On August 20, Hygon Information Technology Co.,Ltd. held its interim results briefing for the 2026 fiscal year, with management fielding investor questions on key topics including DCU market penetration, compute ecosystem development, margin management, and its positioning in the AI computing segment. The company reiterated that its dual-pronged "CPU+DCU" strategy, combined with the full-chain compute ecosystem built through the Hygon Industrial Ecosystem Cooperation Organization (the "Guanghe Organization"), continues to reinforce its technological moat, expand its customer base, and secure a healthy order pipeline. At the same time, Hygon outlined plans to capitalize on the AI Agent-driven compute transformation through product iteration, supply chain optimization, and ecosystem upgrades.
Domestic Compute Power Sees New Window of Opportunity
"Our product development follows a 'sell one generation, validate the next, and research the one after' approach," said Sha Chaoqun, Director and General Manager of Hygon, in response to a question about the company's roadmap. "Through sustained R&D, we are rolling out CPU and DCU products with higher performance, lower power consumption, and more open ecosystems to meet the diverse demands of the industry."
As AI Agents emerge as the core direction of artificial intelligence evolution, the structure of compute demand is undergoing a seismic shift, creating fresh growth opportunities for domestic chipmakers. Zhang Ruiping, Independent Director of Hygon, explained to investors during the briefing that AI Agents are fundamentally reshaping how compute resources are allocated. The traditional CPU-to-GPU ratio of roughly 1:8 is moving significantly closer to 1:1 or even lower, as the CPU's central role in request handling, task orchestration, and concurrent scheduling is being revalued upward. This trend, Zhang noted, opens up an unexpected incremental growth path for Hygon's CPU offerings.
On the DCU front, Hygon has completed adaptation of its products to more than 400 mainstream large language models, covering 99% of globally available non-closed-source models. The company has also achieved Day-0 adaptation with multiple model vendors, including Tencent's Hunyuan Hy3. Through its "CPU+DCU" synergy and the HSL bus interconnect protocol, Hygon has built both Scale-Up and Scale-Out architectures, providing a full-stack compute foundation for AI Agents in high-concurrency inference and private deployment scenarios across internet, financial services, biomedicine, and intelligent manufacturing.
Ecosystem Expansion Moves to Deep Collaboration
Ecosystem development remains the bedrock of Hygon's strategy to broaden market reach and increase penetration. Xu Wenchao, Vice President and Board Secretary of Hygon, noted that through the Guanghe Organization platform, the company has built a comprehensive, full-chain compute ecosystem with steadily expanding scale and adaptation capabilities. Looking ahead, the next phase of ecosystem building will shift from initial scale aggregation and basic compatibility to deep collaborative innovation and large-scale scenario deployment. Hygon plans to further its "chip + ecosystem" strategy by upgrading its partner enablement programs, pushing upstream and downstream players beyond simple adaptation toward joint technical development, co-built solutions, and on-the-ground implementation. The company will also zero in on core tracks such as AI compute and industry digitalization, leveraging national ecosystem adaptation centers and joint laboratories to refine standardized solutions, build flagship case studies in key sectors, and complete the domestic compute industry loop.
Addressing investor concerns about gross margin sustainability, Sha Chaoqun explained that Hygon's blended gross margin is influenced by a combination of product mix, order types, upstream procurement costs, and competitive dynamics. The compute industry undergoes rapid technological iteration and intensifying competition, so margins will naturally fluctuate with product cycles and market conditions. To manage this, the company is pursuing two fronts: launching next-generation products with higher compute density and better energy efficiency, while also implementing more disciplined supply chain management and leveraging economies of scale to contain costs. Management reiterated its commitment to prudent operational controls to sustain a stable and healthy profitability profile.
Looking forward, Hygon's leadership stated that the company will remain laser-focused on its core compute business, sharpening product competitiveness and refining its market expansion framework. By steadily advancing the integration of technological innovation with commercial deployment, the company aims to unlock sustained industry growth dividends and ensure stable, long-term performance for shareholders.
Comments