China's SOG raises tens of millions in angel round for AI materials research
What's the deal? Shanghai-based AI-for-materials startup SOG (索格智算)Dealroom has a profile for this one. Try Dealroom → has closed a tens-of-millions-of-yuan angel round. Fosun CapitalDealroom has a profile for this one. Try Dealroom → led, with Nantong Chankong Siyuan AI Pioneer Fund, Yunqi PartnersDealroom has a profile for this one. Try Dealroom →, and Zhongying Venture CapitalDealroom has a profile for this one. Try Dealroom → joining. Existing backer Zizhu Science and Technology Investment added funds.
What does it do? Founded in 2025, SOG uses proprietary AI computing engines to speed up new-materials research. It combines an atomic large model, high-performance simulation, and a lab-based experiment loop to shorten development across scales.
Why now? Traditional materials research relies on trial and error, and moving a material from lab to product can take decades at high cost. As batteries and fine chemicals advance, existing materials struggle to meet demand for higher-performance alternatives.
What's the endgame? The startup is building automation for materials design, spanning design, computational validation, industrial simulation, and self-directed research. Its products include the SOGNet atomic model for lithium batteries and polymers, the NanoTitan simulator, the R2DPack device-simulation platform, and an agent called SOG Omni.
The company says it has already run cell-level simulations for a solid-state battery maker using R2DPack, cutting reliance on repeated physical testing. For a listed chemicals firm, it used SOG Omni to recommend polymer formulations now in wet-lab validation.
Where does the money go? Proceeds will fund ongoing development of its automated research pipeline, lab construction, hiring, and recruiting industry talent, with the goal of advancing domestic high-end materials.
Founder and chief scientist Xu Zhenli is a distinguished professor at Shanghai Jiao Tong University and director of its AI new-materials research centre. The 30-plus-person team spans mathematics, AI, and materials, organised into four research centres covering AI algorithms, applied technology, agents, and quantum computing.
Asked about the company's edge, Xu pointed to its algorithms. "Our core barrier is original algorithms," he said (translated from Chinese), citing three systems that address accuracy, speed, and fidelity in modelling how atoms interact.
The signal: AI-for-materials sits at the intersection of AI and advanced materials, a field investors frame as a route to remake how materials get discovered. Lead backer Fosun Capital called it a direction with "strategic value," a sign that capital is flowing toward efforts to make China's high-end materials supply more self-reliant.
Read more: 36kr.com
Image credit: U.S. Army Combat Capabilities Development Command
