## Is the Humanoid Robot Market Really Heading for 26 Million Units a Year?

Berg Insight's answer is an unambiguous yes — eventually. The Swedish research firm forecasts annual humanoid robot shipments to grow from **16,000 units in 2026** to **26.0 million units by 2040**, a compound annual growth rate of **63.7 percent**. The corresponding market value is projected to rise from **US$890 million** to **US$554 billion** over the same period. Those are the headline numbers. The more operationally significant figure buried in the report: cellular subscriptions tied to humanoid robots are forecast to hit **50.6 million** by 2040, up from just **5,000 in 2025** — an 84.9 percent CAGR. That connectivity trajectory is the clearest signal that Berg Insight is modeling a world where humanoids are managed as fleet infrastructure, not one-off deployments.

The forecast covers more than 100 companies Berg Insight identifies as currently developing full-size humanoid robots. Named specifically: [Figure AI](https://humanoidintel.ai/companies/figure-ai), [Tesla (Optimus Division)](https://humanoidintel.ai/companies/tesla-optimus), [Agility Robotics](https://humanoidintel.ai/companies/agility-robotics), [Boston Dynamics](https://humanoidintel.ai/companies/boston-dynamics), [Apptronik](https://humanoidintel.ai/companies/apptronik), [1X Technologies](https://humanoidintel.ai/companies/1x-technologies), [NEURA Robotics](https://humanoidintel.ai/companies/neura-robotics), [Unitree Robotics](https://humanoidintel.ai/companies/unitree-robotics), [UBTECH Robotics](https://humanoidintel.ai/companies/ubtech), [AGIBot](https://humanoidintel.ai/companies/agibot), [Leju Robotics](https://humanoidintel.ai/companies/leju-robotics), [EngineAI](https://humanoidintel.ai/companies/engineai), and [RobotEra](https://humanoidintel.ai/companies/robotera). Most remain at research, prototype, or early pilot stages.

---

## What the 63.7% CAGR Actually Requires

A 63.7 percent CAGR over 14 years sounds dramatic because it is. To put it in context: getting from 16,000 units in 2026 to 26 million in 2040 requires the industry to roughly double or better its output nearly every other year, sustained across more than a decade. That is not unprecedented in consumer electronics — smartphone shipments grew at comparable rates during the early adoption curve — but it has never been achieved in a category defined by complex mechanical assembly, tight tolerance actuator manufacturing, and the need for robust real-world [whole-body control](https://humanoidintel.ai/glossary/whole-body-control).

The skeptical read: Berg Insight's forecast is structurally similar to optimistic projections made for industrial robots in the early 2010s, autonomous vehicles in the late 2010s, and delivery drones more recently. Each of those forecasts required both a technology transition and a deployment ecosystem that proved harder to build than analysts anticipated. Humanoids face both challenges simultaneously — and add a third: the need for software ecosystems capable of [zero-shot generalization](https://humanoidintel.ai/glossary/zero-shot-generalization) across unstructured environments, which remains an open research problem.

Berg Insight itself implicitly acknowledges this. The report notes that what distinguishes this phase is "the simultaneous emergence of dedicated manufacturing facilities, commercial deployments and substantial investment" — a careful way of saying the conditions exist, but the execution is not guaranteed.

---

## Strategic Investors Are Treating This as Infrastructure

The report's list of strategic investors is worth pausing on. Berg Insight identifies NVIDIA, Microsoft, Google, Amazon, Hyundai, and OpenAI as among those increasing involvement in humanoid robotics. These are not purely financial bets — each brings a distinct infrastructure interest. NVIDIA has an obvious stake in compute and simulation. Microsoft and Google bring cloud and AI model distribution. Amazon's warehouse deployment appetite is well documented. OpenAI's interest maps directly to [Physical AI](https://humanoidintel.ai/glossary/physical-ai) and the vision-language-action model stack that would make generalist humanoids viable. The presence of this cohort signals that the platform dependencies for humanoid scale — not just the robots themselves — are being competed for aggressively.

Berg Insight also flags that partnerships and M&A are becoming more central to strategy, as companies seek to acquire embodied AI technology, engineering talent, manufacturing capacity, and deployment opportunities. That consolidation signal matters: a 100-plus-company field does not survive at scale, and the report implies buyers are already circling.

