13 May 2026

 

French robotics startup Genesis AI has unveiled GENE-26.5, a robotics foundation model designed to give machines what the company describes as “human-level physical manipulation capabilities”. Announced in May 2026 alongside a highly dexterous robotic hand, the system is intended to help robots perform complex physical tasks requiring coordination, adaptability and fine motor control. In demonstrations released by the company, robots powered by GENE-26.5 cooked meals, cracked eggs one-handed, solved a Rubik’s Cube and played piano pieces at human-like speed.

The launch marks Genesis AI’s first major product announcement since emerging from stealth in 2025 with a $105 million seed funding round backed by investors including former Google chief executive Eric Schmidt, French entrepreneur Xavier Niel and France’s public investment bank Bpifrance. The company was founded by Zhou Xian, who holds a robotics PhD from Carnegie Mellon University, and Théophile Gervet, formerly a research scientist at Mistral AI.

While Genesis AI initially attracted attention because of the scale of its early funding, GENE-26.5 offers a clearer picture of the company’s broader ambitions: building a general-purpose robotics AI model capable of operating across different robotic systems and physical environments.

That places Genesis AI inside one of the fastest-moving areas of the Artificial Intelligence sector, where companies are attempting to extend foundation model concepts beyond software and into physical machines.

 

Robotics moves beyond narrow automation

 

For decades, industrial robots have largely depended on highly structured environments and repetitive workflows. Automotive factories, warehouses and electronics assembly lines have all benefited from robotic automation, but most systems remain limited to narrowly defined tasks.

A robot programmed to weld a specific car component or move a particular package may struggle when objects shift position slightly, materials change or environments become less predictable.

That limitation has become increasingly important as the robotics industry attempts to move beyond fixed industrial automation into more adaptable systems capable of handling real-world variability.

Foundation models are central to that effort.

In software, foundation models are large AI systems trained on broad datasets that can generalise across different tasks. OpenAI’s GPT models and Google’s Gemini are among the best-known examples. Robotics companies are now attempting to apply similar principles to physical machines.

The challenge is significantly harder.

Unlike text-based AI systems, robots must interpret space, force, movement, object interaction and physical consequences in real time. Small variations in texture, lighting, weight or positioning can cause robotic systems to fail.

GENE-26.5 is Genesis AI’s attempt to address that problem through large-scale multimodal training and what the company describes as a “robotic brain” capable of operating across multiple tasks and hardware systems.

According to Genesis AI, the system is designed to run not only on the company’s own robots, but also on third-party robotic platforms.

 

Why robotic hands have become strategically important

 

The centrepiece of the announcement was not only the AI model itself, but the robotic hand accompanying it.

While humanoid robots often attract headlines because of their human-like appearance, many robotics researchers consider dexterous manipulation to be one of the field’s most difficult technical challenges.

Human hands combine more than 20 degrees of freedom with highly sensitive touch feedback, fine motor coordination and continuous adaptation. Replicating that mechanically — while simultaneously interpreting environmental feedback through AI systems — remains extremely difficult.

Genesis AI says its robotic hand was designed to mirror human anatomy more closely than traditional industrial grippers. That design serves a practical purpose beyond appearance.

One of the largest bottlenecks in robotics is collecting useful training data. Text-based AI systems can train on internet-scale datasets, but robots require physical interaction data linked to movement, force and spatial reasoning.

Genesis AI says it addresses this by collecting data from humans wearing sensor-equipped gloves that track finger, wrist and hand movements. Because the robotic hand closely mirrors human hand structure, the company says that data transfers more effectively into robotic systems.

The startup is also using internet video, simulation environments and head-mounted cameras to generate training data.

Genesis AI is not alone in pursuing dexterous robotic manipulation. A growing number of robotics companies now view robotic hands — rather than locomotion alone — as one of the key bottlenecks preventing broader deployment of humanoid and general-purpose robots.

Several companies are already competing in this space.

UK-based Shadow Robot Company has spent years developing highly dexterous robotic hands used by research institutions including NASA and OpenAI. Canadian startup Sanctuary AI has focused heavily on tactile sensing and fine manipulation for industrial humanoid robots, while US startups Figure AI and Physical Intelligence are building broader robotics foundation models intended to generalise across tasks and environments.

Tesla’s Optimus programme has also invested heavily in robotic hand development, with recent versions reportedly featuring 22 degrees of freedom designed for delicate object handling.

