From underlying chips and neuromorphic computing to autonomous robot deployment, physical intelligence in extreme environments, and modeling of the Earth's physical environment, cutting-edge AI technologies are advancing industrial applications along multiple distinct paths. How to enable original laboratory achievements to bridge the distance between engineering validation and commercialization has become a key issue in translating scientific innovation into industrial practice.
On October 26, the first roadshow of the "Yuanchuang Program" released by the Beijing Academy of Artificial Intelligence (BAAI) was held at the Zhongguancun International Technology Trading Center. Five projects—Linjie He Keji (Critical Core Technology), Lingshu Qizhi, Tensor Field Technology, Cangqiong Jizhi (Sky Extreme Intelligence), and Latent Vector Technology—took the stage collectively, focusing respectively on 3D self-selecting V-RRAM chips, neuromorphic embodied intelligence, Hamiltonian world models, reliable physical intelligence for extreme environments, and a unified world model of the surface physical environment. The event also included a project-fund negotiation session to facilitate exchanges and connections between startup teams and investment institutions.
From "source innovation" to "industrial validation": what the Yuanchuang Program aims to connect is more than just funding
At the event, Yang Yang, Vice President of the Beijing Academy of Artificial Intelligence (BAAI), introduced the positioning and support mechanisms of the "Yuanchuang Program". As an incubation program for original technology entrepreneurship primarily targeting the BAAI Scholar community, the Yuanchuang Program focuses on original research topics in the field of artificial intelligence, follows the principles of "source innovation, rapid decision-making, small-step piloting, and full-process compliant transformation", and integrates technology, talent, capital, and industry resources to support research teams on their journey from the laboratory to industry. The program not only focuses on funding connections, but also strives to build a collaborative mechanism linking original research, entrepreneurial teams, and industrial markets, promoting scientific research achievements from technological discovery to engineering validation and commercial application.
The launch of the Yuanchuang Program also responds to the practical needs of artificial intelligence industry development and the commercialization of scientific and technological achievements. At present, AI is at a critical stage of moving from technological breakthroughs toward industrial application, and the engineering validation, scenario adaptation, and commercialization of original technologies are receiving growing attention. At the same time, policies and mechanisms related to the commercialization of scientific and technological achievements continue to improve, providing new conditions for research institutions to explore the transformation of original results. The Yuanchuang Program hopes to seize this window of opportunity to further strengthen the connection between scientific innovation and industrial demand, and to support more promising original technologies in reaching the market.
Five projects presenting diverse explorations of original AI technology
1. Critical nuclear technology: from original devices to software and hardware platforms
Linci Nuclear Technology focuses on 3DV-RRAM chips, building a portfolio spanning underlying self-selecting RRAM devices, chips, the FlagOS software ecosystem, and system applications, committed to creating a full-stack product system with software-hardware synergy. The project takes storage-class memory and compute-in-memory as its entry points, exploring a new path for the coordinated development of novel memory and AI computing power, aiming to break through the memory and computing bottlenecks faced by existing computer architectures. The team is led by a national-level young talent from Tsinghua University and won a provincial and ministerial-level Natural Science First Prize (2025), combining strengths in both fundamental research and engineering development.
2. Lingqiu Inspiring Intelligence: Reconstructing the Perception Architecture of Embodied Intelligence with Neuromorphic Computing
Lingshu Qizhi focuses on cross-scale multimodal perception and neuromorphic embodied intelligence, and is committed to building physical perception and brain-inspired computing infrastructure for embodied intelligence. The project extends from underlying optical devices and perception-computation integrated chips to model algorithms, exploring a full-stack technology architecture from perception to decision-making.
Its core goal is not only to enable agents to perceive the external environment, but also to understand material properties and human intentions, and on this basis to achieve millisecond-level autonomous decision-making. Through the co-design of devices, chips, and algorithms, the project explores new paths to improve embodied intelligence's environmental understanding and real-time decision-making capabilities.
3. Tensor Field Technology: Advancing Autonomous Robot Deployment with Hamiltonian World Models
Tensor Field Technology targets high-value scenarios with high human-factor risks where engineers struggle to iterate through repeated debugging, building a robot deployment engine driven by a Hamiltonian world model, offering full-chain services including environment modeling, policy generation, edge-side deployment, and continuous evolution.
The team proposes a Hamiltonian world model architecture that improves the physical consistency of model outputs through physics-based priors. According to the project introduction, its pretraining data efficiency is improved by 3-4 orders of magnitude compared with mainstream approaches. The model can adapt to multiple robot embodiments, and the project also proposes a "Deployment-as-a-Service" model, planning to develop in coordination with robot body manufacturers, hardware suppliers, and cloud computing platforms. The team won third place in the Tsinghua University President's Cup and has completed angel round financing; the company was registered in Haidian, Beijing in August 2026.
