Humanoid Horizon: Why Robots Are Our Next Inescapable Neighbours And Maybe Heirs Too

The tech world, buzzing with excitement over advances in artificial intelligence, saw three major headlines last week:

—On March 15, American robotics company Figure AI unveiled BotQ — a first-generation, high-volume manufacturing facility capable of producing up to 12,000 humanoid robots per year.

—Then, on March 16, Chinese start-up Unitree made headlines with its acrobatic humanoid — affectionately dubbed the "kung fu bot" — impressing the industry by executing the first-ever successful robotic side flip.

These rapid-fire announcements from both the US and China signal one thing loud and clear: the technology behind humanoid robots — AI-powered machines with humanlike forms — is maturing more rapidly than many expected. The dream of mass production may no longer be a distant fantasy but an imminent reality.

Robots, of course, are nothing new. The very word "robot" first appeared over a century ago, in Karel Čapek's 1920 play Rossum's Universal Robots, describing human-like machines designed for mundane labour.

Since then, technological innovation has pushed robotics far beyond basic industrial functions. Today, the idea of humanoid robots in our workplaces and homes is no longer far-fetched.

Though still in development, humanoid technology is accelerating — fuelled by the powerful convergence of robotics and artificial intelligence (AI) — bringing industry leaders closer to mass production, real-world applications, and widespread market entry.

But First, What Is A Humanoid?

Before diving into why the world is racing to create humanoid robots, let’s break things down: What exactly makes a robot "humanoid"? Does it just need to look human, or does it need to act human too?

Of course, “resembling the human form” can mean different things to different people. Some robots simply feature a head, torso, arms, and legs, while others go further — imitating human expressions and movements.

What truly sets humanoids apart, though, are three defining traits: locomotion, dexterity, and intelligence.

Locomotion (or gross motor skills) lets these robots navigate human-oriented environments, such as navigating steps and uneven floors and avoiding obstacles.

Dexterity, or fine motor skills, enables them to interact in useful ways with their environment to perform specific tasks such as picking up fragile objects without damaging them.

Intelligence allows humanoid robots to independently perceive and engage with the world around them and is increasingly powered by AI.

To fully appreciate these advanced machines, it helps to peek inside — at both the “brain” and the “body” that work in harmony to bring humanoid robots to life.

The "brain" of a humanoid robot isn’t made of neurons — it’s built from a powerful combination of semiconductors and software. At its core are advanced Generative AI (GenAI) models, driving the robot’s autonomy and decision-making. Supporting these are simulation models which serve as virtual training grounds — helping the robot learn and refine its skills before ever interacting with the physical world.

The "body" is a complex network of high-tech components. It houses an array of sensors — from cameras and lidar to force, torque, and magnetic sensors — providing detailed awareness of its surroundings.

Actuators — made up of motors, encoders, bearings, screws, and reducers — translate commands into smooth, precise movements. This entire system is linked by an intricate network of wires and connectors, powered by a centrally mounted lithium-ion battery.

On the outside, most humanoid robots are encased in a blend of lightweight aluminum alloys and plastics — keeping them strong, durable, and agile without adding unnecessary weight.

Why Humanoids?

Many readers might wonder: "Why do we need robots shaped like humans?" Why engineer something as complex as hands, arms, or legs?

The answer: It’s not just about aesthetics.

To be sure, there are strong arguments for robots to take on specialised forms — like robot arms, snake-shaped robots, robot dogs, or even robotic dust — adapting to specific tasks in countless configurations.

However, many robotics and AI experts argue that the most compelling case for humanoid robots is rooted in the world we've already built. Our environment is "brownfielded" for human bodies — meaning it’s designed by humans, for humans — and humanoid robots can seamlessly integrate into these spaces with minimal modifications.

As Nvidia points out, humanoid robots excel precisely because they match the physical blueprint the world was designed for.

Nvidia CEO Jensen Huang explains it this way: “The easiest robot to adapt into the world are humanoid robots because we built the world for us. We also have the most amount of data to train these robots than other types of robots because we have the same physique."

