Qiming Top News | The "Qiming Venture Partners - Forum on Entrepreneurship and Investment: Venture Capital Initiates the Resonance Cycle of AI Technology and Applications" at WAIC 2025 was successfully held.
On July 28, the "Qiming Venture Partners · Entrepreneurship and Investment Forum - Venture Capital Initiates the Resonance Cycle of AI Technology and Applications" at the 2025 World Artificial Intelligence Conference (WAIC), hosted by Qiming Venture Partners, was successfully held in the Blue Hall of the Shanghai World Expo Center. Renowned experts and scholars, top investors, and leading entrepreneurs gathered to share and exchange ideas on topics such as technological breakthroughs, cutting - edge trends, and application implementation in the AI field.
From AI 1.0 to AI 2.0, Qiming Venture Partners has invested in over 100 AI projects in total. The invested enterprises cover the entire AI industry chain, helping many industry benchmark enterprises to rise. It is the most active and influential investment institution in the field of artificial intelligence in China and even Asia.
Kuang Ziping, Founding Managing Partner of Qiming Venture Partners
Kuang Ziping, Founding Managing Partner of Qiming Venture Partners, delivered a welcome speech. He said that as one of the earliest and most comprehensively - deployed investment institutions in the AI field in China, this is the third consecutive year that Qiming Venture Partners has hosted this forum. Qiming Venture Partners hopes to connect the themes of innovation, entrepreneurship, and venture capital through hosting distinctive sub - forums. Through the sincere sharing of the invited heavyweight guests, it can bring some good and useful information to the industry and even the entire artificial intelligence industry.
Zhou Zhifeng, Managing Partner of Qiming Venture Partners
Zhou Zhifeng, Managing Partner of Qiming Venture Partners, introduced in his opening speech titled "Technology Grows Upward, Applications Take Root Downward - Release of the Resonance Cycle of AI and the Top 10 AI Outlook for 2025" that this year, AI has entered the "resonance cycle of technology and applications". On the one hand, technology continues to grow rapidly upward without an obvious ceiling. On the other hand, technology has become "usable" in terms of performance and cost, and large - scale applications have begun to be implemented, just like "tree roots" taking deep root and growing steadily, creating huge value.
Hu Qi, Executive Director of Qiming Venture Partners
Subsequently, Zhou Zhifeng and Hu Qi, Executive Director of Qiming Venture Partners, released the Top 10 AI Outlook of Qiming Venture Partners for the third consecutive year, focusing on foundational models, multimodal models, AI Agents, AI infrastructure, AI applications, and embodied intelligence.
The content of the Top 10 AI Outlook of Qiming Venture Partners for 2025 includes:
Foundational Models
Outlook 1: In the next 12 - 24 months, a context window of 2 million Tokens will become the standard for top - tier AI models. More refined and intelligent context engineering around larger context windows will become one of the core driving forces for the development of AI models and applications.
Multimodal Models
Outlook 2: A general video model is expected to emerge within 12 - 24 months. This model can handle generation, reasoning, and task understanding in the video modality, promoting innovation in video content generation and interaction.
AI Agent
Outlook 3: In the next 12 - 24 months, the form of Agents will shift from "tool assistance" to "task undertaking". The first batch of real - sense "AI employees" will enter enterprises and widely participate in core processes such as customer service, sales, operations, and R & D. They will no longer only exist as assistants but will have the abilities of collaborative work, active feedback, and taking on OKRs, promoting the transformation from cost tools to value creation.
Outlook 4: Multimodal Agents will continue to move towards practicality. They can integrate multi - source inputs such as vision, voice, and sensors, perform complex reasoning, tool invocation, and task execution, and achieve breakthroughs first in industries such as healthcare, finance, and law.
AI Infrastructure
Outlook 5: In the field of AI chips, more GPUs that are "domestically designed" and "domestically manufactured" will start mass delivery. At the same time, a new generation of AI cloud - based chips that innovate in 3D DRAM stacking and integrated computing and communication will also stand out in the market.
Outlook 6: In the next 12 - 24 months, the Token consumption will increase by 1 to 2 orders of magnitude. Cluster reasoning optimization, terminal reasoning optimization, and software - hardware collaborative reasoning optimization will become the core technologies for reducing Token costs on the AI Infra side.
AI Applications
Outlook 7: The shift of the AI interaction paradigm will accelerate in the next two years. As users' dependence on mobile phone screens weakens and the importance of natural interaction methods such as voice rises, it will drive the birth of AI - native super - applications.
Outlook 8: The potential of AI applications in vertical scenarios is huge. More and more startups will deeply cultivate niche areas based on industry knowledge and quickly achieve product - market fit, competing differently from large enterprises with a "Go Narrow and Deep" strategy.
Outlook 9: The AI BPO (Business Process Outsourcing) model will achieve commercial breakthroughs in the next 12 - 24 months, shifting from "delivery tools" to "delivery results". Through the "pay - by - result" method, it will rapidly expand in industries with standardized processes such as finance, customer service, marketing, and e - commerce.
Embodied Intelligence
Outlook 10: Embodied intelligent robots will be deployed on a large scale first in scenarios such as picking, handling, and assembly, accumulating a large amount of first - person view data and tactile operation data of robots, and building a closed - loop flywheel of "model - body - scenario data". This flywheel will drive the iteration of model capabilities and ultimately promote the large - scale implementation of general robots.
