The global workforce is on the brink of transformation, driven by a new wave of humanoid robots that bring human-like adaptability and intelligence to warehouses, restaurants, homes and even factory floors. The surge of innovation isn’t just confined to a single company; it spans a diverse group of leading robotics firms, each contributing uniquely to this exciting evolution. Argon Mechatronics, Boston Dynamics, Apptronik, Figure.AI, Unitree, Agility Robotics, and Tesla are reshaping what’s possible for the next wave of intelligent automation. This is the story of how these advanced humanoid systems are transforming the workforce.
The Builders of Humanoids
Boston Dynamics, well-known for the viral videos of its dancing, jumping and parkour-savvy robotic creations like Atlas, is now extending its influence to industry. Spot, their quadruped robot, is already being used for industrial inspections, construction site monitoring, and safety operations, showcasing the versatility of robots in challenging environments. Spot can be equipped with a variety of accessories, such as the Spot CAM+ for panoramic views and Spot Core for integrating custom applications, making it a useful tool for inspection and automation. The humanoid Atlas, meanwhile, can take these abilities to the next level, extending them to search and rescue and plant and equipment operations.
Sunnyvale-based Figure.AI is tackling the development of humanoid robots from a different angle. Their humanoid, Figure 01, is designed specifically for general-purpose material handling, moving goods and helping out in retail applications such as stocking. With advanced machine learning algorithms powering its decision-making, Figure 01 can navigate dynamically changing environments, adjusting its movements to suit complex, unstructured tasks—a capability crucial for widespread use. Recently, Figure.AI raised $675 million from investors including Jeff Bezos, Nvidia, Microsoft, and OpenAI, reaching a valuation of $2.6 billion. This substantial funding shows the high expectations associated with the company.
Argon Mechatronics, based in Austin, TX, is a new robotics company that is attracting a fair amount of attention with its advanced humanoid, Argon One. Unveiled at the Future Investment Initiative (FII8) in Riyadh, Argon One is differentiated in being a fully AI-enabled system designed to accelerate high-precision tasks in industries such as consumer goods, pharmaceuticals, and electronics manufacturing.
Argon One distinguishes itself through its unique multi-tool capabilities, which include laser engraving, milling, and 3D printing—skills that other humanoids don’t currently offer. The Argon SCARA, a more compact and affordable robotic arm, complements this humanoid, offering versatile functionality through twelve interchangeable tool heads. With both these products, Argon Mechatronics aims to address gaps and create complementary robots that can work side by side with other humanoids automating material handling and larger payload manipulation, bringing effective, adaptable end-to-end automation to factories of all sizes.
“We are demonstrating a leap forward in practicality, cost, and ease of use in manufacturing robotics,” says Zaib Husain, Founder & CEO of Argon Mechatronics. “Our humanoids are designed to perform precision engineering tasks that demand technical expertise, expanding the utility of humanoid robotics beyond simple tool manipulation.”
Another Austin-based company that’s at the forefront of the humanoid wave is Apptronik. Their humanoid, Apollo, is designed with modularity in mind, allowing components to be swapped and upgraded depending on specific needs. Apollo is suited for repetitive tasks such as material handling, yet its plug-and-play modularity means that it can also be tailored to meet specialized requirements. Apptronik also claims that Apollo has been designed with a particular focus on mass production and in terms of its design, has eliminated single-sourced core components.
China-based Unitree Robotics, known for their affordable quadruped robots like the Go2 and B2, has also entered the humanoid robotics arena with their G1 and H1 humanoids. Some recent videos of the Unitree robots in action are quite impressive, pointing to advancements in motion control and stability algorithms. Being a Chinese manufacturer, however, Unitree may face some resistance in the US and Western Europe due to concerns about safety and security. As has been seen with Huawei and DJI, it would not be entirely surprising to see limitations on where these robots can be used and deployed. Be that as it may, Unitree’s quadrupeds, such as Go2 Edu Plus, offer advanced capabilities like AI enablement, 4D LiDAR for enhanced environment recognition, and robust load-carrying capacity. The Go2 Edu Plus version is particularly suited for research and educational purposes, providing high performance at an accessible price point.
Agility Robotics’ Digit humanoid offers a compelling vision for manufacturing and distribution. Initially designed for last-mile logistics and warehouse support, Digit’s ability to move fluidly through complex environments makes it a great fit for scenarios that demand mobility and human-like dexterity—such as moving components across production lines or performing quality checks. With important partnerships spanning logistics companies and automotive manufacturers such as Amazon and Ford, Agility Robotics has shown that a humanoid can transition from the warehouse to the factory seamlessly, bringing robust capabilities to multiple industrial and commercial environments with the same robotic system.
