SenseTime launched the Galaxy Project to scale domestic AI chip infrastructure in China through an ecosystem of nearly 20 partners. Unveiled by company co-founder Yang Fan, the initiative connects chip-level technology, infrastructure partnerships and commercial deployment to support locally produced AI computing capacity.
Partners include Cambricon, Huawei Ascend, Moore Threads, Biren Technology and other domestic chip and infrastructure companies. SenseTime also introduced a Tokens Per Watt efficiency metric and a Computing-Power Collaboration Agent for resource scheduling, electricity price prediction and energy optimization. The project includes plans for one token factory, five large-scale computing clusters, collaboration across 10 technology areas and support for 200 AI startups.
For enterprises, the initiative aims to simplify workload migration across different domestic chips through full-stack hardware and software adaptation. The rollout reflects China’s broader push to commercialize domestic AI processors, improve energy efficiency and build more resilient local computing supply chains.
Image Credit: SenseTime
What's Driving This Trend
- Hybrid AI Clusters
- Multi-vendor computing environments create room for orchestration platforms that simplify workload migration across heterogeneous domestic chip architectures.
- Tokens-per-watt Metrics
- Energy-linked AI performance benchmarks introduce new ways for enterprises to evaluate compute efficiency, infrastructure costs and sustainable model deployment.
- AI Resource Scheduling Agents
- Autonomous systems for electricity forecasting and workload allocation are reshaping how large-scale AI clusters optimize capacity, cost and uptime.
Who This Affects Most
- AI Infrastructure
- Domestic cluster ecosystems are expanding opportunities for localized compute platforms that reduce reliance on global chip supply chains.
- Semiconductors
- Commercial demand for AI processors is accelerating innovation in chip compatibility, full-stack adaptation and performance optimization across diverse hardware vendors.
- Energy Management
- AI-driven power prediction and efficiency tracking are creating new intersections between data center operations, grid pricing and intelligent energy use.