Consider this: human hands are remarkably versatile — capable of handling delicate objects, using tools, and operating heavy machinery. Most workstations, tools, and everyday objects are designed around human proportions. A robot that mirrors our form can step into those roles immediately — without requiring a complete redesign of the world around it.

And let’s face it — a machine that looks familiar (and maybe even friendly) has a better chance of social acceptance than something that seems alien or out of place.

Huang sums it up perfectly: "Robots look like people because … the most important reason, is that we built the world for ourselves, and so the workstations of a factory, the manufacturing line of a factory, was really created for people."

Breaking Down the Humanoid Robot Value Chain

Companies in the "Brain" category supply advanced semiconductors and AI software models — the driving force behind humanoid autonomy. These models handle both training (teaching the robot how to think and act) and inference (helping it make decisions in real-time).

Next comes the "Body". Companies here specialize in hardware — from sensors and actuators for movement to batteries, wiring, and other essential components that bring the robot to life.

Finally, we have the "Integrators" — companies currently building full humanoid robots or that have the reasonable potential to do so given their expertise in manufacturing other robotics or automation platforms.

To analyze the value chain, Morgan Stanley’s team selected an extensive set of publicly traded companies with significant involvement in the humanoid market.

They identified 22 companies in the "Brain" category, 64 companies supplying critical components for the "Body" and 22 companies classified as "Integrators" — those most likely to deliver full-scale humanoid robots to market.

One striking takeaway from Morgan Stanley’s research? Asian companies — particularly those in China — are leading the humanoid robotics race.

Out of 100 publicly traded companies worldwide that Morgan Stanley identified as actively developing humanoids, 56 per cent are based in China. Even more notably, 45 per cent of the world’s robot integrators — companies that customize robots to meet specific end-user needs — also call China home.

“A common refrain we hear from investors is the lack of Western firms to add to their humanoid portfolio outside of Tesla and Nvidia,” Morgan Stanley analysts led by Adam Jonas noted in the research paper.

China “continues to show the most impressive progress” in humanoid robotics due to strong government support, domestic start-ups’ access to established domestic supply chains, and the plethora of businesses willing to put the machines to use, the paper, titled 'Humanoid 100', found.

“In our view, this is important information in and of itself as it represents the reality of the current humanoid ecosystem, which we expect may need to change materially over time,” the bank said in the paper.

The Use Case

Humanoid robots first gained traction in the automotive industry.

In a post on X, Tesla highlighted its key achievements since 2018 — notably announcing the successful deployment of two Optimus robots in one of its factories.

Similarly, Hong Kong-listed UBTech Robotics revealed in early March that it had completed a trial involving dozens of Walker S1 robots at the Zeekr EV factory in the Chinese port city of Ningbo, supporting “multitask” and “multi-site” operations.

Meanwhile, Figure AI, backed by Nvidia, plans to start training its robots in BMW factories later this year.

So far, most humanoid robots in real-world use are concentrated in industrial environments like logistics and manufacturing, while others support research and education.

Their presence in the service sector remains limited, primarily appearing in commercial roles — such as tour guides at museums and exhibitions.

Notably, last month, 16 humanoid robots from Hangzhou-based Unitree dazzled audiences with a synchronized dance performance during China’s Spring Festival Gala — the world’s most-watched annual TV broadcast — on Lunar New Year’s Eve.

Beyond entertainment, countries are accelerating efforts to develop humanoid robots for military purposes, though large-scale deployment remains a distant goal.

These robots are being tested in high-risk scenarios — from battlefield reconnaissance to disaster response — leveraging their human-like mobility and decision-making abilities. The US leads this charge, with Boston Dynamics' agile search-and-rescue robot, Atlas, and SAFFiR, a firefighting robot designed for the US Navy.

Humanoid robots are also venturing into space. The US, Japan, and Russia have deployed them on missions to gather data and operate in extreme environments.

Yet, for everyday people, humanoid robots remain more of a spectacle than a helping hand — their true potential still waiting to be unlocked.

The Urgency Is ‘Palpable’

Beyond the technological marvels and intricate engineering, one crucial question looms: Is the world truly ready for humanoid robots?

