# Is PsiBot's $100M Round a Sign That Industrial Capital Is Taking Over Chinese Humanoid Funding?

More than $100 million. That's the headline figure from PsiBot's latest funding round — and the investor list tells a story as important as the number itself. The Chinese embodied-AI startup, which focuses on [dexterous manipulation](https://humanoidintel.ai/glossary/dexterous-manipulation), closed the round with participation from Ningbo Tuopu Group, a Chery Holdings-backed fund, Lens Technology, 37 Interactive Entertainment, Wuhu Investment Holding Group, and Fosun Fortune Capital. Existing shareholder Zhuhai Technology Industry Group also recommitted capital, according to TechNode reporting published today.

The round will fund three specific workstreams: embodied-world-model research, human-operation data collection, and deployments targeting logistics and advanced manufacturing. On the technical side, PsiBot is building a dual-system architecture pairing its Psi-R2 operation-policy model with the Psi-W0 action-conditioned world model — a structure that places the company squarely in the world-model camp of embodied-AI development, alongside international peers like [Physical Intelligence (π)](https://humanoidintel.ai/companies/physical-intelligence).

The investor mix — dominated by an automotive-parts supplier, a precision glass manufacturer, and a regional state-backed holding group — marks a notable departure from the venture-capital-heavy rounds that characterized the 2024–2025 Chinese humanoid boom. Industrial strategic capital is now actively writing large checks into foundation-layer manipulation AI.

---

## Who Are These Investors and Why Does It Matter?

The composition of PsiBot's cap table deserves close reading. Ningbo Tuopu Group is a Tier-1 automotive supplier with established manufacturing relationships across the Chinese EV sector. Lens Technology is a precision components manufacturer best known for producing display glass. Neither is a typical software or deep-tech VC.

Their presence signals something specific: end-users and manufacturing infrastructure players are beginning to co-invest in the AI stack that will govern their own automation roadmaps. This is not patient financial capital betting on a 10-year exit — it's strategic capital seeking preferential access to manipulation policy models that could run on future production lines.

The Chery Holdings-backed fund adds an automotive OEM angle. Chery, one of China's major passenger-car manufacturers, has a direct interest in factory-floor automation as it competes on cost structure with domestic EV rivals. A stake in a dexterous manipulation AI startup is a hedge on labor cost exposure and an option on future in-house deployment.

Wuhu Investment Holding Group and Zhuhai Technology Industry Group are both municipal state-backed vehicles — a pattern common in Chinese deep-tech funding where local governments use investment as an industrial policy instrument to anchor technology companies in their jurisdictions.

Fosun Fortune Capital and 37 Interactive Entertainment represent more traditional financial and conglomerate capital. 37 Interactive's presence is unusual for a hardware-adjacent robotics round, but Chinese gaming and entertainment conglomerates have increasingly diversified into [physical AI](https://humanoidintel.ai/glossary/physical-ai) bets over the past two years.

---

## The Dual-System Architecture: What Psi-R2 and Psi-W0 Actually Do

PsiBot's technical framing — pairing an operation-policy model (Psi-R2) with an action-conditioned world model (Psi-W0) — is the most substantive technical disclosure in the source material, and it's worth unpacking.

An action-conditioned world model generates predictions about how the environment will change in response to specific robot actions. This is architecturally distinct from a pure [vision-language-action model](https://humanoidintel.ai/glossary/vision-language-action-model), which maps perception and language directly to motor commands. By separating the world model (Psi-W0) from the policy model (Psi-R2), PsiBot is betting that a robot can plan more robustly by simulating the consequences of its own actions before committing to them.

This dual-system framing has parallels in cognitive science — System 1 / System 2 thinking — and has been gaining traction in robotics research as a path toward better [zero-shot generalization](https://humanoidintel.ai/glossary/zero-shot-generalization) in novel environments. The key technical question PsiBot will need to answer: does the world model generalize beyond its training distribution, or does it simply memorize the structured environments used during human-operation data collection?

