无问芯穹与华环电子签署战略合作协议，共同探索国产异构算力AI基础设施的新方向。

On September 17, Shanghai Wu Wen Xin Qiong Intelligent Technology Co., Ltd. (“Wu Wen Xin Qiong”) and Beijing Huahuan Electronic Co., Ltd. (“Huahuan Electronics”) signed a strategic cooperation framework agreement. Leveraging their respective industrial and technological strengths, the two companies will deepen collaboration around AI infrastructure innovation and industrial applications, jointly exploring new models for the coordinated development of computing power and networking. This signing marks a new phase in their synergistic efforts within the AI infrastructure sector, driving the deep integration of computing, networking, and AI-native infrastructure to provide more efficient, stable, and open foundational support for the large-scale application of artificial intelligence.

The agreement was signed by Guo Meng, Party General Branch Secretary and Chairman of Huahuan Electronics, and Zeng Shulin, General Manager of Wu Wen Xin Qiong’s Shanghai branch. The signing was witnessed by Zhen Jia, Party Secretary and Chairman of Tsinghua Holdings Co., Ltd.; Wang Yu, Tenured Professor at the Department of Electronic Engineering, Tsinghua University and IEEE Fellow; and Li Boxun, Chief Technology Officer of Wu Wen Xin Qiong.

As large language models and intelligent agent applications accelerate, AI infrastructure is shifting from a single focus on computing resources toward systemic coordination across computing, networking, software platforms, and model services. Huahuan Electronics, a national-level key “Little Giant” specialized and sophisticated enterprise, shares deep historical ties with Tsinghua University’s Department of Electronic Engineering and its broader technology ecosystem. For over 30 years, Huahuan has focused on information communications and optical networks, with its optical transmission equipment long serving China Mobile, China Telecom, and China Unicom, and widely deployed in dedicated communication networks for sectors such as power and transportation. Similarly originating from Tsinghua’s Department of Electronic Engineering, Wu Wen Xin Qiong is a leading provider of AI-native infrastructure. Relying on core technologies including multi-source heterogeneous computing, software-hardware co-design, and autonomous AI, it has built the Agentic Infra autonomous infrastructure platform and the Agentic MaaS large model service platform, having already served over 1,000 AI enterprises and research institutions.

Under the agreement, Wu Wen Xin Qiong will leverage its advantages in heterogeneous computing management and AI software platforms, while Huahuan Electronics will contribute its capabilities in network communications, hardware R&D, and engineering implementation and operations. Together, they will explore collaborative infrastructure solutions tailored for domestic heterogeneous computing, establishing end-to-end complementary capabilities spanning computing scheduling to network transmission, and from software platforms to hardware engineering.

Regarding smart computing center solutions, the two parties plan to jointly advance scheme design, software-hardware adaptation, integrated delivery, and operational maintenance. Wu Wen Xin Qiong has developed a cross-domain training system for large models tailored to heterogeneous clusters. By employing pipeline orchestration that overlaps computation and communication, adaptive communication flow scheduling, and heterogeneous computing scheduling mechanisms, the system systematically resolves challenges in computing power aggregation. It has already been deployed nationwide, delivering over 37,000P of computing capacity and supporting 16 mainstream chip architectures. Huahuan Electronics will provide high-speed interconnection and stable transmission support in optical transmission and network access. The companies will also establish a normalized cooperation mechanism to continuously conduct technical validation and technology transfer.

The partners plan to further explore a new infrastructure supply model standardized around tokens as output, driven by AI application demand, with a focus on optimizing the closed loop of “domestic large models + domestic chips.” Wu Wen Xin Qiong’s Agentic MaaS platform is dedicated to the large-scale, high-quality production of tokens, having driven a tenfold reduction in inference costs over the past year. Huahuan Electronics’ capabilities in hardware R&D, network communications, and engineering operations will provide the engineering support needed to scale the “Token Factory” from construction to daily operation.

The partnership between Huahuan Electronics and Wu Wen Xin Qiong represents both a continuation of the longstanding “Tsinghua family” relationship into a new era and a fresh launch toward emerging opportunities in AI infrastructure. Both companies aim to follow the pathway of “domestic chips → heterogeneous computing infrastructure → token services → enterprise AI applications,” creating a replicable and scalable model for building and operating smart computing centers. This effort will accelerate the transformation of domestic software and hardware capabilities into practical AI productivity for real-world business applications.

This signing marks a new stage in their collaborative efforts within AI infrastructure. The deepened cooperation between Wu Wen Xin Qiong and Huahuan Electronics is expected to provide a pragmatic path of “software-hardware synergy and computing-network integration” for the large-scale deployment of domestic heterogeneous computing infrastructure. In the future, both parties will drive the integrated development of computing networks and AI-native infrastructure, ensuring that domestic heterogeneous computing power is truly converted into AI productivity accessible across all industries, thereby offering more efficient, stable, and open foundational support for the large-scale application of artificial intelligence.

This article was provided by Wu Wen Xin Qiong and reprinted with authorization by Quantum Bit. All views belong to the original authors. All rights reserved. Unauthorized reproduction or use in any form is strictly prohibited.