Founded by a former Tesla team, with an investment from the first angel investor of OpenAI, tens of millions of dollars are being bet on industrial embodied intelligence | Exclusive from 36Kr

Author | Huang Nan

Editor | Yuan Silai

36Kr learned that IndustrialNext, an industrial general embodied intelligence enterprise, recently completed tens of millions of dollars in Series A financing. This round was led by global top venture capital firm Khosla Ventures, with existing shareholders Y Combinator and Feidian Capital continuing to increase their investments. Gaojie Capital served as the exclusive financial advisor. Khosla Ventures, as the first institutional investor in OpenAI, has a keen insight into technological trends and a long - term investment style, and is a globally renowned technology - investment - driven VC.

The funds from the Series A financing will mainly be used for team expansion, R & D investment, mass production and delivery, and global market expansion.

Currently, IndustrialNext has completed a total of three rounds of investment. The investors include a diverse group of shareholders, including industrial investors Lenovo and Xiaomi, and financial investors such as Khosla Ventures, Y Combinator, Feidian Capital, and Miracle Plus.

Since its establishment, IndustrialNext has been focusing on the R & D and application of embodied intelligence technology in the industrial vertical field. Based on its self - developed end - to - end embodied AI algorithm, it starts from the actual industrial needs to connect the perception - decision - control closed - loop, and provides a general embodied intelligent manufacturing platform for the core scenarios of the manufacturing industry and enterprise customers.

As a cross - domain innovation group, IndustrialNext is composed of elite engineers from global technology companies. The core founder and CEO, Allen Pan, was formerly the person - in - charge of Tesla's AI autonomous factory project. The co - founder and CTO, Lukas Pankau, was the chief software architect at Tesla's Autopilot. The founding team is mainly composed of members from Tesla and Google. They have successfully applied the underlying AI technology of autonomous driving to multiple projects on Tesla's production line. From 2017 to 2019, they witnessed the construction of Tesla's Model 3, the world's first fully automated automobile production line, from scratch, significantly reducing production costs and Bill of Materials (BoM) costs.

The company has also recruited core algorithm experts from Google's DeepMind RT series of embodied models. Many of its core technical backbones come from technology giants such as SpaceX, Boston Dynamics, Mercedes - Benz, and Momenta, and have significant advantages in cutting - edge AI technology, manufacturing insights, and a global perspective. Currently, the team size is still expanding.

In the past, in the hardware - centric industrial automation field, affected by the fluctuations in the trade environment and the frequent changes in upstream products, the efficiency of fixed - asset investment has been greatly reduced, and the production line utilization rate has been in trouble. During this period, due to the characteristics of traditional equipment that often require pre - set tasks and repetitive execution, this model lacks the ability of autonomous learning and dynamic iteration, and has clearly reached its capacity limit.

Specifically in the production process, on the one hand, the production line adjustment cycle is long. Adjusting production line processes and deploying new products rely on manual programming by engineers and hardware modification, accompanied by a large amount of debugging. It often takes weeks or even months from the proposal of requirements to implementation, making it difficult to keep up with the current rapid product iteration rhythm.

On the other hand, the intelligence level of traditional automation equipment is limited, and its ability to perform complex tasks is insufficient. For example, in the face of common unstructured scenarios in production, such as adaptive grasping and placing of irregular workpieces, fastening and assembly of process parts, and dynamic locking of complex curved surfaces, it often requires the coordinated operation of embodied intelligent equipment with a perception - decision - execution closed - loop to meet the needs of flexible production.

Traditional manufacturing platforms cannot improve their intelligence level through data accumulation and algorithm iteration. Any fluctuations different from the pre - set input will lead to a decrease in the accuracy of output results or an increase in the failure rate, let alone complete complex task operations beyond the pre - set programs. This has also become a key bottleneck restricting the long - term upgrade of the manufacturing industry to high - end and intelligent levels.

Noticing the common pain points in the industry, such as inefficient equipment debugging, limited intelligence, and poor adaptability to flexible production, IndustrialNext hopes to further optimize and promote the production line technology that has been verified in Tesla's autonomous factory.

In the second half of 2024, IndustrialNext launched the first - generation general embodied intelligent manufacturing platform for the manufacturing industry. This platform reconstructs the underlying software architecture through the embodied AI algorithm, combined with a plug - and - play modular hardware design, and self - develops an embodied intelligent manufacturing platform integrating hardware and software. It has the capabilities of senior engineers and skilled operators, and can learn quickly, schedule flexibly, and configure autonomously.

Allen Pan told 36Kr that in general production processes such as general assembly, testing, and packaging, due to reasons such as high requirements for process flexibility, frequent changes in upstream demand, and low input - output ratio, there are still a large number of "blind spots" that cannot be covered by automation in some production lines. After introducing IndustrialNext's embodied intelligent manufacturing platform, by learning actual production data and simulation data, robots can efficiently learn the assembly processes of manual or automated equipment and be quickly deployed on the production line. Robots can replace operators to complete flexible and complex on - site production tasks. Experienced operators do not need to leave their posts but instead take on the roles of operation guidance and process correction, realizing a functional upgrade from "executors" to "mentors".

The advantage of this model is that it not only avoids the repetitive investment in fixed assets of equipment during product production line iteration but also directly reduces the comprehensive cost and improves production flexibility. It can effectively fill the gap in the capabilities of traditional automation solutions in high - dynamic and unstructured scenarios.

In terms of its commercialization strategy, IndustrialNext aims at the global market. On the one hand, relying on the advantages of China's manufacturing industry, it jointly develops and iterates products with key accounts (KA) to quickly build product capabilities. On the other hand, for overseas manufacturing customers, it launches a standardized, ready - to - use embodied intelligent manufacturing platform to quickly achieve commercial scale - up.

36Kr learned that IndustrialNext's embodied intelligent manufacturing platform has been verified by global top - tier customers in the 3C and automotive industries. In some production line scenarios with high flexibility and rapid iteration, it can achieve a shorter delivery cycle, higher production rhythm, and lower losses, and has received small - batch orders worth tens of millions of yuan.

In the future, IndustrialNext will gradually expand to high - end manufacturing scenarios in multiple industries and accelerate the global implementation of industrial embodied intelligence.