## Does the Humanoid Form Factor Actually Serve Enterprise Needs?

The answer, according to Erik Nieves, CEO and Co-Founder of Plus One Robotics, is: not yet — and possibly not in the way the industry assumed. Speaking to UC Today, Nieves argued that the humanoid sector made an early strategic error by prioritising bipedal locomotion over the capability that actually unlocks enterprise value: coordinated, [dexterous manipulation](https://humanoidintel.ai/glossary/dexterous-manipulation) between two hands. His benchmark for genuine progress is deliberately concrete — "The day a humanoid can clap its hands behind its back is the day we know we will have achieved true hand-to-hand coordination." That is not a capability any shipping platform demonstrates today. Meanwhile, many enterprise tasks are completed from a fixed position, with materials delivered directly to the worker, making wheeled mobility arguably more practical than a full bipedal [gait cycle](https://humanoidintel.ai/glossary/gait-cycle) for a significant slice of the addressable market. The near-term commercial opportunity Nieves identifies is "fractional tasks" — short-duration jobs across multiple workstations that individually cannot justify dedicated automation but collectively build a viable business case for a general-purpose machine.

---

## Why "Over-Indexing on Legs" Was a Costly Detour

Nieves's critique lands with some precision. The bulk of humanoid demonstrations through 2024 and into 2025 showcased robots walking, balancing, and carrying totes — all impressive proofs of locomotion control, but a relatively thin argument for an enterprise procurement officer evaluating capital expenditure.

"A lot of the demonstrations were just a robot carrying a tote," Nieves told UC Today. The implicit criticism: locomotion is table stakes for a humanoid brand story, but it does not differentiate value in a warehouse or factory where conveyors, AMRs, and purpose-built fixtures already handle material transport efficiently.

The harder problem — and the one that would genuinely expand the addressable task space — is bimanual coordination. Assembling components, folding garments, handling irregular items, or performing any task that requires two hands to work in concert demands a fundamentally different capability than walking across a factory floor. This is where the gap between demonstration and deployment remains widest across the industry.

Nieves's framing implicitly challenges teams building [whole-body control](https://humanoidintel.ai/glossary/whole-body-control) policies that prioritise loco-manipulation fluency. The locomotion side of that equation may be close enough for many enterprise environments. The manipulation side is not.

---

## The "Fractional Tasks" Thesis and Its Commercial Logic

The most commercially interesting idea in Nieves's analysis is the fractional tasks framework. Traditional automation economics are straightforward: high-volume, high-speed, repetitive processes justify purpose-built systems — a dedicated industrial arm with specialised tooling will always outperform a general-purpose humanoid on throughput for a single task.

But enterprises carry a long tail of lower-intensity work. A two-hour job at one station, a few hours supporting a different process, intermittent quality checks — none of these individually cross the return-on-investment threshold for bespoke automation. Collectively, routed across a shift to a single deployable humanoid, they might.

This is the utilisation argument, and it is where the general-purpose form factor earns its premium over specialised machines. If a humanoid can be redeployed across task types within a single facility — without retooling, reprogramming from scratch, or physical reconfiguration — then the per-task cost calculus changes.

The caveat is that this thesis depends entirely on the software stack. [Zero-shot generalization](https://humanoidintel.ai/glossary/zero-shot-generalization) across novel task instances, or at minimum rapid few-shot adaptation, is a prerequisite. That is precisely the capability that companies like [Physical Intelligence (π)](https://humanoidintel.ai/companies/physical-intelligence) and [Skild AI](https://humanoidintel.ai/companies/skild-ai) are racing to deliver at the policy layer. The hardware may be approaching sufficient capability for fractional tasks; the AI generalization is the binding constraint.

---

## Security Risk Is Not Theoretical

Nieves's security commentary deserves more attention than it typically receives in humanoid coverage. He identifies two distinct threat vectors: unauthorized access to sensitive operational data flowing through connected robots, and the potential for a compromised system to disrupt industrial operations directly.

