics engineer who did machine learning work at Bosch and BMW and reinforcement learning research at RoboTUM, he now leads a team deploying bimanual robots that pack, kit, and fulfill orders, with remote human operators stepping in on edge cases so the line never stops.
6thSense
6thSense builds wearable tactile sensing rigs that record how human hands actually touch the world - pressure, grip, contact, force, depth, vision and 3D hand pose - out in real homes rather than labs. It packages that multimodal, model-ready data and sells it to teams training the next generation of dexterous robots, positioning itself as a 'nervous system for physical AI.' The company is part of Y Combinator's Summer 2026 batch and is based in San Francisco.
InLoop Robotics
InLoop Robotics deploys intelligent bimanual robotic arms into warehouses as a service - handling packing, kitting, and order fulfillment for a flat monthly fee with no upfront hardware cost. Instead of waiting for a perfect model, InLoop ships imperfect policies into real warehouses, detects failures in real time, recovers automatically when it can, and hands off to a remote human operator when it can't. Every intervention becomes training data, so the system keeps getting faster and cheaper with each order.
Remy AI
Remy AI is a San Francisco robotics startup in Y Combinator's Winter 2026 batch building dexterous, AI-powered robots for e-commerce third-party logistics (3PL) warehouses. Its core bet is data: the team captures real human dexterity on real job sites using head-mounted cameras paired with wearable tactile sensors, recording synchronized vision, motion, and contact force. That data trains robotics foundation models so a bi-manual mobile robot can pick, pack, kit, and receive across changing inventory without reprogramming, deploy in weeks rather than months, and target roughly half the cost of legacy automation via a robot-as-a-service model.
Sensei Robotics
Sensei Robotics (YC S24) is a San Francisco startup building the data layer for robot learning. It pairs a sub-$300 sensorized exoskeleton arm with a distributed network of trained human operators - 'Senseis' - who perform real-world manipulation demonstrations on demand. Robotics and AI teams submit a task like folding cloth or loading a dishwasher, and Sensei returns curated, labeled human-demonstration data. The company positions itself as 'the Scale AI for robotics training data,' claiming roughly a tenth of the cost and twice the speed of conventional teleoperation setups.
Sureform
Sureform is a San Francisco startup (Y Combinator X25) that collects real-world, multimodal human data to train robots and world models. It equips workers at hotels, warehouses, restaurants, construction sites, and laundromats with head-mounted cameras, stereo depth rigs, and handheld grippers that capture time-synced video, audio, pose, depth, and tactile data while they do their jobs. Each session is delivered to robotics teams as structured, synchronized, training-ready data.
Steve Xie
Steve Xie is the founder and CEO of Lightwheel, a physical AI infrastructure company building simulation and synthetic data systems that teach robots to learn and operate in the real world. A physicist by training with a Columbia PhD, he led autonomous driving simulation at Cruise, NVIDIA, and NIO before starting Lightwheel in 2023. The company supplies SimReady 3D assets, egocentric human data, and robot evaluation tools to labs including NVIDIA, Google DeepMind, and Figure, and raised roughly $137.5 million in 2026 to expand its data and evaluation infrastructure.
Apptronik
Apptronik is an Austin-based robotics company building Apollo, a general-purpose humanoid robot designed to work alongside people in manufacturing, logistics and other physically demanding settings. Spun out of the University of Texas at Austin's Human Centered Robotics Lab in 2016, the company grew from custom actuator work for NASA into one of the most heavily funded humanoid robot developers, with commercial pilots at Mercedes-Benz and GXO and an AI partnership with Google DeepMind's Gemini Robotics team.
LE Robotics
LE Robotics is a Chengdu-founded industrial robotics company building AI-driven welding and cutting robots that combine 3D vision, SLAM autonomous navigation and welding-specific AI algorithms. Its systems locate weld seams in seconds and operate without pre-programming or manual teaching, letting robots 'think and act like workers.' Spun off from the Loyalty Enterprises industrial group in 2022 by CEO Haojie Hu, the company serves Fortune 500 manufacturers and large state-owned enterprises across nuclear, oil and gas, rail and heavy machinery, and is expanding internationally through subsidiaries in Canada, the US, Brazil and Singapore.
Lightwheel
Lightwheel is a Physical AI infrastructure company building the data layer that teaches robots to work in the real world. It produces physically accurate SimReady 3D assets, egocentric human demonstration data, and evaluation platforms so robots can learn safely and cheaply in simulation before they are deployed. Founded in 2023 by former autonomous-driving-simulation lead Steve Xie, Lightwheel partners with NVIDIA and supplies simulation assets and synthetic datasets to leading robotics teams.
Nullmax
Nullmax is an AI company building vision-only, map-free autonomous driving software for mass-produced cars. Founded in Silicon Valley in 2016 by former Tesla Autopilot engineer Lei Xu and headquartered in Shanghai, it develops MaxDrive, a chip- and sensor-agnostic driving stack, and Nullmax Intelligence (NI), an end-to-end, multimodal system that reads the road with cameras instead of lidar or high-definition maps. Its technology ships in production vehicles from major Chinese automakers.
