20 companies | 9 countries | $3.4B+ raised combined
From warehouse floors in Stuttgart to shipyards in California, a new generation of startups and SMEs is deploying intelligent machines that can see, reason, and act in messy, chaotic, and unpredictable environments.
For these start-ups and SMEs, the next generation of physical intelligence represents both a profound opportunity and a pressing question: which players will rewrite the four pillars of physical intelligence: the humanoid systems, industrial autonomy, synthetic training and seamless integration layers?
These companies are more than well-funded bets; they are signals toward the industry shift to humanoids that operate outside data centers and inside the physical world alongside humans.
EUROPE
Sereact - Stuttgart, Germany 🇩🇪
Sereact is redefining how robots interact with objects via its PickGPT platform. By applying Vision-Language-Action (VLA) models to robotic picking, warehouse robots can handle entirely new, unseen objects without retraining. For SMEs running fulfillment operations, Sereact offers a compelling path to flexible, AI-driven automation that adapts as SKUs change (a chronic pain point for traditional robotic picking systems).
Wandelbots - Dresden, Germany 🇩🇪
Wandelbots is tackling the software silo in automation, where most industrial robots typically require specialist engineers and weeks of setup time. Its no-code Wandelbots OS platform allows manufacturers to teach robots new tasks quickly using intuitive interfaces rather than lines of code. Having raised over $123M, Wandelbots is now expanding its ecosystem to include generative AI features that allow robots to self-program from simple human demonstrations.
Inbolt - Paris, France 🇫🇷
Inbolt builds AI-powered 3D vision software that gives industrial robots the ability to perceive and respond to dynamic environments in real time. Their GuideNOW technology company enables robots to locate and manipulate parts in random orientations, dramatically reducing setup time and integration costs and solving one of the oldest frictions in factory automation.
Enchanted Tools - Paris, France 🇫🇷
While most humanoid robotics companies target factories, Enchanted Tools is engineering socially intelligent robots for human-facing environments such as hospitals, hospitality, and retail. Its signature Mirokaï robot features an emotive display face and is designed to navigate real-world spaces and interact with people with social fluency.
Generative Bionics - Milan, Italy 🇮🇹
Founded in 2024 by Daniele Pucci, this startup has secured a landmark €70M seed round. Drawing on biomechanics and generative design principles, Generative Bionics is architecting humanoid robots for physically demanding industrial tasks, engineering bodies that move more naturally than typical robotic platforms.
Mimic Robotics - Zurich, Switzerland 🇨🇭
Mimic Robotis is tackling dexterous manipulation, the ability to handle objects with the nuance and adaptability of a human hand, head-on. By combining advanced hardware design with learned manipulation policies, they are unlocking fine-motor applications in assembly, logistics, and laboratory automation. Breakthroughs in dexterous manipulation will unlock entire categories of physical AI applications that are currently out of reach.
ASIA-PACIFIC
RLWRLD - Seoul, South Korea 🇰🇷
RLWRLD is building foundation models for physical AI systems. Using reinforcement learning and real-world industrial data, they train robots to generalize across a wide range of tasks and environments. Their recent $41M seed extension reflects strong investor conviction in their data-first approach to robotics manufacturing.
Galexea Dynamics - Shenzhen, China 🇨🇳
With ~$434M raised and a fresh Series B unicorn valuation, Galaxea is architecting a full-stack platform for humanoid and embodied AI, spanning hardware, simulation, environments, and AI models. Their G0 and G0 plus dual-system models power the R1 pro humanoid, enabling it to think slowly for complex reasoning and act fast for real-time task execution.
Spirit AI - Shenzhen, China 🇨🇳
Founded in 2024 by former Rokae CEO Han Fengtao, the company has quickly secured unicorn status with a $290M seed round. Their Spirit v1.5 embodied AI model uses a VLA architecture to power their general-purpose humanoid robots capable of operating in unstructured environments.
Astribot - Shenzhen, China 🇨🇳
Astribot’s S1 humanoid is targeting both consumer and industrial markets. Its humanoid has demonstrated impressive dexterous manipulation capabilities in public showcases, designed to perform everyday tasks spanning cooking, cleaning, and sorting. Their dual-market strategy is a bold bet that home and factory robots will eventually converge.
Linkerbot - Shenzhen, China 🇨🇳
If the humanoid robot is the body, Linkerbot is building the hands. This Beijing-based startup has turned what was once a fragile, expensive lab component into a mass-produced industrial product. Its Linker Hand series spans 11 to 42 degrees of freedom across multiple actuation technologies, with flagship models achieving sub-millimetre precision suitable for complex assembly and surgical assistance tasks. Recently raising nearly $217M, Linkerbot is the first company worldwide capable of mass-producing over 1,000 dexterous units per month.
