In his influential 2011 novel, Zero to One: Notes on Startups, or How to Build the Future, entrepreneur and controversial figure, Peter Thiel famously said, "We wanted flying cars; instead, we got 140 characters." Thiel was disappointed with progress in the early 21st century and argued that innovation had largely stagnated in the physical realm, limiting breakthroughs to incremental advances in the digital world.

Almost 15 years after the book, the stagnation is over, we're stepping into a future where the physical world will witness dramatic changes. Although mass market flying cars remain an elusive goal, we are going to get self-driving cars this year, and humanoid robots will quickly enter our reality. 2025 is a special year because it’s when many robots move from labs to real-world deployment with pilots in industrial and domestic settings.

Why are humanoid robots becoming a reality now? What has changed?

Three key factors are converging to make humanoid robots inevitable: explosive progress in artificial intelligence, breakthroughs in robotics hardware, and urgent global challenges.

‍

The Explosive Progress In AI

The bedrock of this rapid evolution lies in recent advancements in artificial intelligence and machine learning. Made visible by the groundbreaking release of ChatGPT in November 2022, the momentum around AI has only intensified. AI applications have proliferated across diverse fields, including image and video generation with tools like DALL·E and Midjourney, protein folding with AlphaFold, and autonomous driving with systems like Tesla’s Full Self-Driving. Multiple fields within AI are growing simultaneously, and they are all relevant for humanoid robots.

Large Language Models (LLMs) are enabling machines to understand human languages, let them generate human-like text, answer complex questions, and assist in tasks like coding and creative writing. Speech recognition and text-to-speech (TTS) are relatively old fields. Being personally involved in the development of those technologies 15 years ago, its usability was limited to very specific applications. 

Today, speech recognition is so advanced that it allows machines to transcribe speech into text with remarkable accuracy, even in noisy environments. Meanwhile, text-to-speech models, like those from ElevenLabs or Google’s WaveNet, let machines produce voices that are nearly indistinguishable from humans. These technologies will allow people to interact naturally with self-driving cars, humanoid robots, or smart home systems.

As Brett Adcock, CEO of Figure, put it,

'We can literally talk to our robot, and it can execute the tasks you ask of it. The ultimate goal is for speech to become the default interface.'

This vision underscores the potential for voice-driven AI to serve as an intuitive bridge between humans and machines, simplifying complex actions and democratizing access to advanced technology.

‍

Vision-Language-Action models (VLAs) are another revolution, those models integrate diverse inputs, such as plain-language commands, visual data from cameras, or environmental states; and translate them into actionable sub-tasks and precise physical movements. 

As outlined in recent research arXiv:2405.14093v4, VLAs are entering a transformative era, trailing only slightly behind the advancements of Large Language. VLA models eliminate the traditional siloing of robotic functions, where perception, reasoning, and movement were handled by separate systems. Instead, a single multimodal model now processes raw pixel inputs from cameras, interprets natural language instructions, and directly outputs torque targets to control every joint of a robot. This end-to-end approach enables fluid, context-aware interactions, allowing robots to dynamically adapt to complex environments in real time.

‍

World simulation is another critical enabler. Increasingly, robots are trained in virtual environments using foundational world models, which simulate physics, interactions, and scenarios with high fidelity.

These models, powered by platforms like NVIDIA’s Isaac Sim, generate billions of synthetic training scenarios, allowing robots to learn complex behaviors without physical wear or risk. This shift to simulation-driven training accelerates development, reduces costs, and enables robots to generalize across diverse tasks, from warehouse logistics to caregiving.

‍

Hardware Breakthroughs in Robotics and AI

Hardware is progressing extremely fast, and once again there are two completely different parts of hardware that are moving at breakneck speed: computing hardware and robotics hardware.

The progress in computing hardware might be less surprising but the xx are still mind-blowing.

Nvidia dominates compute. Their Jetson Orin NX already runs at 275 TOPS (tera-operations per second) inside a chip that fits in your palm. The next iteration of Jetson, Thor, triples this.

Computing Hardware Built for Real-Time Robotics

In computing hardware, the rapid escalation of performance is nothing short of remarkable. NVIDIA continues to dominate, setting new benchmarks with its powerful yet compact chips.

The NVIDIA Jetson Orin NX, for example, achieves a staggering 275 tera-operations per second (TOPS) in a palm-sized form factor, enabling real-time processing for complex AI tasks such as vision, language, and motion planning. Its successor, the Jetson AGX Thor, introduced in 2025 and built on NVIDIA’s cutting-edge Blackwell GPU architecture, triples this computational capacity. Featuring a specialized transformer engine, the AGX Thor is explicitly designed to meet the intensive demands of humanoid robotics, providing the raw AI horsepower needed for multimodal Vision-Language-Action (VLA) models.

