AI inference deployment solutions company Qingmao Intelligence raises tens of millions of RMB in funding | Oasis Capital leads round

Qingmao Intelligence, a provider of AI inference deployment solutions, announced today that it has raised tens of millions of RMB in an angel round led exclusively by Oasis Capital. The funds will be used primarily for algorithm R&D, product development, and team expansion. This marks Qingmao's second investment round within just a few months.

An investment lead at Oasis Capital said: "The underlying inference deployment technology that the Qingmao team has built can significantly lower the barriers and costs for model deployment, enabling rapid development of downstream applications and defining the core value at the foundation of this new wave. In less than a year since its founding, the company has already secured demand validation from multiple customers and ecosystem partners. Foundational AI technology is a long-distance race that rewards deep accumulation. Oasis Capital will continue to track next-generation technological shifts and innovation opportunities in the market. We look forward to seeing the Qingmao team keep exploring, growing quickly, and continually pushing technical boundaries to accelerate the industry's development."

Chaoyu Guan, founder of Qingmao Intelligence, said: "Qingmao Intelligence is a foundational model inference deployment solutions provider, dedicated to building an optimization and inference engine that adapts to any hardware environment, enabling large models to run everywhere. Oasis Capital is a pure, long-term-value-oriented investor focused on corporate vitality. They have a deep understanding of artificial intelligence, especially foundational technology, and share our belief that AI development must start from the bottom up to achieve 'deep accumulation and explosive growth.' This aligns perfectly with Qingmao Intelligence's technical path and vision. AI infrastructure is a critical domain in building the large model ecosystem. Qingmao Intelligence will continue to deepen its commitment to technological innovation and long-term value creation, fully unleashing the productivity of large models."

Qingmao Intelligence's founding team hails from Tsinghua University's Department of Computer Science, with over a decade of experience in edge computing and automated model compression. Founder Chaoyu Guan is a recipient of Tsinghua's prestigious Graduate Student Special Award. During his graduate studies, he led the open-source release of AutoGL, the first automated graph learning platform, and captained teams to multiple top international AutoML competition victories, defeating strong international teams from UC Berkeley, Stanford, and MIT.

Guan explained that Qingmao Intelligence's mission is to reduce the cost of AI adoption and deployment, including for foundational models. By optimizing the inference and deployment pipeline, the company aims to lower the barrier to using AI models, with the ultimate goal of enabling AI to run on any device.

According to Guan, the industry currently employs multiple approaches to address the high resource consumption and deployment challenges of large models. From a software engineering perspective, cluster efficiency can be improved at the resource scheduling layer. From an algorithm optimization angle, model compression and related techniques can reduce model complexity. From a hardware design or selection standpoint, companies can choose machines that offer better cost-performance for target models, or design hardware and modules better suited to those models.

Currently, Qingmao Intelligence has launched MLGuider, an inference optimization toolchain for AI models that adapts to and optimizes for mainstream hardware environments. Beyond Nvidia, MLGuider also supports optimization for AMD, Qualcomm, and Ascend chips, giving customers alternatives beyond Nvidia. Overall, MLGuider aims to become an optimization and inference engine that adapts to any hardware environment, unlocking hardware productivity. Guan noted that in customer engagements, MLGuider has been validated to deliver performance on AMD's MI200 series that matches Nvidia's A100 for large language models.

Based on MLGuider, Qingmao Intelligence will also launch an LLMOps platform centered on foundational models, deeply embedding itself in every stage of large model development and deployment. The company has recently completed initial commercial exploration and refinement of its AI toolchain. Going forward, it will continue to refine its product capabilities through practice.

Source: 36Kr, Zhen Zi

Nourishing Vitality

What do you think vitality is?

For AI developers, vitality is the instinct to keep exploring, the bold practice of technological innovation, and the sustainable development of long-term value. For Qingmao, vitality is the cornerstone of long-term growth and the driving force behind value realization. That's why Qingmao Intelligence focuses on building foundational infrastructure to democratize large model deployment, working alongside AI developers everywhere to co-create a new era of large models.

— Chaoyu Guan, Founder and CEO of Qingmao Intelligence

Oasis Capital is a new-generation Chinese venture capital firm dedicated to discovering the most vital entrepreneurs of the next decade and growing alongside them to create long-term value. "Nourishing vitality" is Oasis Capital's vision and mission. This vitality — Vitality — represents both the direction of structural transformation in our era and the resilience and evolutionary power of entrepreneurs.

Oasis Capital focuses on early and growth-stage investments, writing checks from $3 million to $30 million, with emphasis on healthcare, biotechnology, enterprise services, and other technology-enabled service sectors, supporting China's technology-driven service upgrade.