Infinigence AI Closes Nearly 500 Million Yuan Funding Round to Build Infrastructure for the Agentic Era | Oasis Vitality

Infinigence AI, an Oasis Capital Pre-A portfolio company, announced today the completion of its nearly RMB 500 million Series A+ funding round.

This round was led by Zhuhai Technology Group and Fortera Capital (Yuan Chuang Future Fund), with participation from Huiyuan Capital, Shangqi Capital, and HLC. Existing investors including Hongtai APlus, Fortune Venture Capital, Shangshi Capital & Haitang Fund, Lenovo Capital and Incubator Group, Legend Capital, Shenwan Hongyuan, Xuhui Sci-Tech Investment, and Yuanzhi Future also continued to increase their stakes.

This financing lineup brings together national industrial capital and leading market-oriented funds. The dual endorsement of "state capital + market capital" affirms the company's technological innovation in addressing major national strategic needs while strongly encouraging Infinigence AI to continue solidifying its value creation in AI industrial deployment and infrastructure ecosystem building.

Over the past two and a half years since its founding, Infinigence AI has consistently focused on building high-performance AI infrastructure for cutting-edge AI technology and application deployment. Through software-hardware co-optimization and heterogeneous computing services, the company has developed the "Infini AI Cloud" and "Infini Edge Intelligence Solutions" for AI development and services, breaking through computing bottlenecks in AI industrial applications. It currently serves leading AI and agent companies including Baichuan, Moonshot AI, Lenovo Group, Liepin, Li Auto, Lovart, Sand.ai, Shengshu Technology, Soul, VAST, H3C, Zhipu AI, China Mobile, and ZTE Terminals, as well as AI research institutions such as Beijing Zhongguancun Academy, Shanghai Artificial Intelligence Laboratory, Shanghai Algorithm Innovation Research Institute, Zhejiang Lab, and BAAI — helping each of them unlock the full potential of every computing resource.

Produce Agent, Collaborate With Agent, Serve Agent

Into Agent, With Agent, For Agent

The proceeds from this round will be deployed across three main priorities: first, continuing to expand Infinigence AI's technical leadership in software-hardware synergy and heterogeneous computing; second, driving scaled commercial expansion of AI cloud products and AI edge solutions across industries; and third, increasing R&D investment in agent infrastructure to build a world-class agent service platform with supporting cloud and edge infrastructure, accelerating the scaled and democratized deployment of agents in both digital and physical worlds.

AI is currently evolving from "conversational tools" to "agents of action," with agent-based applications poised to become a new form of productive force in future society. Infinigence AI believes that infrastructure must always face the frontier of intelligent development: it serves as the production line for agent development and iteration, the ideal testing ground for agent deployment, and the foundation supporting scaled industrial application and continuous evolution of agents.

Dr. Lixue Xia, Co-founder and CEO of Infinigence AI, stated: "The paradigm shift toward Agentic AI represents not only a strategic opportunity but a mission entrusted to us by the era. Thanks to our years of accumulated technical expertise and engineering capabilities in AI systems, Infinigence AI has nimbly completed its transformation into agent-native infrastructure. The completion of this new funding round will inject even stronger momentum into our growth. Facing the agent era, Infinigence AI will continue to rely on software-hardware co-optimization as our technical foundation and building a new generation of learnable, evolvable Agentic Infra as our strategic core — producing agent, collaborating with agent, serving agent (Into Agent, With Agent, For Agent) — to achieve deeper breakthroughs in AI infrastructure system optimization and ecosystem building."

Agent Infrastructure × (AI Cloud + Edge Intelligence)

Cloud-Edge Concurrent Agentic Infra Development

Agent-based AI technology is gradually moving from the digital world into the physical world, influencing reality across various production and labor scenarios through cloud infrastructure, mobile devices, and embodied robots. As an internationally leading AI infrastructure company, guided by the philosophy of "produce agent, collaborate with agent, serve agent," Infinigence AI has further developed a complete "Agent Infrastructure × (AI Cloud + Edge Intelligence)" technical and product architecture to support breakthroughs in agent application performance and continuous evolution across digital and physical worlds.

In the cloud, Infinigence AI has accumulated deep technical expertise and infrastructure experience in AI-native cloud infrastructure — the Infini AI Cloud. It maintains technical leadership in heterogeneous computing management and scheduling, AI training, and inference optimization, providing standardized product capabilities including an agent service platform and AI service platform within Infini AI Cloud, building an end-to-end commercial service stack. Currently, it has completed management of over 25,000P of computing power nationwide, covering 53 core data centers across 26 cities, providing native toolchain support and abundant high-performance computing resources for agent and AI R&D and hosting, enabling cost reduction and efficiency gains across the full lifecycle of AI application development and deployment for over 100 AI companies.

At the edge, Infinigence AI has developed "edge model + edge software + edge IP" integrated edge intelligence solutions — Infini Edge Intelligence. It has launched Infini-Megrez, the world's first on-device full-modal understanding model, whose large language version achieves 21B-level intelligence with only 3B computational cost and 7B memory requirements, breaking through edge resource constraints. The Infini-Mizar edge inference acceleration engine achieves 3× latency reduction, 40% energy savings, and 40% memory usage reduction on mainstream hardware, maximizing edge computing potential. The self-developed edge inference LPU IP, Infini-Merak, doubles energy efficiency while further substantially reducing large model inference costs.

Recently, Infinigence AI also successively released two agent products: Infra Agents, a cloud infrastructure agent swarm driven by Agentic AI, and Kernel Mind, a universal edge inference acceleration optimization platform, along with supporting agent technologies including RLinf, a reinforcement learning framework supporting continuous agent evolution, and Cache to Cache, a communication framework enabling more efficient and lossless agent interaction. The goal is to make cloud and edge infrastructure themselves proving grounds for agent deployment, actively helping agents move from demos to scaled new productive forces, becoming foundational resources as natural as water and electricity that flow into thousands of industries and millions of households.