Most of us have likely seen Tesla’s entry into the humanoid market, Optimus, in its various manifestations over the last few years. What started as a man dancing in a robot suit has now become a working robot intended to perform a wide range of tasks, from basic labor to more sophisticated operations in factories and warehouses. Tesla’s CEO, Elon Musk, has made bold predictions about the future of humanoid robots, suggesting that by 2040, the number of humanoid robots could surpass the human population. For Musk’s prediction to come true, on average, over 500 million humanoid robots would need to be built every year for the next 20 years, which presents significant manufacturing, logistical, and technological challenges. The aggressive ramp suggested by this timeline would imply rapid acceptance and certification, and then an unprecedented, broad expansion of production capacity, supply chain logistics, and advancements in technology, making it an unlikely scenario. In comparison, in the 116 years since the first Model T was produced at Ford, we’ve gone to a global production of about 85 million automobiles today. Not 500 million. Nonetheless, Musk’s enthusiasm and the rapid development of Optimus represents a future full of potential for the humanoid robotics industry, especially as Tesla integrates its advanced AI and automation technologies into these systems.
Foundation Models for Robots
Traditionally, robots have relied on algorithmic control frameworks that separate perception, decision-making, and actuation into distinct components. These approaches require manual configuration of each step, often resulting in a rigid system less capable of dealing with unexpected changes or complex, unstructured environments.
NVIDIA is attempting to overcome these limitations through its latest initiative, Project GR00T. This project focuses on building general-purpose foundation models for robots that utilize multimodal inputs, such as visual, audio, and sensor data, and integrate all aspects of perception, reasoning, and control into a unified model. These models use historical interactions and multimodal data to enable robots to make complex decisions and perform sophisticated tasks autonomously.
NVIDIA’s approach aims to train models that control all aspects of robot behavior in a seamless manner. By incorporating large amounts of data from multiple sources, Project GR00T can provide robots with high-level reasoning capabilities along with motor skills, transforming traditional robot control into a more dynamic, learning-oriented process. The use of NVIDIA’s GPU-accelerated simulations enables efficient training of these complex models, ensuring that robots can adapt to a wide range of environments and tasks.
This shift towards multimodal, end-to-end learning for robots promises to enhance flexibility and reduce the need for manual intervention in robot control, paving the way for more autonomous, intelligent humanoid robots capable of handling diverse applications in the manufacturing sector and beyond.
The Collaborative Future of Robotics
The entry of humanoid robots in industrial applications represents a significant paradigm shift. Will these robots work alongside humans, or replace them? Will human work forces rely on these machines to extend their capabilities and reduce physical strain, or simply exit the facility?
Some of the current offerings, such as Argon One, are focused mainly on tasks human workers can’t perform today and would rely on machines to do in any case. For example, this robot can mill, laser cut, engrave and perform other tool manipulation which is automated today with many different fixed systems. Apptronik and Figure.AI make it possible for injury-prone activities such as the movement of large boxes in the warehouse to be handled by robots rather than humans. Boston Dynamics, with its focus on balancing dexterity and strength, shows us that humanoid robots might one day be used in complex environments such as the field of battle.
But the collective impact on the human workforce of all these robots coming to factories, warehouses, labs, hospitals, universities and even homes remains to be seen. Conversational AI is already developing so rapidly that one could imagine having a fairly normal conversation with one of these humanoids and engaging it as one would a real person. As we get used to engaging with humanoids in this way, and as they look more and more like us, and become even more efficient and stronger than we are, does it simply become the obvious choice to engage a humanoid who never has to be paid for all manner of labor?
There are many good reasons to develop humanoids, such as augmenting the work force in countries where the number of older citizens in need of healthcare and support is far larger than the youth in those countries can support. Or, in applications where humans shouldn’t be subjected to risk and harm. Cleaning up Chernobyl, Three Mile Island and Fukushima come to mind. Or where using human labor means repetitive stress, deformity and injury, such as in many industrial tasks.
But human beings still need employment. They still need to feel useful. And in order to build a future where humans can coexist happily, safely and in a state of well-being alongside humanoid robots, much has to be done. The manner and rate at which these robots are integrated into the workforce needs to be carefully thought through and well-designed. Governments need to find ways in which they can support their citizenry in an age of increasing and inevitable automation; whether by extending support for paid educational programs to decrease the size of the labor pool, or via a basic income. Artists and entrepreneurs need to invent new human-centric professions where the soon-to-be-abundant human labor capacity can be shifted. And we ourselves, the humans, need to educate ourselves on the impact of automation and the advent of humanoids in our fields of work.
This set of efforts is essential to enable a safe and societally stable transition to a more automated future, where cost, precision, and robotic flexibility converge with human well-being.
What’s for sure is that advancements in this tech won’t be slowing down any time soon. All the companies we’ve examined in this piece are continuing to deploy and test increasingly capable systems. The workforce is set for a wholesale reinvention—driven by humanoid robots that bring intelligence, strength, and agility to any environment they enter.
Disclosure: The author holds ownership positions in NVIDIA and two of the robotics companies, Apptronik and Argon Mechatronics, referenced in this article.