NVIDIA Jetson AGX Thor, based on NVIDIA's Blackwell GPU architecture, delivers ultra-high-performance AI compute and a new transformer engine. This delivers the necessary AI superpower at the edge to enable the new generation of humanoids

‍

Robotic Systems That Match Human Precision

The development of humanoid robots depends on the development of extremely advanced mechanical systems, and this is often overlooked amidst the hype surrounding artificial intelligence.

As Elon Musk noted in a recent X post,

“The progress in robotic mechanics is often overshadowed by AI, but it’s the precision of actuators and sensors that makes a humanoid truly functional.”

The good news is that robotics hardware is undergoing a profound transformation as well, driven by innovations in sensors, actuators, and control systems.

The mechanical complexity of humanoid robots, especially in mimicking human anatomy, is challenging. The human hand is a marvel of natural engineering, with its 27 bones, 34 muscles, and an intricate network of nerves. It enables tasks as diverse as grasping delicate objects, manipulating tools, and expressive gesturing. Replicating this in robotics demands breakthroughs in multi-degree-of-freedom actuators, lightweight yet durable materials, and sophisticated haptic feedback systems that simulate the sense of touch.

For instance, Sanctuary AI, a Canadian humanoid robotics startup, has prioritized developing a hand capable of sensing and handling fragile objects with human-like finesse, underscoring the hand’s pivotal role in robotic functionality.

Similarly, Proception AI, a Y Combinator-backed startup, has chosen to focus exclusively on creating a humanoid hand, bypassing the development of a full robot to tackle this singular challenge. This illustrates the immense technical hurdles and specialized focus required for even a single body part of the humanoid robot.

‍

The robotics hardware sector is seeing so much interest and investment, that there’s been as much progress in the past 12 months in robotics hardware than in the previous 12 years combined.

Robotics Hardware Is Becoming More Affordable

The affordability of robotics hardware is improving dramatically, lowering barriers to adoption. Companies like Unitree are leading, offering models like the G1 humanoid for approximately $20,000—a price point comparable to a mid-range car.

While not yet production-ready, these systems demonstrate a clear path toward mass-market humanoids, making them accessible to researchers, businesses, and eventually consumers. This cost reduction, driven by economies of scale and manufacturing innovations, is accelerating the democratization of robotics.

Why the Race for Humanoid Robots Is Accelerating

Global Demand is Driving Humanoid Adoption

Humanoid robots are arriving at a critical juncture. Aging populations in countries like Japan, China, and Europe, coupled with labor shortages in manufacturing, healthcare, and logistics, are creating an urgent demand for robotic solutions. In healthcare, robots could assist with eldercare, performing tasks from mobility support to medication management.

In manufacturing, they could address worker shortages, boosting productivity. This economic imperative is driving governments and industries to prioritize humanoid development, with China emerging as a leader.

Progress Fuels Progress

Successful pilot programs and compelling demonstrations are attracting significant attention from industry giants and investors. Capital is no longer a bottleneck.

In 2025, Figure AI secured a $2 billion funding round to scale its Figure 02 robot for industrial applications, while Apptronik raised $403 million from investors, including Mercedes-Benz and Google, to deploy its Apollo robot in automotive and logistics sectors. These investments are fueling a virtuous cycle: breakthroughs drive visibility, which attracts funding, which accelerates innovation.

Competition is Speeding Up Innovation

Global competition is intensifying, with startups, tech giants, and governments vying for dominance in the humanoid robotics race. China’s aggressive push, supported by state-backed initiatives, contrasts with Western efforts led by companies like Tesla, Figure AI, and Boston Dynamics. This rivalry is spurring rapid advancements, as each player seeks to outpace the others in technology and deployment.

While those three factors are progressing in parallel, unprecedented investment is accelerating this transition, with startups securing billions in funding to expedite development and commercialization.

Figure AI raised $2 billion in 2025 to scale its Figure 02 robot for industrial applications, while Apptronik secured $403 million from investors such as Mercedes-Benz and Google, aiming to deploy their Apollo robot in sectors like automotive and logistics. Let’s dive into those factors.

This Is the Start of the Humanoid Revolution

Humanoid robots were once relegated to the realm of science fiction. We never thought we would see them in our lifetime. Now that they are within reach, their arrival is met with skepticism by many who underestimate the significance of early demonstrations and pilot programs. The reality is that we are on the precipice of a revolution, and the pace of change is staggering.

We are still early. If you’re captivated by this transformation, you’re part of an elite group poised to witness, and contribute to, a world redefined by intelligent machines.

‍

Bullish on robotics? So are we.

XMAQUINA is a decentralized ecosystem giving members direct access to the rise of humanoid robotics and Physical AI—technologies set to reshape the global economy.

Join thousands of futurists building XMAQUINA DAO and follow us on X for the latest updates.

‍