The Humanoid Landscape

While the AI software revolution dominated headlines in 2023 and 2024, something equally transformative has been quietly developing in labs and factories around the world: humanoid robots capable of performing complex physical tasks with unprecedented dexterity and intelligence.

We're witnessing the early days of a profound shift in labor automation that could make today's AI text and image generators look like mere precursors to the real revolution.

This analysis explores the rapidly accelerating humanoid robotics landscape, the key players driving innovation, and what this means for the future of work, the economy, and investors.

Let's dive in.

The $38 Billion Question

The global market for humanoid robots, once a niche corner of robotics, is exploding. Goldman Sachs recently revised their projections, forecasting a market size of $38 billion by 2035 – up dramatically from their previous $6 billion estimate. They expect shipments to reach 1.4 million units, with applications spanning manufacturing, healthcare, logistics, and eventually, consumer homes.

What's driving this sudden acceleration? Three key factors:

1. The AI Breakthrough: Advanced multimodal models like those powering generative AI are now being applied to robotics, enabling robots to understand natural language commands, reason about their environment, and control complex movements.

2. Hardware Economics: The cost to build high-end humanoid robots has plummeted by approximately 40% to around $150,000 per unit, dramatically improving the economics for factory applications.

3. Real-World Capability: Recent demonstrations from companies like Figure AI and 1X show humanoid robots performing genuinely useful tasks – from manipulating unfamiliar objects to collaborating with other robots.

Morgan Stanley has gone even further in their projections, suggesting the humanoid robot population in the U.S. could reach 40,000 by 2030 and swell to a staggering 63 million by 2050 – with a potential $3 trillion impact on wages.

The economics are compelling. While human workers typically provide around 2,000 productive hours annually, humanoid robots can potentially work up to 7,000 hours per year – operating 20 hours daily with minimal downtime. Even at current prices of $30,000 to $150,000 per unit, the math makes sense for many applications.

When we consider the total addressable market for labor replacement and augmentation across sectors, we're potentially looking at a multi-trillion dollar opportunity that could approach 5% of global GDP by 2035-2040.

That’s a pretty big market.

The Race for the Humanoid Future

The competition to build the definitive general-purpose humanoid robot has intensified dramatically, with heavyweight contenders and agile startups all competing for a slice of the pie.

Figure AI: The Fundraising Phenomenon

In just three years, Figure AI has emerged as perhaps the most talked-about company in the space. Founded by Brett Adcock (who previously founded and sold Vettery for $100M and took Archer Aviation public), Figure has demonstrated remarkable progress with its humanoid robot platform.

Their latest breakthrough came in February 2025 with the release of Helix, a Vision-Language-Action (VLA) model that represents several industry firsts:

• Full upper-body control: Controlling the entire humanoid upper body, including individual fingers

• Multi-robot collaboration: Operating simultaneously on two robots to solve shared tasks

• Generalized object manipulation: Picking up virtually any small household item via natural language prompts

• Commercial-ready deployment: Running entirely on embedded GPUs

What makes Figure's approach unique is their "System 1, System 2" architecture that combines a fast reactive visuomotor policy (operating at 200Hz) with a slower but more capable vision-language model for scene understanding and decision-making.

The company has secured impressive partnerships, including deployments at BMW's manufacturing facility where Figure's robots are handling tasks like inserting sheet metal parts into assembly fixtures.

Most notably, Figure has attracted extraordinary investment interest. After raising $675 million in 2024 from investors including Microsoft, OpenAI, NVIDIA, and Jeff Bezos, the company is reportedly in talks for a new funding round at a staggering $39.5 billion valuation – a 15x increase from its previous round.

While such a valuation has raised eyebrows given the early stage of the technology and limited revenue, it underscores the immense potential investors see in the humanoid robotics space.

Tesla's Optimus: The Manufacturing Powerhouse

While Figure AI dominates headlines, Tesla's Optimus project represents perhaps the most credible path to mass production of humanoid robots.

Initially dismissed as a distraction when announced in 2021, Tesla's Optimus has made remarkable progress. The company unveiled its Generation 2 prototype in December 2023, demonstrating end-to-end AI control from scenario observation to task execution.

Tesla's key advantage lies in its manufacturing expertise and vertical integration. The company plans to produce several thousand Optimus robots in 2025 (what they call "Production Design 1"), primarily for internal use in Tesla factories. By 2026, with "Production Design 2," they aim for a dramatic production ramp, potentially reaching towards millions of units annually.

