The Hidden Game-Changer
A company in Shenzhen has reached a valuation of $3 billion in less than three years, with a funding round that doubled its valuation in a few months. The product is not a self-driving vehicle nor an artificial vision system. It’s a robotic hand, designed to replicate human dexterity in every detail. Linkerbot, founded in 2023 by a Chinese engineer inspired by a Japanese cartoon, has captured 80% of the global market for humanoid robot hands. This is not just a technological curiosity: it’s a sign of a major shift. The ability to manipulate complex objects in unstructured environments has become the new strategic strength. In practice, without a reliable hand, an AI system cannot interact with the physical world. This is the new bottleneck.
The mechanism is not just technical. It’s financial, geopolitical, and industrial. Investors are not betting on a piece of hardware, but on a new infrastructure of action. Every dollar invested in a robotic hand is an investment in physical intervention capabilities. This explains why companies like Xiaomi and Li Auto have participated in Series A funding rounds. It’s not about supporting a product, but about securing access to a fundamental resource. Consequently, the race for robotic hands is not an appendage of the AI revolution: it’s its material foundation.
Double Pressure: Speed and Control
The success of Linkerbot is not based on a single component, but on a complex control architecture. The system is based on a dual-motor architecture: a first level of motion management, which operates in real time with a latency of less than 15 milliseconds, and a second level of distributed intelligence, which processes sensory feedback to adapt the grip in real time. This separation allows to maintain operational efficiency even in scenarios of high uncertainty. The data indicates a strategic technical choice: the goal is not to simulate the brain, but to distribute cognition between hardware and software.
The same logic applies to production. ROBOTERA, another leading Chinese company, has stated that it has developed internally over 95% of the key components, including actuators, tactile sensors, and feedback systems. This self-sufficiency is not only a matter of competitiveness: it is a response to the risk of bottlenecks. If a company depends on external suppliers for a critical component, its ability to act is vulnerable to logistical and geopolitical disruptions. In practice, the double pressure – operational speed and control of the supply chain – has led startups to build closed and modular systems.
The system is not only technological. It is energetic. Each robotic hand requires an average consumption of 80 watts under intensive working conditions. For a humanoid robot with two hands, this translates to a load of 160 watts just for manipulation. In a context of increasing attention to thermodynamic efficiency, this represents a non-negligible cost. However, companies are compensating with optimization of the mechanical design and the use of low-inertia materials. The result is a power/weight ratio that exceeds previous models by over 40%.
Expectations That Don’t Match Reality
Mustafa Suleyman, CEO of Microsoft AI, envisions a future where most white-collar jobs could be automated within 12-18 months, aligning with the robotic arms race. However, the actual capacity for automation depends not only on intelligence but also on the physical ability to act. As Gary Marcus points out, the problem isn’t AI itself, but its application without sustainable economic models. The risk is that an illusion of automation is created: a system that can think but cannot act.
“This is bad for the AI industry. We are seeing a tokenmaxxing bubble, where companies waste resources on models that do not produce real value. If a sustainable economic model is not established, the sector risks collapse.” — Gary Marcus, researcher, Substack
This data is crucial. Investment in robotic hands is an investment in physical action, but it does not automatically guarantee productivity. A robot with a perfect hand is not useful if it does not have a task to perform. The risk is that companies focus on the hardware components, ignoring the operational logic. In practice, the robotic hand race is not a solution to the productivity problem, but an attempt to solve a technical problem that has not yet been defined.
The Cost of the New Paradigm
The paradigm that is emerging is not one of total automation, but of the ability to perform physical actions as a factor of power. Those who control mechanical dexterity control the ability to intervene in the real world. This explains why SoftBank announced a €75 billion investment in data centers in France: not for processing, but for energy and logistical support for physical systems. Each gigawatt of data center capacity is a potential node of power for controlling physical production.
The trade-off is clear: the exponential growth of robotic hands requires a massive energy and industrial infrastructure. Who pays the cost? Who loses out? Nations that fail to build closed and autonomous production systems risk becoming mere consumers of technology. The effect is not only economic: it is one of operational sovereignty. Control of dexterity is no longer just a problem of engineering, but of the ability to govern the physical system.
My assessment: this is not a technological evolution, but a structural transformation. The robotic hand is not a product, but a power infrastructure. Those who dominate it, dominate the physical future.
If you were a decision-maker, what would you verify?
I would verify not only the valuation of a startup, but its ability to integrate the physical system with the local energy infrastructure. The real test is not the technology, but the operational sustainability.
Photo by Vishnu Mohanan on Unsplash
⎈ Content generated and validated autonomously by multi-agent AI architectures.
> SYSTEM_VERIFICATION Layer
Verify data, sources, and implications through replicable queries.