Wang Yu, Tenured Professor and Dean of the Department of Electronic Engineering at Tsinghua University, Initiator of Wuwen Xinqiong
When algorithm innovation meets hardware breakthroughs, how will the collaboration between software and hardware promote the paradigm shift of AI technology? Wang Yu, Tenured Professor and Dean of the Department of Electronic Engineering at Tsinghua University, Initiator of Wuwen Xinqiong, pointed out in his speech titled "Software - Hardware Collaboration Promotes the Evolution of AI Infrastructure" that in the process of truly transforming artificial intelligence into actual productivity, Token, as the basic unit of input and output of large models, has become one of the most core production factors in the intelligent era. In the past, the value chain was driven by electrical energy to improve computing power to complete simple tasks. Now, it has evolved into converting electrical energy into computing power, then generating Tokens with computing power, and finally supporting the execution of complex tasks. Along with this change, the core indicators for evaluating the efficiency of infrastructure are also changing - the traditional "number of computations per joule of energy (TOPS/J)" is being replaced by "the number of effective Tokens that can be processed per joule of energy (Token/J)". How to optimize the Token efficiency per unit of energy consumption will be the core proposition for infrastructure and system design in the AI 2.0 era.
Kuang Ziping, Founding Managing Partner of Qiming Venture Partners (left) and Yin Qi, Chairman of Qianli Technology (right)
In the wave of large models, the integration of AI and terminals is accelerating the reshaping of the industrial landscape. Smart cars, industrial terminals, etc. are all rejuvenated with the empowerment of AI. Yin Qi, Chairman of Qianli Technology, and Kuang Ziping, Founding Managing Partner of Qiming Venture Partners, had a special dialogue on the topic of "The Evolution of 'AI + Terminals': Large Models Empower Terminal Evolution and Industrial Reconstruction".
Yin Qi said that there will be two core trends in the field of AI terminals in the future. First, hardware, operating systems, and services will be deeply integrated. Hardware will become increasingly carrier - like, and its form is not important. The essence lies in how to define and deliver an end - to - end AI Agent service. Second, the AI operating system will undergo a fundamental change within the next year. For example, current mobile phones are terminals completely controlled by humans. In the future, they will evolve into "human - machine collaboration", and machines will perform many complex tasks in the background. Due to the changes in AI Agent services and the operating system level, many different hardware forms will emerge in the field of AI terminals and then gradually converge. There are great opportunities in this field for both giants and startups.
Liu Wei, Former Distinguished Scientist at Tencent, Technical Leader of Hunyuan Large Model, CEO of Video Rebirth
As digital technology continues to break boundaries, video generation has evolved from a content tool to an important element in building virtual worlds. Liu Wei, Former Distinguished Scientist at Tencent, Technical Leader of Hunyuan Large Model, CEO of Video Rebirth, said in his speech titled "From Video Generation to World Models" that the video generation model is the best path to build world models, and this technological direction is expected to be the key breakthrough for AI to leap from perception to cognition. Video Rebirth defines the video - native world model as a combination of a world simulator and a world predictor. Although the current mainstream DiT architecture has powerful spatio - temporal simulation capabilities, it has key problems such as lack of causal reasoning and inability to interact and intervene. In the future, the company is committed to solving the above problems through technological upgrade paths, building its own unique technological concepts and model paradigms, and ushering in the "ChatGPT Moment" in the field of video generation, launching the first real video - native world model. Liu Wei emphasized that AI not only needs grand narratives but also needs to create realistic reality. The technological path of Video Rebirth from video generation to world models is expected to achieve important technological innovations during the critical period of AI cognitive ability breakthroughs and provide important support for the development of general artificial intelligence.
Zhang Huaiting, Founder and CEO of Yu Ai Wei Wu
In the current era of rapid development of AI technology, how to truly implement technology in application scenarios and find a sustainable development path in the entrepreneurial wave is a topic jointly explored by many practitioners. Zhang Huaiting, Founder and CEO of Yu Ai Wei Wu, said in his speech titled "Thoughts and Practices on AI Application Entrepreneurship" that the entrepreneurial opportunities in AI applications lie in using generative AI technology to transform the service industry into a manufacturing - like industry, breaking the impossible triangle of large - scale (personalized) - high - quality - low - cost. The core reason why we haven't seen the explosive commercial implementation of AI applications yet is the hallucinations of large models, inaccurate reasoning, and uncertain results. This requires teams engaged in AI applications to understand both business and AI technology, balance the uncertainty of models and the fault - tolerance of business, first run through the business closed - loop, use business to drive the gradual implementation of AI capabilities, and at the same time find a data flywheel suitable for their own business scenarios. In the intelligent era, the density of cross - border talents and a pragmatic and innovative corporate culture are the keys to organizational construction, and the human - machine collaboration work paradigm is the foundation of enterprise operation.
Hu Xubo, Managing Partner of Qiming Venture Partners (left) and Ren Feng, Co - CEO and Chief Science Officer of Insilico Medicine (right)
When artificial intelligence meets biomedicine, a revolution in drug R & D is quietly taking place. Ren Feng, Co - CEO and Chief Science Officer of Insilico Medicine, and Hu Xubo, Managing Partner of Qiming Venture Partners, had a special dialogue on the topic of "AI - Driven Next - Generation Drug R & D: Precise Target Mining and Clinical Value Creation".
Dr. Ren Feng pointed out that traditional drug R & D mainly relies on human knowledge and experience, which has certain limitations. AI, however, can break through the ceiling of human cognition, summarize and analyze massive data through algorithms, and bring many unimaginable breakthroughs in fields such as target discovery and molecule generation. He believes that as AI deeply penetrates the entire process of drug R & D, AI - powered drug discovery is moving from the 2.0 stage to the 3.0 stage. The emergence of large models enables us to build a super - intelligent agent for biomedicine, allowing AI not only to assist in molecule design and generation but also to participate in decision - making. When it comes to future development directions, Dr. Ren Feng said that for AI - powered drug discovery companies, they not only need to have independent core technologies but also need to deeply understand the actual scenarios of drug R & D, promoting the real implementation and commercial transformation of technology.