According to recent earnings calls, Tesla CEO Elon Musk believes:

• Optimus requires 10x more training compute than autonomous vehicles

• Humanoid robots have "1,000 times" more potential use cases than cars

• The long-term revenue opportunity could exceed $10 trillion

• Production costs could fall below $20,000 per unit at scale

Tesla's expertise in power electronics, batteries, AI, and mass manufacturing gives them significant advantages in scaling production. Additionally, their experience with global regulators from their automotive business helps in navigating the complex regulatory landscape for robotics.

The company is significantly increasing their investment in training infrastructure for Optimus, signaling their seriousness about the project. While Musk's timelines have historically been optimistic, Tesla's track record in scaling complex technology products can't be dismissed.

1X: The Home-First Pioneer

Taking a different approach is Norwegian robotics company 1X (formerly Halodi Robotics). Their NEO Gamma humanoid robot, unveiled in February 2025, prioritizes home usability over industrial applications.

NEO Gamma's innovations include:

• Consumer-friendly design: Featuring "Emotive Ear Rings" for improved communication and a minimalist aesthetic

• Natural movement: Walking with human-like gait and arm swings, able to squat and sit in chairs

• Safety focus: Soft covers to reduce impact and increase overall safety

• Advanced manipulation: Visual models capable of picking up diverse objects in unfamiliar environments

• Conversational AI: An in-house language model enabling natural conversations

This homes-first approach stands in contrast to most competitors who are targeting industrial applications first. 1X has raised over $23.5 million in funding, with OpenAI as a notable investor – suggesting potential future integration of OpenAI's language models.

Boston Dynamics: The Veteran Player

No discussion of humanoid robots would be complete without Boston Dynamics, the company whose viral videos of dancing robots have captivated audiences for years.

While their Atlas robot demonstrates arguably the most advanced mobility of any humanoid – capable of running, jumping, and performing complex parkour movements – Boston Dynamics has struggled to find commercial applications at scale. Now owned by Hyundai, the company has focused more on their quadruped robot Spot for commercial applications.

Their approach to bipedal robotics has emphasized raw physical capabilities over general-purpose intelligence and manipulation. It remains to be seen whether they'll pivot to compete more directly with newer entrants like Figure AI and Tesla in the general-purpose humanoid space.

Other Notable Contenders

The field is rapidly expanding, with several other important players making significant progress:

Apptronik: This Austin-based company is developing Apollo, a humanoid robot designed for commercial applications. They've secured partnerships with NASA and recently raised substantial funding to accelerate development.

Agility Robotics: Known for their bipedal robot Digit, Agility has taken a focused approach targeting warehouse automation. They've already secured major customers including Amazon and are building a factory capable of producing 10,000 robots annually.

Sanctuary AI: This Canadian company is developing robots with advanced cognitive architecture aimed at human-like reasoning and dexterity.

Unitree: This Chinese manufacturer has made waves with impressively agile and surprisingly affordable robots, though they currently lack the manipulation capabilities of some Western competitors.

Other Big Tech's Strategic Moves

The humanoid robotics race isn't limited to startups and Tesla. Major tech companies are positioning themselves strategically:

Amazon: Logistics Leader

While Amazon hasn't announced plans to build their own humanoid robots, they've made significant investments in the space. Most notably, they're working with Agility Robotics to deploy Digit robots in their warehouses.

Amazon also recently unveiled "Ocelot," a quantum computing chip developed with Caltech to tackle quantum error correction – technology that could eventually support advanced robotics AI.

Google: DeepMind's Robotics Push

Google's DeepMind has been a pioneer in applying AI to robotics. The RT-X project, launched in 2023, brings together 32 robotics laboratories to pool data and resources for developing general-purpose robots.

Their approach focuses on connecting physical robot skills to the semantic and symbolic knowledge in vision-language models, giving robots a degree of common sense that could enable them to understand complex commands.

Microsoft and OpenAI: Key Investors

Rather than building their own robots, Microsoft and OpenAI have invested strategically in the space. Both are investors in Figure AI, positioning them to potentially integrate their AI models with leading hardware platforms.

Microsoft also recently unveiled its Majorana 1 quantum chip, representing another technological avenue that could eventually enhance robotics AI capabilities.

Meta's New Interest

Meta is reportedly exploring entry into the humanoid robotics market, looking to leverage their expertise in AI, sensors, and software. After spending nearly $100 billion on AR/VR through Reality Labs, the company may see robotics as another pathway to create enterprise AI revenue streams.

Apple's Quiet Moves

According to recent reports, Apple is actively exploring both humanoid and non-humanoid robot development, particularly aimed at smart home applications. Given Apple's expertise in consumer hardware, their entry could eventually drive wider consumer adoption. Though we haven’t seen anything yet, a successful product launch from a consumer driven company like Apple has potential to be extraordinarily successful.

