Xingfan Technology provides one-stop AGI industry technology services centered on large models, selling integrated training–inference appliances and cluster products and developing middleware to accelerate large-model performance on heterogeneous chips. Its core product, AI OS, is based on large-model and Agent architectures and automates intent recognition, workflow decomposition, model/tool/compute scheduling, end-to-end training, and application deployment. The company says AI OS can reduce model training configuration from days to minutes (bank practice example: conventional platforms take five days per engineer; AI OS can configure in ten minutes) and supports virtual-pooling and multi-tenant elastic compute to improve utilization. Xingfan also offers a ‘‘compute appliance’’ product line and has deployed industry-specific appliances for education, energy, and security, with plans to expand into construction, cultural-tourism, and media. The firm emphasizes automatic adaptation across different chips to support heterogeneous compute without manual migration steps. It is investing in a full-stack toolchain for large-model training and inference and in performance-adaptation middleware and model intelligentization for compute-center operations. Xingfan Technology develops a proprietary metaverse engine and related digital infrastructure to serve commercial and city metaverse scenarios. The company leverages big data, blockchain, AIGC, AI, data visualization and digital-twin technologies as core capabilities. It aims to enable immersive commercial experiences using virtual spaces and virtual humans and to provide digital-twin and AIGC rapid-production platforms for smart-city verticals. Founded in October 2021 in Chengdu, the founding team combines members from a smart-city technical team with new team members and builds on prior digitalization and AI resources. In July 2022 it completed an angel round of nearly RMB 10 million to fund metaverse engine and AIGC R&D and market expansion. Xingfan positions itself to assist government and enterprise digital transformation and to act as foundational digital infrastructure for the Web3.0 era.

Total raised

$15M

Funding rounds

2

Latest round

Series A

Latest activity

Apr 2024