The humanoid robot market hit $2.16 billion in 2026 and is on track to reach $15.26 billion by 2030 at a CAGR of 39.2%. But behind these impressive numbers lies a stark divide: companies that are actually manufacturing and delivering robots versus those still in the perpetual demo stage.
This article cuts through the marketing noise to give you the real state of humanoid robotics in April 2026.
The Shipping Tier
These companies have moved beyond demos to actual production and delivery.
Tesla Optimus (Gen 3)
Tesla converted part of its Fremont factory to Optimus manufacturing in early 2026. While exact shipment numbers remain undisclosed, industry analysts estimate 2,000-5,000 units produced for internal factory use, with limited external sales expected in H2 2026.
Key development: Optimus Gen 3 demonstrated reliable box-moving and parts-sorting tasks in Tesla factories, running 8-hour shifts with minimal human intervention.
AgiBot (Shanghai)
The clear front-runner in China’s humanoid robot race. AgiBot shipped over 5,000 units from its Shanghai facility by Q1 2026, primarily to automotive and electronics manufacturers.
Key development: Standardized “robot-as-a-service” pricing model at approximately ¥15,000/month per unit, making economic payback achievable within 12-18 months for factory deployments.
Unitree H1 / G1
Unitree took a different approach — shipping affordable, research-grade humanoids at price points 10-50x lower than competitors. The H1 ($90,000) and G1 ($16,000) have become the de facto platforms for university research labs worldwide.
Key development: Estimated 10,000+ cumulative units shipped across all robot models, with the G1 becoming the “Raspberry Pi of humanoid robots” for the research community.
The Scaling Tier
Companies with working products that are actively scaling production.
Figure AI (Figure 02)
Figure AI secured a partnership with BMW for warehouse logistics. Figure 02 robots are deployed at BMW’s Spartanburg plant, but volumes remain in the low hundreds. The company is building dedicated manufacturing capacity.
1X Technologies (NEO Beta)
Norwegian company 1X shipped NEO Beta units to select enterprise customers for home and office assistance tasks. Production remains limited but the design-for-manufacturing approach suggests rapid scaling potential.
Fourier Intelligence (GR-2)
China-based Fourier deployed GR-2 units primarily in rehabilitation and healthcare settings, with a few hundred units in the field. Strong in medical applications but limited manufacturing scale.
The Demo Tier
Companies with impressive technology that haven’t yet crossed the manufacturing chasm.
Boston Dynamics (Atlas Electric)
Atlas Electric continues to produce the most visually impressive humanoid demonstrations. However, Boston Dynamics’ historical pattern of amazing demos followed by limited commercialization persists. No mass production plans announced.
Sanctuary AI (Phoenix)
Sanctuary AI’s Phoenix robot demonstrated remarkable dexterous manipulation in controlled settings. The company’s focus on “general-purpose” capabilities may be slowing their path to production, as manufacturing applications require reliability over generality.
Key Market Dynamics
The 95% to 99.9% Reliability Gap
This is the industry’s biggest challenge. Most humanoid robots demonstrate 95%+ success rates in controlled lab settings. But factory deployment requires 99.9% reliability — a gap that represents orders of magnitude more engineering work than the initial 0-95% journey.
Data as the Real Moat
The companies shipping the most units (AgiBot, Unitree, Tesla) are accumulating the most real-world operational data. This data flywheel — more deployments → more data → better models → more deployments — is becoming the primary competitive advantage.
China vs. US: Different Strategies
China: Aggressive scaling, competitive pricing, government subsidies, focus on manufacturing applications. AgiBot and Unitree are moving fast.
US: Higher R&D investment per unit, focus on general-purpose capabilities, venture-funded. Tesla has manufacturing scale but is primarily deploying internally.
Investment Implications
The humanoid robot supply chain is where the real near-term opportunity lies:
- Actuators and motors — Every humanoid needs 20-40 high-performance actuators
- Sensors — LiDAR, force/torque sensors, tactile sensing
- Edge compute — NVIDIA Jetson, Qualcomm Robotics, custom ASICs
- Simulation — NVIDIA Isaac Sim, MuJoCo, custom environments
What to Watch in H2 2026
- Tesla Optimus external sales announcement
- AgiBot’s expansion beyond China
- First humanoid robot IPO (rumored for several Chinese companies)
- Regulatory frameworks for humanoid robots in workplaces
- The 100,000 cumulative units milestone (projected for 2027)
The humanoid robot industry is no longer a science fiction story — it’s a manufacturing scaling challenge. And that shift changes everything about which companies will win.