The Economics of a Robot Workforce

The fundamental economic case for humanoid robots rests on their ability to replace or augment human labor across multiple sectors. Several factors drive this economic transformation:

Production Economics

Manufacturing costs for sophisticated humanoid robots have fallen dramatically. Current costs range from $30,000 to $150,000 per unit, representing a 40% decrease from previous ranges of $50,000 to $250,000.

At scale of one million units or more, Tesla projects costs could fall below $20,000 per unit, though R&D expenses remain substantial, often exceeding $1 billion for advanced platforms.

The cost decrease stems from several factors: maturing technology, economies of scale, better manufacturing processes, and crossover benefits from advancements in related fields like electric vehicles.

As production volumes increase, the economics become increasingly favorable for wider deployment.

Labor Economics

The economic argument for humanoid robots becomes compelling when comparing their productivity to human workers. Robots can operate up to 20 hours daily, providing approximately 7,000 hours annually, while human workers typically provide around 2,000 productive hours per year.

Current battery limitations of around 5-6 hours runtime require multiple robots for continuous operation, but despite higher upfront costs, the long-term economics increasingly favor automation.

For tasks with clear parameters and physical requirements, the return on investment calculation increasingly favors robotic solutions – particularly in sectors with labor shortages or high wages. The ability to work continuously without breaks, vacation, or sick days creates a compelling economic case even at current price points.

Unprecedented Economic Change

Countries heavily investing in robotics could see annual GDP growth exceeding 10% by the early 2030s, and by the 2040s, robot labor could match or exceed human labor capacity. The "Robotics as a Service" (RaaS) model is emerging as a preferred business strategy, allowing companies to access robotic capabilities without massive upfront investments.

The market could eventually approach a substantial percentage of global GDP as humanoid robots become integrated across manufacturing, logistics, healthcare, retail, and eventually consumer applications.

This economic analogy often made is to the automobile's impact on economies and households – a fundamental reshaping of how work gets done in both labor market and overall economic structures.

Technical Challenges and Breakthroughs

Despite the optimism and progress, significant technical hurdles remain. Understanding these challenges helps contextualize the different approaches companies are taking.

The Data Bottleneck

Unlike large language models that can train on vast text datasets from the internet, robotics faces a fundamental data challenge:

1. Physical Interaction Data: Robots need abundant data from physical interactions to learn manipulation skills

2. Real-World Variety: The diversity of real-world environments and objects is difficult to capture in training

3. Collection Cost: Generating this data is expensive and time-consuming

Though challenges in data processing and storage remain sizeable, the recent capex from large companies like Google, Meta, and Microsoft to build enormous data centers make this seem less daunting.

Several approaches are emerging to address this:

• Simulation: Training robots in virtual environments before deploying to the real world

• Collaborative Data Collection: Projects like Google's RT-X aim to pool data across many robots

• Transfer Learning: Applying knowledge from vision-language models to physical tasks

• Self-Supervised Learning: Allowing robots to learn from their own experiences

Balancing Intelligence and Hardware

The optimal balance between software intelligence and hardware capability remains unresolved. Companies are taking different approaches:

• Boston Dynamics: Prioritized advanced mobility and hardware capabilities

• Figure AI: Emphasized integrated AI and sophisticated hand dexterity

• Tesla: Focused on scalable manufacturing and end-to-end AI systems

• 1X: Prioritized safety, aesthetics, and home integration

This diversity of approaches suggests the market may eventually support multiple specialized types of humanoid robots optimized for different use cases.

Form Factor Debates

The very premise of humanoid robots – machines that replicate human form – remains contested. Critics argue that bipedal robots are unnecessarily complex for many tasks:

• Factory floors are flat and could be served by wheeled robots

• Many tasks could be accomplished by fixed robotic arms or specialized machines

• The human form might be suboptimal for many industrial applications

Proponents counter that humanoid form enables versatility across environments designed for humans, from navigating stairs to operating tools and machinery built for human hands.

While the global economic potential is massive, the many different applications and useful form factors will be beneficial for competition, allowing numerous companies to thrive that prioritie

Investment Implications

While much of the humanoid robotics sector remains private, several investment angles warrant consideration:

Direct Investments

A limited number of public companies offer direct exposure:

Tesla (TSLA): Though primarily an automotive company, their Optimus program represents significant optionality and future revenue potential. Elon Musk has repeatedly emphasized that humanoid robots could eventually generate more revenue than Tesla's automotive business, with projections exceeding $10 trillion long-term.

UBTECH (China): One of the few publicly traded pure-play humanoid robotics companies, UBTECH has developed a comprehensive portfolio of robots for service, education, and commercial applications, giving investors focused exposure to the sector's growth.

Hyundai: As the owner of Boston Dynamics, Hyundai provides investors with exposure to some of the most advanced mobility and manipulation technologies in robotics. The company is integrating these capabilities into their broader manufacturing and mobility strategy, potentially creating synergies across their business lines.