4. Cangqiong Jizhi (Sky Extreme Intelligence): Reliable Physical Intelligence in Extreme Environments
Targeting extreme environments such as deep space and deep underground, Cangqiong Jizhi focuses on physical intelligence infrastructure construction, attempting to solve problems in extreme scenarios such as data being difficult to obtain, models being difficult to train, and intelligent systems being difficult to deploy.
The team proposes the AI4π (AI for Physical AI) technology paradigm, building a technical closed loop covering agent design, training, testing and improvement through a physically embedded simulation engine, a self-evolving hierarchical brain training-and-inference system, and a pan-ontology operating system, exploring ways to break through the data and engineering bottlenecks in physical intelligence R&D.
The team is led by national-level leading talent and draws on the research resources of Beihang University and BAAI; it has already won more than ten international awards, including an Outstanding Paper at a top conference and the global third place at the World Humanoid Robot Games. Currently, the project is exploring commercialization through simulation data subscriptions and system licensing, and has already secured orders worth tens of millions of yuan in fields such as park maintenance and commercial aerospace.
5. Potential Vectors Technology: Unifying the Virtual and the Real with a Unified World Model of Surface Physical Environments
QuantVector Technology focuses on a unified world model of the surface physical environment. Its team independently developed EWM (Earth World Model), which incorporates external driving factors such as atmospheric forcing and solar radiation, surface processes such as soil, vegetation, water bodies, mountains, and the carbon cycle, as well as human systems such as power stations, farmland, industry, and infrastructure into a unified latent space for representation, exploring the construction of a general-purpose physical foundation model that is intervenable, invertible, and falsifiable.
The project adopts a token-based billing model for its model services, and its new-energy scenario has completed empirical validation at 12 sites across 5 countries. In photovoltaic scenarios, the team uses its self-developed Boson series cleaning and operations robots as execution terminals, working with cloud models to build a closed loop of "perception—decision—execution—feedback," exploring a path for the collaborative deployment of surface physical environment modeling and embodied intelligent systems.
Connecting original technology with capital, driving research achievements toward industry
From original chips to world models, from brain-inspired embodied intelligence to autonomous robot deployment and physical intelligence in extreme environments, the five projects present diverse explorations of original AI technology across underlying hardware, computing architecture, model development, and physical applications. What they have in common is an attempt to transform cutting-edge research capabilities into verifiable, deployable, and sustainable technologies and products.
For hard-tech companies founded by research teams, technological breakthroughs in the laboratory are only the starting point. The subsequent engineering validation, product development, scenario adaptation, and exploration of business models all require sustained resource investment and industrial collaboration. How to effectively connect scientific innovation, entrepreneurial teams, capital, and the market is a crucial link in translating original technologies into industrial applications.
The Source Innovation Program roadshow, through project presentations and fund negotiations, built a platform for exchange and matchmaking between startup teams and investment institutions, further promoting the connection of original technologies with capital and industrial resources. The event showcased the original explorations of BAAI scholars in different technical directions, and also presented diverse paths from mining research achievements to nurturing entrepreneurial projects.
From the laboratory to the front lines of industry, the transformation of original technology requires not only sustained scientific breakthroughs, but also the combined support of engineering capability, application scenarios, and industrial resources. The Yuanchuang Initiative will continue to focus on cutting-edge original achievements in artificial intelligence, promoting the effective connection between scientific innovation and industrial demand, and supporting more promising original technologies on their way to industrial practice.
BAAI Scholars: From "Dream Team" to "New Industrial Force"
The BAAI Scholars Program is a flagship initiative by the Beijing Academy of Artificial Intelligence (BAAI) to build a high-level team of researchers in cutting-edge basic research. It aims to select and cultivate academic leaders with international influence and young academic talents with development potential, establish collaborative innovation teams oriented toward major research tasks and future frontiers, lead the future direction of artificial intelligence technology, and build an ecosystem for AI industry innovation. Over the past seven years, the BAAI Scholars Program has evolved through three stages—free exploration, task-driven development, and innovation leadership—and has become an important vehicle for attracting and cultivating top talent in the field of artificial intelligence. The number of BAAI Scholars has exceeded 180, forming a global AI 'Dream Team'.
More importantly, this "dream team" is becoming an important source of AI entrepreneurship. In recent years, BAAI has incubated more than 20 innovative companies with core technological capabilities, including Zhipu AI, ModelBest, Galbot, Xingyuan Zhi, Inverse Matrix, and Ziwuxian Guangnian. From large models to embodied intelligence, from underlying vision to safety governance, behind a group of representative AI startups one can see the presence of BAAI scholars and the BAAI ecosystem. The launch of the first roadshow of the Yuanchuang Program is the latest footnote to the transformation of this talent potential into industrial momentum.