About Infinigence AI

About Infinigence AI

Infinigence AI is a leading AI Infrastructure Service Provider, dedicated to accelerating the intelligent transformation of industries through next-generation Agentic Infrastructure.

Leveraging capabilities in Software-Hardware Co-Optimization, we have developed a scalable "M × N" architecture that enables interoperability between diverse AI models and heterogeneous hardware, improving performance and cost efficiency across cloud and edge deployments. Our industry-leading Agentic Infra Platform and Agentic MaaS support training, inference, and reinforcement learning for ultra-large-scale AI workloads. Processing hundreds of billions of tokens per day, our platforms serve top enterprises and research institutions, empowering the full lifecycle of model development and deployment.

Driven by the mission to accelerate the future of AGI, we are committed to achieving a 10,000x reduction in AI costs through technical innovation, extending the benefits of intelligence from the digital world to the physical world and making it as accessible as water and electricity to every industry and household.

Milestone

2026

2025

2024

2023

2026

Jul.

Unveiled the strategic layout of "Front Store, Back Factory & One Hub" Agentic Infrastructure, laying the infrastructure for all future AI productivity operations in both digital and physical realms.

Jan.

Infinigence AI has joined hands with Alibaba Cloud and China Mobile to jointly establish the Jing'an District "AI+ Immersive Audio-Visual" Public Service Platform, supporting the high-quality advancement of the industry.

2025

Dec.

The Infinigence AI agent service platform was officially released, providing enterprises with end-to-end support from agent customization and optimization to deployment, hosting, and commercialization.

Sep.

Infinigence AI launched its agent infrastructure swarm powered by Agentic Infra, supporting the industrial deployment of next-generation AI agents.

Jul.

Infinigence AI released its AI full-scale performance leap solution along with three core products, exploring the limits of intelligent efficiency across multiple computing scales.

Mar.

Infinigence AI partnered with Beijing Jingneng digital industry to officially launch the Beijing Public AI Compute Platform.

Mar.

Infinigence AI partnered with Zhejiang Compute Technology to officially launch the Hangzhou Compute Resource Scheduling Service Platform .

Mar.

A custom video generation model IP jointly developed by Infinigence AI, the Department of Electronic Engineering at Tsinghua University, and Shanghai Jiao Tong University was recognized as Best Paper at FPGA 2025.

Feb.

SMC and Infinigence AI launched China's first "compute ecosystem supermarket"—the SMC Compute Ecosystem Platform.

Feb.

Infinigence AI entered into a strategic partnership with Lenovo to jointly advance an integrated intelligent terminal solution combining edge models, edge software, and edge IP, enabling next-generation AI PCs and related devices.

2024

Dec.

Megrez-3B-Omni, he world’s first fully multimodal understanding model designed for edge deployment, was open-sourced, enabling high-speed on-device inference for image, audio, and text workloads.

Jul.

The world’s first heterogeneous hybrid training platform supporting single-task training at the 1,000-GPU scale was launched, with cluster utilization reaching up to 97.6 percent.

Jan.

FlightLLM, the first customized inference IP for large language models, was released, achieving efficient large-model inference on a single FPGA for the first time.

2023

Oct.

The Infini-Megrez large model was released, extending the maximum context length to 256k and setting a new benchmark.

Oct.

FlashDecoding++ was officially released, accelerating GPU inference by two to four times.

May

With support from the Department of Electronic Engineering, Tsinghua University, Infinigence AI was officially established.

Infinite Computing, Accelerating the Future of AGI

Focus on Goals

Every action we take should be guided by the company's long-term objectives. This requires clarity of responsibility, sustained focus, and alignment between individual work and overall strategy. By uniting individual efforts, we create a collective force that drives the company forward.
Pursuit of Excellence

We encourage innovation and continuous improvement in everything we do, from product development and customer service to internal processes. Excellence is not merely an outcome, but a mindset defined by self‑transcendence and the pursuit of higher standards.
Open and Honest

We value open, transparent, and respectful communication within the organization. Through candid dialogue, we build trust, strengthen collaboration, and enable teams to work more effectively together.
Accountability

We expect every team member to take responsibility for their work and for collective success. When challenges arise, we confront them directly, take ownership, and work together to find solutions and move forward.

Leadership

Infinigence AI brings together leading academic research achievements and core industry resources. The company is supported by a strong core team with deep expertise in both software and hardware R&D, as well as proven large-scale engineering execution capabilities. The team combines rigorous academic foundations in artificial intelligence with extensive real-world industry experience.
Core team members come from top domestic and international universities, leading technology companies, and research institutions. over 63% hold master’s or doctoral degrees, and over 60% Graduates from QS Top100 Team members have previously held key roles at organizations such as Alibaba Cloud, Baidu Research, SenseTime, Huawei, Tencent, ByteDance, and Microsoft, and have contributed to widely adopted open-source projects including Apache, TensorFlow, ONNX, PyTorch, and PyG.