Jiangxing Intelligence has consecutively secured hundreds of millions of yuan in Series C and Series D strategic financing, with participation from industrial capital and strategic investors. The raised capital will be mainly used for recruiting top-tier talents, launching computing power centers, establishing data collection factories and commissioning the production line for embodied AI controllers, so as to further strengthen the company’s full-stack closed-loop capabilities in the Physical AI Brain sector.

The Physical AI industry is witnessing a crucial division: players focusing on physical embodiments and those dedicated to developing AI brains. Embodiments are designed to act upon the physical world, while AI brains are built to understand the physical world — and we solely focus on the latter. Serving as a reusable, cross-embodiment and cross-scenario layer, a sophisticated AI brain deployed on a robotic arm today can be applied to quadruped robots and drones tomorrow. Capabilities trained in substation scenarios can be migrated to photovoltaic stations, mines and industrial production lines. Such compound value gains are unattainable for hardware embodiments, yet achievable with AI brains.

In this sense, our business model follows the "asset-light non-vehicle manufacturing" logic. Instead of competing with hardware manufacturers, we empower their devices by embedding our AI brains to enable reliable field operations. The more hardware partners we collaborate with, the wider the deployment coverage of our AI Brain solution.

While the industry was debating whether Physical AI is a false proposition, we have already deployed multimodal large models to process field data, train algorithms and refine commercial implementation in real industrial scenarios.

2026 is our year of commercial delivery and value realization.

All our previously invested technological research achievements have been transformed from academic papers into commercial products, converging into two core product lines: JX-Phi World and JX-Phi Brain.

MoW (Mixture-of-World Models) and OC-STORM (Object-Centric STORM), two research papers accepted by ICLR 2026, tackle two core challenges: whether an AI brain can accurately perceive on-site environments and predict upcoming events. OC-STORM has been listed as one of the world’s five leading technical routes in authoritative industry reviews on world models, demonstrating that we are not merely following existing technological paths but pioneering one of the core paradigms.

DyGRO-VLA, accepted by ICML 2026, empowers AI brains to accurately execute tasks perceived from real scenarios. World models and Vision-Language-Action (VLA) frameworks are not competing technical routes but complementary components of an integrated system. JX-Phi World understands the physical world while JX-Phi Brain interacts with it, and their closed-loop collaboration constitutes a complete Physical AI Brain.

JX-Phi Brain enables robotic arms to complete panel button operations

 

Physical AI has moved beyond the technical feasibility verification stage and entered an era that demands stable, large-scale and replicable commercial delivery. What clients need is not an eye-catching technical demo, but an AI Brain that can be reliably deployed hundreds of times across production lines.

In the next 12 to 18 months, our progress can be measured against four clear milestones: assembling a top-tier R&D team, putting computing power centers into official operation, completing the construction of data collection factories, and launching the embodied AI controller production line.

This round of financing is not intended for blind scale expansion to boost short-term revenue. Instead, it aims to build, deepen and consolidate the full-chain closed loop covering scientific research, mass production and commercial delivery.

Quadruped robots equipped with JX-Phi Brain Physical AI system perform autonomous operations at hydropower stations

Physical AI belongs to a resource-intensive track, where algorithmic advantages alone cannot drive sustainable industrial iteration.

We have integrated four core strengths: industrial partners provide diverse real-world scenarios and massive field data; global top scientists and engineers set the capability ceiling of our AI Brain; government authorities offer policy support for our data collection factories and computing infrastructure. Our full-stack self-developed technology system, paired with robust industrial infrastructure, accelerates the iterative upgrading of Physical AI.

Our core competitive barrier in Physical AI is not a single technical threshold, but a self-amplifying flywheel formed by the synergy of these four forces.

Spectacular demos do not represent core industrial competitiveness, and published academic papers do not equal commercial delivery. All our research outcomes have been fully deployed to industrial sites this year.

There are no shortcuts or superficial gimmicks in Physical AI. The most credible performance credential lies in deploying our AI Brain to consistently deliver practical operational results across numerous industrial scenarios.