Robotera - Shenzhen, China 🇨🇳
Robotera has emerged as a powerhouse in the end-to-end embodied intelligence race, reaching a valuation of over $1.4B in early 2026. The company’s flagship L7 bipedal humanoid and XHAND1 dexterous hand are powered by a sophisticated VLA architecture that enables robots to generalize across complex tasks like pharmaceutical sorting and cross-border logistics. By integrating "slow" cognitive planning with a "fast" motor-action system, RobotEra is transitioning from laboratory demonstrations to high-volume commercial deployments in partnership with global logistics leaders.
NORTH AMERICA
Genesis AI - Palo Alto / Paris 🇺🇲🇫🇷
Genesis AI is building the simulation-to-reality infrastructure that the next generation of physical AI systems will depend on. Their platform allows robot developers to generate physics-accurate synthetic training data, solving the data scarcity problem for physical AI. Having raised $105M at the seed stage, high-fidelity simulation may be the quiet backbone of the physical AI era.
Persona AI - San Francisco, California 🇺🇲
Persona AI is targeting one of the most physically demanding and labour-short sectors in the global economy: shipbuilding. The start-up has launched pilot programs in Louisiana Shipyards, using humanoid welders and inspectors to navigate the confined, structurally complex environments of naval and commercial shipyards.
Collaborative Robotics - Santa Clara, California 🇺🇲
Founded by former Amazon Robotics VP Brad Porter, Cobot is bringing AI-native design principles to autonomous mobile robots for warehouses and manufacturing environments. Their Proxie platform is built for genuine human-robot collaboration, designed to work fluidly alongside people without the need for safety barriers or separate zones.
Dyna Robotics - San Francisco, California 🇺🇲
Dyna Robotics is pursuing a general-purpose foundation model for robotic behaviour rather than task-specific programming. Dyna’s models allow robots to execute complex manipulation tasks from natural language instructions or visual demonstrations, without extensive per-task engineering. Its trajectory reflects the broader convergence of LLMs and physical robotics.
The Bot Company - San Francisco, California 🇺🇲
The Bot Company is building home robots designed to perform domestic tasks like tidying and food preparation. Their home robots demonstrate the fluency and adaptability that previous generations of home robots conspicuously failed to deliver. Having raised over $300M at Series B, they are doubling down on the conviction that foundation model breakthroughs have finally made home robots commercially viable.
Viam - New York 🇺🇲
Viam is the “Microsoft of machines.” Founded by former MongoDB CTO Eliot Horowitz, it is building a modular, cloud-connected Robot OS that allows any team to build, deploy, and manage robots without deep robotics expertise. By simplifying the SDK and data pipeline layer for physical machines, it positions itself as a strong infrastructure for a generation of physical AI applications.
Bedrock Robotics - Austin, Texas 🇺🇲
Bedrock Robotics is bringing autonomous AI to construction. While the construction sector has historically been resistant to automation due to the variability and physicality of the work, Bedrock believes that AI has now reached the capability threshold to change that equation. Its AI systems enable massive equipment like excavators, graders, and other machinery to operate with minimal human supervision on complex job sites.
Machina Labs - Los Angeles, California 🇺🇲
Machina Labs is reimagining metal manufacturing for the aerospace and defence sectors using AI-driven robotic forming. Using a two-armed robotic system and real-time AI to shape metal sheets into complex 3D geometries, they can produce parts in days rather than producing parts in months at an expensive cost. Machina Labs is now working with key aerospace primes and defence contractors to produce parts that were previously impossible or uneconomical to manufacture at low volumes.
The Companies Powering the Physical AI Revolution
Physical AI is not a single product category; it is a matrix of humanoids, autonomous systems, robotics infrastructure, training environments, and deployment frameworks. The 20 companies profiled collectively span hardware, foundation models, simulation infrastructure, and vertically-focused applications. Combined, they represent an industry accelerating across all layers of the stack.
At MACHINA Summit 2026, we convene the architects programming the infrastructure robotic layer. We believe progress in embodied intelligence is not defined by a single breakthrough. It emerges from the dialogue between what exists, what’s emerging, what’s challenging, and what’s next.
It is within this dialogue that intelligence evolves, boundaries shift, and the future takes form.
Join us at STATION F this July 7, 2026, to bridge the gap between lab-based research and industrial scale deployment. Secure your ticket.