Goldman Sachs certainly thinks so. The firm projects the total addressable market (TAM) for humanoid robots to hit a staggering $38 billion by 2035 — with much of this growth driven by their deployment in structured environments like manufacturing. Picture EV assembly lines, precision component sorting, and other repetitive industrial tasks.

But the real promise of humanoids stretches far beyond factory floors. Thanks to AI-driven adaptability, these robots are poised to tackle hazardous, high-risk jobs — tasks with high fatality rates and low human willingness. Goldman Sachs highlights that customers may be more willing to pay a premium for robots capable of handling dangerous operations where human lives are at stake.

While factories may be the first to embrace humanoid robots, analysts like RBC’s Tom Narayan believe the biggest breakthrough will come when — or if — these machines become an integral part of everyday life. The real opportunity, he says, is in “more than just cutting vegetables,” he said.

He envisions humanoid robots evolving into household companions — personal assistants capable of helping raise children or even babysitting. "In most manufacturing plants in the U.S., robots are already working. The real value? It’s in people’s homes,” Narayan said in an interview.

Morgan Stanley’s Adam Jonas echoes this, predicting humanoid robots will become a multi-decade, trillion-dollar industry — with adoption potentially outpacing autonomous cars on public roads.

Far more capital will be pumped into the humanoids given the thousands of repetitive and dangerous human tasks that could be made significantly easier with robots being incorporated into factory lines, warehouses and kitchens, Jonas wrote in a note to clients.

This sense of inevitability is shared by corporate giants already doubling down on the humanoid future.

Musk’s focus on Tesla’s humanoid robot project, Optimus, has only intensified. First introduced on September 30, 2022, Optimus remains a top priority at Tesla’s engineering headquarters in Palo Alto.

As Tesla grappled with plummeting profits last year — driven by a brutal EV price war in China and slowing demand in Europe — Musk increasingly spotlighted Optimus as a cornerstone of the company's future.

During Tesla’s fourth-quarter earnings call in January, Musk voiced confidence that the company could produce thousands of humanoid robots by year-end.

He explained that the bots would first take on "the most boring and annoying tasks" in Tesla’s own factories, with deliveries to external customers expected to begin in the second half of next year.

A ChatGPT Moment

Humanoid robots are evolving — not just in form, but in intelligence. Powered by machine learning algorithms, they can analyse data, learn from past actions, and process information from sensors to make decisions in real time.

However, traditional robotics programming remains rigid. These robots still require extensive human expertise to code, and they struggle with unfamiliar situations.

That’s beginning to change. The advent of the large language models (LLMs), coupled with growth of generative AI — software that can create plausible and sophisticated text, images and computer code at a level that mimics human ability — is reshaping robotics.

These technologies are accelerating how physical machines learn — using natural language, imitation, and simulation. Robotics, it seems, is having its own "ChatGPT moment, says Adam Jonas.

For years, the fields of AI and robotics were seen as distinct — LLMs driving language and knowledge, while robotics focused on movement and mechanics. Now, those worlds are colliding, and the impacts are profound.

Robots are increasingly capable of learning new tasks — not through painstaking manual programming, but by tapping into powerful foundation models trained on massive datasets of both real-world and simulated experiences.

These models — built by some of the world’s leading tech giants like NVIDIA, Alphabet, Baidu, Meta, and Microsoft — will supercharge humanoid robot development.

On 18 March, NVIDIA unveiled Isaac GR00T N1— the world’s first open, fully customizable foundation model for generalized humanoid reasoning and skills— to speed robot development.

“The age of generalist robotics is here,” said Jensen Huang, CEO of NVIDIA. “With NVIDIA Isaac GR00T N1 and new data-generation and robot-learning frameworks, robotics developers everywhere will open the next frontier in the age of AI.”

Vincent Vanhoucke, Senior Director for Robotics at Google DeepMind, captured the magnitude of this shift, recalling how experts now look back on 2022 as "the good old days" as he explains how LLMs and genAI have very abruptly flung the field of robotics from an isolated "robot island" firmly onto the "AI flywheel".