The obstacles to scaling up humanoids

Vision vs. Current Reality

• One camp argues the “vision” is a single general-purpose robot that can do thousands of tasks at ~80% of a bespoke machine, yielding huge economies of scale and easy redeployment.
• Critics counter that current humanoids are “worse than humans on every metric”: clumsy, slow, low dexterity, unable to do simple real-world tasks like making a sandwich without heavy staging.
• Many see current marketing and pricing (hundreds of thousands per unit) as wildly out of line with demonstrated capability.

Humanoid Form Factor Debate

• Pro-humanoid side: the world is built for human shape—stairs, doors, tools, cars, cramped kitchens—so a human-like body best exploits existing environments and human tools, and avoids redesigning infrastructure.
• Skeptics: much work is better served by wheeled platforms, fixed arms, AGVs, dishwashers, etc. For factories and warehouses, “robot arm on a mobile base” or other non-humanoid bodies may be simpler, safer, and more efficient.
• Some argue that if robots became genuinely useful, environments might adapt to them (e.g., dumbwaiters instead of stairs), weakening the “must be humanoid” premise.

Economics, Wear, and Maintenance

• Several comments note robots must beat low-wage human labor on total cost, not just wage: productivity, lifespan, training, and maintenance dominate.
• Industrial arms today have low wear-and-tear costs relative to labor and are proven in high-throughput settings; humanoids must match or beat that.
• Debate over longevity: some think heavy wear will make humanoids uneconomical; others argue we could engineer very long-lived machines, but market incentives haven’t favored maximal durability.
• Concerns raised about vendor lock-in: the risk that robot suppliers can “turn off” or throttle an entire workforce via software updates.

AI, Software, and Data

• Broad agreement that hardware is improving but that fine motor control, dexterous hands, and robust, general-purpose control software remain major bottlenecks.
• Supporters point to transfer learning and real-world data from teleoperation and industrial tasks as a path to rapid improvement.
• Timelines are contested: some foresee burger-flipping in ~10 years; others see humanoids as comparable in difficulty to or harder than self-driving cars and expect multi-decade horizons.

Safety, Consumers, and Demand

• Industrial safety and liability, especially for unstable bipedal machines, are seen as major hurdles; relevant standards are still emerging.
• Consumer interest in “chore bots” (laundry folding, cleaning) is acknowledged as huge in theory, but reliability, safety, and price must improve dramatically.
• Several conclude that, today, demand is low because no humanoid robot can yet do anything reliably useful at a competitive cost.