Scout AI
Scout AI is a Silicon Valley defense-tech company building Fury, a Vision-Language-Action foundation model that turns unmanned robots - ground, air, sea, and space - into autonomous agents that perceive the world, understand natural-language commands, and act in GPS- and comms-denied environments. Founded in 2024 by Colby Adcock and Collin Otis, the company raised a $100M Series A in April 2026 (which it billed as the largest defense-tech Series A in U.S. history) after emerging from stealth in 2025 with a $15M seed and early Department of Defense contracts.
Anvil Robotics
Anvil Robotics is a Palo Alto-based hardware and software company building a composable platform for Physical AI. It sells ready-to-run robot devkits - most notably the open-source OpenARM bimanual manipulator and the OpenYAM arm - bundled with teleoperation, data-collection, and training tooling so teams can train end-to-end robot policies on day one instead of spending six months sourcing and wiring parts. Backed by a $6.5M seed round, Anvil manufactures in-house in Taiwan and has shipped hundreds of robots to customers including Nvidia, Google, and university labs.
Dexmate
Dexmate is a Santa Clara robotics startup building Vega, a general-purpose mobile robot with high-payload arms, dexterous hands and a foldable torso that reaches over seven feet. Founded in 2024 by a team out of MIT, UCSD and CMU, the company pairs custom hardware with a data-driven 'flywheel' learning approach - combining simulation and real-world data - to tackle dexterous manipulation for logistics, manufacturing, retail and research. Backed by NVIDIA and LG Technology Ventures, Dexmate has opened preorders for Vega at $89,999.
Integral AI
Integral AI is a Japan-US startup building foundation world models - AI systems meant to learn, plan, and act in the real world the way people do, with an emphasis on energy efficiency and continuous, autonomous skill learning. Founded in 2021 by ex-Google AI leaders Jad Tarifi and Nima Asgharbeygi, the company positions its interactive world models as infrastructure for an AI-based economy, spanning robotics, universal simulators, and multimodal interfaces. In December 2025 it made headlines by claiming to have tested the world's first AGI-capable model, TOWA.
Mbodi AI
Mbodi AI is a New York robotics-software company building an embodied AI platform that lets people teach industrial robots new skills through plain natural language and short demonstrations - no code and no specialized engineers. Founded in 2024 by two former Google Public DNS engineers, the platform combines generative AI, agent orchestration and symbolic reasoning to turn everyday instructions into precise, reliable robot actions that run in production within minutes and transfer across an entire robot fleet. Mbodi is a Y Combinator company with a joint commercialization partnership with ABB Robotics.
Proception
Proception is a Palo Alto and Mountain View robotics company building the dexterous robotic hand that humanoids are missing. Its flagship ProHand 1.0 packs 22 degrees of freedom, tendon-driven fingers and skin-like tactile sensors into a hand designed with input from surgeons - close enough to a human hand that robots can learn directly from human demonstration. A companion sensor glove, ProGlove, captures real human manipulation data at scale to train those movements. Founded in 2024 by former Tesla Optimus engineers Jay Li and Jack Xu, the company raised an $11M seed led by First Round Capital and is shipping its first hands to research labs and robotics companies.
Jackie Wu
Jackie Wu is the founder and CEO of Corvus Robotics, a company that builds fully autonomous drones that fly indoors and count warehouse inventory without beacons, GPS, Wi-Fi, or a human at the controls. A Northwestern-trained roboticist who spent time inside warehouses on four continents, he grew tired of watching people climb ladders with clipboards and barcode guns, and decided to teach a drone to do it instead. Corvus designs and manufactures its drones in California, went through Y Combinator, and raised a Series A in 2024 to push warehouses toward a lights-out future.
Anyware Robotics
Anyware Robotics is a Fremont, California startup building deployable physical AI for industrial work. Its first product, Pixmo, is a mobile robot that drives into a shipping container or trailer and unloads boxes - up to about 1,000 cases per hour - using a force-sensing collaborative arm and an AI stack the company calls AnywareOS. Founded in 2023 by former UC Berkeley researchers and FANUC engineers, the company aims to take over the back-breaking jobs that warehouses can no longer reliably staff.
Jad Tarifi
Jad Tarifi is the co-founder, CEO and Chief AI Scientist of Integral AI, a Tokyo- and Silicon Valley-based startup building energy-efficient, interactive world models aimed at scalable artificial general intelligence and embodied robotics. Before founding the company in 2021, he spent nearly a decade at Google, where he founded and led the company's first generative AI team and worked on neocortex-inspired learning from limited data. Born and raised in Lebanon, he holds a PhD in computer science and AI from the University of Florida and an MS in quantum computing. In December 2025 he drew global attention - and skepticism - by claiming Integral AI had built the world's first AGI-capable model, TOWA. He is also the author of the four-volume 'Freedom Series' on AI and human potential.
Sebastian Peralta
Sebastian Peralta is a physicist-turned-roboticist and co-founder of Mbodi AI, a New York startup teaching industrial robots new skills through plain language and quick demonstrations instead of weeks of reprogramming. A former core engineer on Google Public DNS (8.8.8.8) with a triple major in electrical engineering, computer science, and physics from UPenn, he builds embodied-AI systems where one robot's lesson instantly becomes the whole fleet's. Mbodi won ABB Robotics' global AI Startup Challenge, reached the Startup Battlefield Top 20 at TechCrunch Disrupt 2025, and is rolling out factory deployments with a Fortune 100 consumer-goods partner.