The funding allocation toward human-operation data collection is telling. High-quality teleoperation data remains a genuine bottleneck across the entire manipulation stack — it's expensive, slow to collect, and notoriously difficult to scale. The fact that PsiBot is dedicating a specific budget line to this suggests their world-model training is still substantially data-hungry, rather than relying primarily on [sim-to-real transfer](https://humanoidintel.ai/glossary/sim-to-real-transfer).

---

## What This Means for the Broader Chinese Embodied-AI Landscape

PsiBot's round is the latest data point in a structural shift in how Chinese embodied-AI companies are being capitalized. Earlier rounds in this sector were predominantly led by technology-focused VCs and state guidance funds with generalist mandates. The 2026 vintage is increasingly characterized by industry-specific strategic investors — automotive, manufacturing, logistics — who are making bets tied to their own operational timelines.

This creates a different kind of pressure than pure VC funding. Strategic investors want deployment visibility, not just research milestones. PsiBot's stated focus on logistics and advanced manufacturing deployments — rather than domestic service applications — aligns with what these investors want to see.

For comparison, [AGIBot](https://humanoidintel.ai/companies/agibot) and [Unitree Robotics](https://humanoidintel.ai/companies/unitree-robotics) have each pursued different investor mixes and deployment priorities. The Chinese market is fragmenting into distinct segments: general-purpose humanoids chasing broad deployment, and more focused manipulation-AI companies like PsiBot targeting specific industrial verticals with a tighter technical thesis.

The skeptical read: $100 million is substantial but not exceptional for this stage in the Chinese humanoid funding cycle. What matters more is whether PsiBot can demonstrate that Psi-W0's world-model predictions translate into measurable manipulation reliability improvements over a standard end-to-end VLA approach — and whether its industrial investors will give it the factory-floor access needed to prove it.

---

## Key Takeaways

- PsiBot has closed a funding round of more than $100 million, per TechNode reporting dated August 28, 2026.
- Investors include Ningbo Tuopu Group, a Chery Holdings-backed fund, Lens Technology, 37 Interactive Entertainment, Wuhu Investment Holding Group, Fosun Fortune Capital, and existing backer Zhuhai Technology Industry Group.
- The investor base skews heavily toward industrial strategic capital — automotive suppliers, precision manufacturers, and regional state vehicles — rather than pure-play VC.
- Funds are allocated to embodied-world-model research, human-operation data collection, and logistics/advanced manufacturing deployments.
- PsiBot's core technical bet is a dual-system architecture: the Psi-R2 operation-policy model paired with the Psi-W0 action-conditioned world model.
- The round reflects a broader pattern of end-user industries co-investing in the foundation-layer AI they expect to deploy — a structurally different dynamic than the VC-led rounds of 2024–2025.

---

## Frequently Asked Questions

**What does PsiBot do?**
PsiBot is a Chinese embodied-AI startup specializing in dexterous manipulation. The company is building a dual-system AI architecture — combining an operation-policy model (Psi-R2) with an action-conditioned world model (Psi-W0) — aimed at deployment in logistics and advanced manufacturing.

**How much has PsiBot raised in its latest round?**
PsiBot raised more than $100 million in its latest funding round, according to TechNode reporting published on August 28, 2026.

**Who invested in PsiBot's latest round?**
The round included Ningbo Tuopu Group, a Chery Holdings-backed fund, Lens Technology, 37 Interactive Entertainment, Wuhu Investment Holding Group, Fosun Fortune Capital, and existing investor Zhuhai Technology Industry Group.

**What is an action-conditioned world model in robotics?**
An action-conditioned world model is an AI component that predicts how the environment will change in response to a robot's planned actions. By simulating the consequences of potential moves before executing them, it allows a robot to plan more robustly — particularly in novel or unstructured environments.

**How does PsiBot compare to other Chinese humanoid and manipulation AI startups?**
PsiBot's focus on manipulation-specific world models and industrial deployment distinguishes it from broader-platform humanoid companies. Its investor base — heavy on automotive and manufacturing strategic capital — suggests a tighter vertical focus than general-purpose humanoid competitors in the Chinese market.