Infinigence AI partners with Huahuan Electronics on domestic AI compute infrastructure

Infinigence AI partners with Huahuan Electronics on domestic AI compute infrastructure

Infinigence AI (无问芯穹), the Shanghai-based AI-native infrastructure provider, signed a strategic cooperation framework agreement on September 17 with Huahuan Electronics (华环电子), a Beijing-based optical-transmission and telecom equipment maker that has spent three decades serving China Mobile, China Telecom and China Unicom. Both companies trace their roots to Tsinghua University's electronic engineering department. Infinigence will contribute heterogeneous-compute scheduling and its Agentic Infra and Agentic MaaS software platform, which it says has served over 1,000 AI companies and research institutions; Huahuan will contribute optical networking, hardware engineering and field operations. The two plan to co-develop "smart computing center" solutions covering design, integration and ongoing operations, plus a "Token factory" model pairing domestic chips with domestic large models.

The deal sits inside China's push toward chip self-sufficiency and multi-vendor compute pooling. Infinigence's cross-domain training system for heterogeneous clusters is already deployed nationally across more than 37,000 petaflops spanning 16 chip types, according to the article — software that stitches together compute from multiple domestic silicon vendors rather than relying on a single GPU supply line. Pairing that orchestration layer with an established telecom-hardware operator like Huahuan gives Infinigence physical network build-out and field operations capacity it lacks internally. Per the AI Market Watch index, Infinigence has raised $323.3M to date (tracked within AMW's roughly 5,000-company coverage, not a census), indicating this is a funded, scaled player rather than an early-stage bet.

For builders, it signals rising demand for chip-agnostic orchestration software wherever GPU supply is constrained and hardware fleets are mixed. For investors tracking China's domestic-AI-stack thesis, the "Token factory" framing — treating inference output as a standardized product tied to a domestic chip supply chain — is a concrete data point on efforts to decouple AI infrastructure economics from Nvidia-centric supply.

#AIInfrastructure #China #HeterogeneousCompute #DomesticChips #InfinigenceAI #ComputeOrchestration