His reference point is Stuxnet — the malware that targeted industrial control systems rather than conventional IT infrastructure. The analogy is apt. Humanoids deployed in manufacturing or logistics environments will have network connectivity, sensor access to physical processes, and in some configurations direct actuation authority over equipment. That attack surface is qualitatively different from a standard enterprise IoT device.

As US policymakers increase scrutiny of connected foreign-manufactured robots, this is not an abstract concern. The operational technology security challenge for humanoids will require the same rigor that industrial control system vendors have developed over the past decade — and the industry does not yet have established standards for it.

---

## AMR Supply Chain Is the Near-Term Pressure Point

Nieves expects the most immediate near-term disruption from evolving US trade and technology policy to fall on autonomous mobile robots rather than humanoids. AMRs are already widely deployed in warehouse environments, and he noted that domestic manufacturing capacity for them remains limited. The outcome will depend heavily on the specifics of any exemption process.

For humanoid manufacturers, this is an early warning. Supply chain concentration — particularly in actuators, sensors, and compute — is a structural vulnerability that policy shifts can expose quickly. A US robotics strategy that does not address component manufacturing capacity and engineering workforce depth will not succeed regardless of policy intent. Nieves's prescription is direct: "If we get supply chain, engineering capacity and workforce right, it'll be fine."

That conditional is doing significant work. None of those three elements is currently in place at the scale the industry's deployment ambitions require.

---

## Key Takeaways

- **Locomotion vs. manipulation**: Plus One Robotics CEO Erik Nieves argues the humanoid sector over-indexed on bipedal legs when bimanual hand coordination is the harder, more commercially relevant capability.
- **The clapping test**: Nieves's informal benchmark — a robot clapping its hands behind its back — signals the level of coordinated dexterity needed to unlock real enterprise task coverage.
- **Fractional tasks thesis**: Humanoids' near-term commercial case may rest on aggregating short-duration, varied jobs that individually cannot justify dedicated automation.
- **Utilisation is the metric**: For enterprise buyers, throughput utilisation across diverse tasks matters more than peak performance on any single operation.
- **Security is a real risk**: Operational technology attack surfaces in connected humanoids mirror industrial control system vulnerabilities — the Stuxnet analogy is deliberately chosen.
- **AMRs face the first policy pressure**: Domestic supply chain limitations make AMRs more immediately exposed to trade policy disruption than humanoids, but the structural lesson applies to both.
- **Three prerequisites for a viable US strategy**: Supply chain depth, engineering capacity, and workforce investment — none currently at scale.

---

## Frequently Asked Questions

**Why did humanoid robots focus on legs rather than hand dexterity first?**
Bipedal locomotion was the primary technical differentiation signal — it justified the "humanoid" label and generated compelling demonstrations. Nieves argues this prioritisation was misplaced for enterprise use cases, where most tasks require coordinated manipulation rather than locomotion, and where wheeled mobility is often more practical.

**What are "fractional tasks" in humanoid robotics?**
Fractional tasks are short-duration jobs — a few hours each — spread across different workstations or processes within a facility. Individually, none justifies dedicated automation investment. Aggregated across a shift, they may create a viable utilisation case for a general-purpose humanoid that can be redeployed without retooling.

**How significant is the cybersecurity risk for enterprise humanoids?**
Nieves identifies it as material. Connected humanoids with sensor access to physical processes and network connectivity present an operational technology attack surface analogous to industrial control systems. His Stuxnet reference points to disruption of physical infrastructure, not just data theft, as the relevant threat model.

**Will US trade policy affect humanoid robots the same way it affects AMRs?**
Nieves expects AMRs — already widely deployed and reliant on limited domestic manufacturing capacity — to face near-term disruption first. Humanoids face the same structural supply chain vulnerabilities but are earlier in deployment cycles. The details of any exemption process will be decisive.

**What capability would signal that humanoids are genuinely ready for enterprise dexterous tasks?**
Per Nieves: the ability for a robot to clap its hands behind its back — a test of coordinated bimanual reach across the full workspace. It is a proxy for the kind of two-handed coordination required for real assembly, handling, and manipulation tasks that current demonstrations have not yet convincingly shown.