Meanwhile, Chinese robotics companies are scaling rapidly. Startup Linkerbot, according to Reuters, has emerged as one of China’s leading specialised robotic hand companies, reflecting how dexterity has become strategically important within the wider AI and industrial automation race.

Genesis AI’s positioning differs slightly from many of those competitors because the company is attempting to build the full stack simultaneously: the robotic hand, motion-capture systems, simulation environment and foundation model itself.

That vertically integrated approach increasingly mirrors strategies pursued by companies such as Tesla and Figure AI, where hardware, sensing and AI training are treated as tightly interconnected systems rather than interchangeable components.

 

The “sim-to-real” robotics problem remains unresolved

 

Simulation-based robotics training has become more sophisticated in recent years, but major technical limitations remain.

Robots that perform well inside simulation environments often struggle in real-world deployment — a long-standing problem known within robotics as the “sim-to-real gap”.

Material textures, object variation, lighting conditions, friction and unpredictable movement all introduce complications that simulations cannot perfectly replicate.

Genesis AI argues that combining large-scale simulation with real-world human movement data helps reduce that gap.

The company’s demonstrations suggest substantial progress in dexterity compared with earlier generations of robotic systems. In videos released by Genesis AI, robots performed multi-step tasks involving smoothie preparation, laboratory work and multi-object manipulation.

The demonstrations were autonomous rather than remotely controlled, according to the company.

Yet the distinction between controlled demonstrations and scalable commercial deployment remains important.

Many robotics companies have historically produced impressive demonstrations without achieving broad industrial adoption. Physical systems operate under significantly greater reliability constraints than software systems because they interact continuously with unpredictable real-world environments.

Genesis AI itself acknowledged that some delicate manipulation tasks currently succeed at lower rates than others.

That does not diminish the technical progress being made across the sector. Rather, it reflects the broader reality that robotics development cycles typically move more slowly than software because hardware, sensing and physical interaction all introduce additional complexity.

 

Europe enters the robotics foundation model race

 

Genesis AI’s emergence also highlights Europe’s growing interest in advanced robotics and AI infrastructure.

Although the company operates research activities in Silicon Valley, Genesis AI has emphasised Europe as a strategic priority because of both its industrial base and engineering talent.

That positioning aligns with broader European concerns around industrial competitiveness, AI sovereignty and advanced manufacturing.

China has made AI-powered robotics a strategic priority within its industrial policy, while the United States continues attracting substantial investment into robotics startups including Figure AI and Physical Intelligence.

Europe has often struggled to scale technology companies at similar speed.

Genesis AI’s funding round therefore attracted attention not only because of its size, but because it suggested European investors and institutions are increasingly willing to fund robotics infrastructure companies at Silicon Valley scale.

The company is already in discussions with industrial customers in France, Germany and Italy, targeting sectors including automotive manufacturing, logistics, electronics and pharmaceuticals.

Those industries increasingly face labour shortages, supply-chain pressure and growing demand for automation capable of handling more variable tasks.

 

Physical AI becomes the next competitive frontier

 

The broader significance of GENE-26.5 extends beyond robotic hands alone.

Over the past three years, much of the global AI race has centred on language models, cloud infrastructure and generative AI software tools. Robotics represents a shift towards what many researchers now describe as “physical AI” or “embodied AI”: systems capable not only of generating information, but also interacting with the physical world.

That transition introduces entirely different technical and commercial challenges.

Physical systems must deal with safety, energy consumption, reliability, hardware cost and real-world unpredictability in ways software-only AI systems largely avoid.

At the same time, the economic incentives are substantial.

Industrial robotics, warehouse automation, healthcare support systems and domestic robotics all represent potentially significant markets if AI systems become sufficiently adaptable and reliable.

Genesis AI is positioning GENE-26.5 as part of that longer-term transition.

Whether robotics foundation models ultimately achieve the same scale of impact as language models remains unclear. What is already becoming evident, however, is that the AI industry is increasingly moving beyond screens and cloud software into physical systems capable of interacting directly with the real world.

GENE-26.5 reflects how the next phase of the AI race may increasingly depend not only on reasoning and language generation, but also on movement, dexterity and physical capability.

 

Further reading on MoveTheNeedle.news:

China places AI-powered robotics at the centre of industrial strategy