Component and Technology Providers

Companies supplying critical components for humanoid robots may offer more immediate investment opportunities. Investment in the companies that "sell shovels during a gold rush" protect the investment from the failure in any specific robotics company. These include:

NVIDIA (NVDA): Provides AI chips crucial for robot intelligence and training infrastructure. Their specialized GPUs power much of the deep learning behind robot perception and control, and they've developed specific tools for robotics simulation and training that position them as a critical enabler of the entire sector.

Advanced Micro Devices (AMD): Supplies processors for robotics applications with an increasing focus on efficient computing for AI workloads. Their chips are found in various robotics platforms, offering a potential alternative to NVIDIA in certain applications.

FANUC: Leading manufacturer of industrial robots and components with decades of experience in factory automation. Their expertise in precision control systems and manufacturing robotics provides them with opportunities to participate in the humanoid robot supply chain.

Lam Research (LRCX) and ASML (ASML): Produce semiconductor manufacturing equipment essential for advanced robot components. As demand for specialized chips grows with the robotics industry, these companies benefit from increased capacity needs regardless of which robot makers ultimately succeed.

Software and AI

The intelligence layer for robotics offers another investment avenue:

Microsoft (MSFT): Strategic investor in Figure AI with cloud infrastructure for robot training. Their Azure platform provides the computational backbone for much robot development, while their investment in leading robotics companies gives them influence in how the sector evolves.

Alphabet (GOOGL): DeepMind's robotics initiatives and RT-X project represent significant investments in foundational robotics research. Their cross-laboratory collaboration approach to robot learning could establish important standards and capabilities for the broader industry.

Meta (META): Leverages expertise in AI, sensors, and software with sizeable investments in AI and augmented reality. Meta has continuously proven their willingness to enter opportune sectors with minimal friction, and their work on embodied AI and environmental understanding could translate well to robotics applications.

UiPath (PATH): Offers automation software potentially applicable to robotics workflows. Their expertise in process automation could evolve to provide higher-level task planning and coordination for robot fleets, bridging the gap between business processes and physical robot capabilities.

Investment Risks

Several factors could delay or derail the humanoid robotics revolution:

• Technical Challenges: Battery limitations, mobility issues, and manipulation capabilities remain significant hurdles

• Regulatory Concerns: Safety standards, liability questions, and labor regulations could slow adoption

• Economic Viability: High costs may limit widespread deployment in the near term

• Alternative Solutions: Non-humanoid robotic solutions might prove more effective for many applications

Early Days of a Long-Term Trend

While the excitement around humanoid robots is justified, investors should recognize we're in the earliest stages of what will likely be a multi-decade transformation. The most dramatic economic and societal impacts probably won't materialize until the 2030s and beyond.

We don’t recommend investing in any particular company or sector, as we don’t provide investment advice. Though this sector remains full of risk, it is becoming inevitable that robotics will only grow in global market share and will eventually become a regular piece of our daily lives.

With capex from the largest companies in the world to build these robots and tailwinds from the exponential growth in AI capabilities, this isn’t a trend to ignore.

The market is clearly accelerating, with investment dollars flowing in at unprecedented rates. The $40 billion valuation reportedly sought by Figure AI – while raising eyebrows – reflects the enormous potential market if general-purpose humanoid robots become practical reality.

Yes, these will be extremely valuable companies one day, with significant revenues. But it’s pretty hard to justify Figure AI already being worth more than Ford Motors.

Wrapping Up

The humanoid robotics industry stands at a crucial inflection point. After decades of slow progress confined mostly to research labs, we're watching a convergence of AI capabilities, improved hardware, and substantial investment that could finally bring general-purpose robots into factories, businesses, and eventually homes.

While Figure AI's reported $40 billion valuation and Tesla's trillion-dollar revenue projections may seem excessive today, they reflect the enormous potential market if this technology successfully scales.

The total addressable market – potentially representing a significant percentage of global GDP – justifies the investor enthusiasm, even as many technical and economic challenges remain.

Looking Forward

For investors, the opportunity lies not just in the robot manufacturers themselves but in the broader ecosystem of component suppliers, AI providers, and complementary technologies. There WILL be losers — and there IS risk.

Those who thoughtfully navigate this complex landscape may be positioning themselves early in what could become one of the most significant technological transformations of the 21st century.

The humanoid robot revolution won't happen overnight, but its acceleration is unmistakable.

The machines are indeed rising – and they're increasingly capable of understanding us, working alongside us, and perhaps eventually reshaping what it means to be human.

Thanks for reading this Forward Thesis Deep Dive. And remember our weekly market brief will be sent out on Sunday morning.

Until then.