---

## The Connectivity Angle Is Not a Side Story

The cellular connectivity forecast — 50.6 million subscriptions by 2040 from virtually nothing today — deserves more attention than it typically receives in humanoid coverage. Remote monitoring, diagnostics, fleet management, OTA software updates, and access to cloud inference are not optional features at scale; they are operational requirements. A single pilot deployment of five robots can be managed with on-premises tooling. A fleet of tens of thousands distributed across warehouses, logistics hubs, and manufacturing floors cannot.

This reframes who the relevant industry players are. Connectivity management, SIM provisioning, remote software distribution, and fleet visibility at that scale start to look like serious infrastructure businesses — roles currently occupied by IoT platform vendors that have never had to manage an endpoint as mechanically complex as a humanoid. The design problem for OEMs expands accordingly: it is not enough to solve locomotion or [dexterous manipulation](https://humanoidintel.ai/glossary/dexterous-manipulation). The robot must also be a manageable connected endpoint with a credible lifecycle story.

---

## Competition Is Shifting to Software and Manufacturing, Not Just Hardware

Berg Insight's central argument — that competition is moving beyond hardware performance toward embodied AI, software ecosystems, and scalable manufacturing — aligns with what the leading teams are demonstrating in practice. Advances in locomotion or manipulation attract attention in demos, but commercial deployments will depend on whether vendors can reproduce those capabilities economically across large production runs.

That manufacturing constraint is underappreciated. Harmonic drives, precision actuators, and custom sensor arrays do not have the commodity supply chains that, say, smartphone components enjoy. Building dedicated manufacturing at scale requires capital, supply chain development, and yield optimization that most current players have not yet demonstrated. The transition from prototype to repeatable product is where most robotics programs historically stall.

The broader implication: the companies most likely to capture Berg Insight's projected market are those that solve the full stack — hardware, [Physical AI](https://humanoidintel.ai/glossary/physical-ai) software, connectivity management, and manufacturing at cost — not those that lead on any single dimension.

---

## Key Takeaways

- Berg Insight forecasts annual humanoid shipments of **26.0 million units by 2040**, up from **16,000 in 2026**, at a **63.7% CAGR**
- Market value is projected to rise from **US$890 million** to **US$554 billion**
- Cellular connections tied to humanoids forecast to reach **50.6 million by 2040**, from just **5,000 in 2025** (84.9% CAGR)
- More than **100 companies** are developing full-size humanoids; most remain at prototype or pilot stage
- Strategic investors named include NVIDIA, Microsoft, Google, Amazon, Hyundai, and OpenAI
- The near-term constraint is not robot capability — it is reproducible manufacturing, software ecosystems, and connected fleet management
- Consolidation is signaled: partnerships and M&A are becoming more central to competitive strategy

---

## Frequently Asked Questions

**How many humanoid robots will be shipped in 2026?**
Berg Insight estimates annual humanoid robot shipments at 16,000 units in 2026. That figure is expected to grow sharply over the following decade and beyond, reaching 26.0 million units annually by 2040.

**What is the projected market size for humanoid robots by 2040?**
Berg Insight projects the humanoid robot market to reach US$554 billion annually by 2040, up from US$890 million in 2026, driven by volume manufacturing and expanding commercial deployments.

**Which companies are leading the humanoid robot market?**
Berg Insight's report names Figure AI, Tesla, Agility Robotics, Boston Dynamics, Apptronik, 1X Technologies, NEURA Robotics, Unitree Robotics, UBTECH Robotics, AgiBot, Leju Robotics, EngineAI, and RobotEra as companies currently shaping the market, though most remain in early commercial or pilot stages.

**Why does cellular connectivity matter for humanoid robots?**
At scale, humanoid robots become connected fleet endpoints requiring remote monitoring, diagnostics, OTA updates, and cloud inference access. Berg Insight forecasts 50.6 million cellular subscriptions associated with humanoid robots by 2040 — a trajectory that creates significant infrastructure requirements for both OEMs and connectivity providers.

**Is a 63.7% CAGR for humanoid robots realistic?**
It is achievable under the most optimistic scenario — analogous to early smartphone adoption curves — but it requires sustained progress on manufacturing scalability, embodied AI software, and fleet management infrastructure simultaneously. Comparable long-range forecasts in autonomous vehicles and delivery drones have historically overestimated near-term adoption while underestimating infrastructure complexity.