Mao Chengyu, Yunqi Capital: DeepSeek Accelerates AI Application Adoption — Where to Invest in AI in 2025? | Partners' Perspectives

DeepSeek's breakout has pressed the accelerator on AI adoption. As we enter a new year marked by this pivotal shift, how should an early-stage tech investor adapt and position accordingly?

Recently, Mao Chengyu, founding partner of Yunqi Capital, sat down with China Entrepreneur Magazine to share his views on DeepSeek and Yunqi's 2025 AI investment strategy. Below is the full interview.

Key takeaways from this conversation:

1

DeepSeek's mainstream breakthrough has ignited public enthusiasm for AI, accelerating the speed and intensity of AI app adoption by at least one to two orders of magnitude.

2

China's technological breakthroughs and potential in original innovation could translate into positive capital market reactions, such as renewed appetite for Chinese assets in Hong Kong. The vibrancy of Hong Kong's capital markets is critical to the development of China's venture capital industry. Several of Yunqi's early portfolio companies, including Manycore Tech and Zhaogang.com, are currently preparing for Hong Kong IPOs.

3

In 2025, more opportunities will emerge in AI applications — both B2B and B2C — with roughly 70-80% of capital to be deployed in AI-related sectors.

4

Enduring strategies and methodology: forward-looking perspective on technological change, international vision, and deeper, broader roots in industry.

The following is excerpted from China Entrepreneur Magazine, by Kong Yuexin. Click "Read More" at the end for the full article.

How Has DeepSeek Changed Perceptions?

Mao Chengyu: During the Spring Festival holiday, our investment team was constantly discussing DeepSeek in our group chat — from technical angles, from user perspectives. At our first weekly meeting after the break, we spent most of the time analyzing DeepSeek's impact on the industry. Everyone contributed from different dimensions: its technical characteristics, the development of B2C and B2B applications, and upstream hardware related to computing power and data centers. As for its industry impact, I believe it will accelerate application adoption. At the mass awareness level, it has ignited enthusiasm for using AI. Nowadays, if the leader of a major company isn't pushing their organization to try DeepSeek, they risk being seen as "behind the times." I believe the speed and enthusiasm with which people are embracing AI applications has increased by at least one to two orders of magnitude.

Additionally, when people discover and recognize China's breakthroughs and potential in original technological innovation, there will be emotional responses — some of which may translate into positive capital market reactions. This is especially true as international capital markets develop new perspectives on Chinese assets, which will also drive more active sentiment toward Chinese assets in Hong Kong's capital markets. I hope Hong Kong's capital markets will be more active this year than last, particularly on the IPO front. The vibrancy of Hong Kong's capital markets is crucial to the overall development of the venture capital industry. This year, several of our early portfolio companies are preparing for Hong Kong listings, including Manycore Tech and Zhaogang.com.

What Sectors Are You Most Bullish On for 2025?

Mao Chengyu: I think 2025 will offer more opportunities in AI applications. AI was still fairly abstract and vague for most people. Before, aside from the VC circle and industry practitioners, only stock traders paid attention to AI concept stocks. DeepSeek suddenly made everyone aware — every individual might give it a try.

I believe opportunities will definitely emerge in both B2B and B2C, though it's hard to say how large they'll be.

What Industries or Sectors Will You Focus On in 2025?

Mao Chengyu: This year's investment focus will definitely remain AI-related. I expect 70-80% of our capital will go into AI-related sectors. We have a rough framework in mind and will deploy across several directions, including upstream computing-related technologies, AI for Science, and embodied intelligence.

Reflecting on 2024 Performance and the VC Industry

Mao Chengyu: We focused on two important areas in 2024. First, deepening our global perspective. Our entire core team, myself included, went abroad for various exchanges and learning opportunities. As more companies expand overseas, we need to continue broadening our global vision, sensing environmental changes, and better helping our portfolio companies with international expansion. Second, and more importantly, continuing to build around AI. Starting from the AI 1.0 era in 2015-2016, we've maintained a strategy of forward-looking deployment and deep research, making comprehensive investments across the entire AI upstream and downstream ecosystem. Over the past nearly decade, at the infrastructure layer, we've invested in PingCAP and MiniMax; at the application layer, we've positioned in Manycore Tech and XTransfer; and in embodied intelligence, we've backed Astribot and Noematrix. In 2024, we invested in nearly ten new projects around the AI ecosystem and actively sought out various industry opportunities brought by AI.

What Investment Strategies or Methodologies Will You Continue to Uphold?

Mao Chengyu: We primarily do early-stage tech investing, so we need to be forward-looking about technological change. First, our frontline team can actually read and understand technology — when chatting with founders, we can be on the same wavelength as technical entrepreneurs. Second, we need to maintain an international perspective; after all, the US remains an important source of technological development. Third, we must root ourselves deeper and broader in industry. Much technology commercialization must land in specific industry scenarios, so evaluating technological applications requires an industry perspective and understanding.

What's Your Flag for 2025?

Mao Chengyu: On the work front, at our recent year-end gathering we said: if Yunqi is a ship, it has been sailing for ten years. The current environment is relatively more challenging, with bigger storms and waves. I hope we can pull together to keep this ship steadier and sailing farther in the right direction. Personally, first, I want to strengthen my own learning and use of AI. During the holiday, I was constantly using DeepSeek, Doubao, and MiniMax's Hailuo. I hope to learn and experience new technologies more, keeping myself young and vibrant in mindset. Also, I'm still persisting with basketball, even as I get older. I hope to keep playing, guaranteeing at least two hours of full-court basketball every week. This is a habit I formed around 2010 — I've kept it up for 15 years. Click "Read More" at the end for the full article.