Disclosure: My company, Tirias Research, has consulted for AMD, Intel, Nvidia, NXP, Qualcomm, and other companies mentioned in this article.

Unitree G1 Humanoid Robot at CES 2025 – Tirias Research

A Dream from the Past

People have been dreaming about intelligent humanoid robots since the introduction of the beloved family maid Rosey on The Jetsons in the early 1960s. The Jetsons, an animated sitcom, depicted a futuristic world where robots played a significant role in daily life, from domestic chores to assisting with complex tasks. However, bringing such ideas into reality has been a slow and challenging process. Over the past six decades, the world has witnessed the evolution of numerous key technologies required for complex robotic systems, such as semiconductors, sensors, wireless interfaces, hydraulics, electric motors, positioning solutions, gyroscopes, accelerometers, and battery technology.

The Evolution of Robotic Technology

The journey from a fictional maid robot to tangible humanoid robots has been a long one. Initially, the focus was on industrial robots which could perform repetitive tasks in controlled environments such as manufacturing plants. These were not humanoid but significantly advanced in terms of automation and efficiency. Robots slowly made their way into various domains such as warehouses, operating rooms, and even our homes—with the most familiar being the robotic vacuum cleaner.

However, until recently, what defined most of these robots was that they lacked true intelligence. They could perform pre-programmed tasks or react within a narrowly defined set of parameters but did not possess the ability to reason or make autonomous decisions based on a broad set of data.

The Emergence of Agentic AI

The true game-changer has been the advent of what is known as agentic AI. Traditional generative AI enabled systems to think and develop responses based on pre-existing training data. But agentic AI takes a step further by allowing a system to reason and make decisions autonomously based on a broader set of available data, including real-time sensor data about its current environment.

Understanding Generative AI vs. Agentic AI

Generative AI uses massive datasets to train models such as language models (like GPT) that generate text, images, or other data based on the data they were trained on. However, agentic AI not only generates responses but also makes decisions about actions based on an understanding of the current context—which can include real-time data such as sensory inputs.

This makes agentic AI pivotal for real-world, real-time environments where a robot needs to perceive its surroundings, interpret those perceptions, make a reasoned decision, and then act on it. This is what makes “intelligent humanoid robots” a more feasible reality now.

The “Physical AI” Era

At both Computex in Taiwan last year and CES (Consumer Electronics Show) earlier this year, Nvidia CEO Jensen Wong made a compelling case for what he termed as a new transition in artificial intelligence—the “physical AI” era. To make a dramatic point, he replicated an iconic scene from Iron Man where humanoid robots rose onto the stage, emphasizing that the future of AI lies not only in digital realms but also in the physical world.

Humanoid robots on stage with Nvidia CEO Jensen Huang at CES 2025 – Tirias Research

Advanced Processing Solutions

A robust supporting ecosystem has made physical AI viable. Advanced processing solutions are now available from multiple leading vendors like AMD, Intel, Nvidia, and Qualcomm. These companies provide the computational power needed to process complex AI algorithms in real-time.

Advanced sensors are another crucial part of this ecosystem. Companies such as ams OSRAM, Analog Devices, Honeywell, NXP, ON Semiconductor, and STMicroelectronics have developed highly advanced sensors that provide the necessary data inputs for AI processing.

Similarly, wireless communications and positioning solutions provided by companies like MediaTek, NXP, Qualcomm, Silicon Labs, and Texas Instruments allow for real-time data transfer and precise location tracking, making it possible for robots to operate accurately in real-world environments.

Nvidia’s Leading Role in the Physical AI Revolution

Nvidia has been a pioneer in the field of AI for many years, and its role in the physical AI era is no less significant. The company has spent over a decade developing a comprehensive suite of hardware and software solutions specifically designed for advanced robotic applications.

This includes the Jetson edge processing platforms, which provide powerful processing capabilities in a compact form factor suitable for on-board robot processing. On the cloud side, Nvidia’s GPU training solutions offer the computational power needed to train complex AI models used in robotics.

Moreover, Nvidia’s software platforms are equally important. The Isaac Simulator allows for the realistic simulation of robotic environments, enabling the training and testing of robots in virtual settings before they are deployed in the real world. The Omniverse platform provides a collaborative environment for developing and testing AI and robotics applications.

Other tools like Cosmos help in building scalable and reliable robotics applications, while GR00T and Fleet Management solutions make it possible to manage single robots or even entire fleets efficiently. Therefore, it wouldn’t be surprising if significant new announcements related to robotics and physical AI were made at Nvidia’s upcoming GPU Technology Conference (GTC) in San Jose.

Impressions from Trade Shows

Over the past year, I have seen dozens of humanoid robots at major technology trade shows. Watching a robot from Boston Dynamics navigate through an obstacle course was impressive, but a more profound experience was when a robot approached me, shook my hand, and initiated a conversation.

Such interactions underscore an important point: we are on the cusp of a new era where humanoid robots could become an integral part of our daily lives. The COVID-19 pandemic and ongoing labor shortages have only accelerated the push toward AI, autonomous systems, and robots across various industry segments.

However, with the introduction of agentic AI, it seems that the long-awaited “robotic” era has truly begun—a term which Jensen Huang refers to as “physical AI.”

What 2025 and Beyond Hold

It might be premature to call 2025 the “year of the robot” because an incredible amount of innovation is still on the horizon. Nevertheless, the period between now and 2030 promises to make intelligent robotics a pervasive part of nearly every aspect of society.

Anticipating Future Applications

The potential applications for intelligent humanoid robots and physical AI are vast. Here are a few:

1. Healthcare: Robots could assist in surgery, patient care, and rehabilitation.
2. Home Assistance: From Rosey-like maid robots to personal assistants that could help with daily chores and elderly care.
3. Education: Humanoid robots could serve as tutors or assist in classrooms where there is a shortage of teachers.
4. Industrial Automation: Beyond just manufacturing, robots could handle complex logistics and supply chain tasks, warehouse management, and even construction.
5. Public Services: Robots might be used in public safety, such as in disaster response scenarios where human presence could be risky.
6. Entertainment: Robots could be used in the entertainment industry for roles requiring high precision or in places where human actors might be in danger.

Conclusion

The journey from the Jetsons to real-life humanoid robots has been a long road filled with technological advancements and innovations. The emergence of agentic AI and what Jensen Huang calls “physical AI” marks a significant milestone in this journey. An entire ecosystem has come together, led by major technology vendors such as Nvidia, AMD, Intel, Qualcomm, and many others that provide the necessary hardware and software solutions for making intelligent humanoid robots a reality.

With a robust supporting ecosystem and increasing real-world applications, it is only a matter of time before humanoid robots become a common sight in our everyday lives. From healthcare to home assistance, education to industrial automation, the “physical AI” era is set to transform society in ways previously only imagined in science fiction.

It might not be fully realized in 2025, but by 2030, intelligent humanoid بي robots will likely reach nearly every aspect of our daily lives, ushering in a new era of interaction between humans and machines.