Chinese VCs Whiffed on DeepSeek. Do They Still Matter?
The January release of DeepSeek R1, a Chinese-developed open-source large language model, was a watershed moment for China’s artificial intelligence industry. In the weeks since, enterprises and government bureaus across China have raced to deploy the LLM, which can now be accessed everywhere, from the ubiquitous messaging app WeChat to local government websites.
Less talked about has been DeepSeek’s significance in challenging the conventional wisdom around tech start-ups and incubation — and in particular, the role of venture capital. Part of what makes DeepSeek unique is that it’s a self-funded venture by Liang Wenfeng, a hedge fund manager, which took off without any venture capital investment.
Ever since the Silicon Valley boom, it’s become standard to conflate venture capital with entrepreneurship and innovation. This is as true in China as it is in the United States: Venture capital features prominently in the origin stories of Chinese tech giants Alibaba, Tencent, and ByteDance — and all three have since created their own VC arms to stay current with emerging business from food delivery to robotics. The notion that venture capital is necessary for entrepreneurship and innovation is so ingrained that a controversial 2024 article in The Financial Times used declining venture capital funding as a proxy to conclude that entrepreneurship had died in China.
There are problems with conflating venture capital with innovation, however. In the more than a decade I’ve spent in the industry, I’ve watched time and again as VCs followed a herd mentality off the cliff. When peer-to-peer finance lending was hot, everyone wanted to invest in P2P firms. When online-to-offline was trending, everyone pivoted to O2O. When software as a service became popular, companies all rebranded as SaaS. And when AI gained traction, everyone claimed to be doing AI.
If investing is merely about following trends, what’s the point of having investment professionals at all? In hindsight, you could reasonably attribute the past successes of China’s venture capital industry to broader economic trends rather than savvy picks. In other words, we were lucky, enjoying the fruits of an era of across-the-board growth rather than making smart bets on competitive companies.
In addition, traditional VC models work best for projects with clear-cut commercialization paths or clear initial public offering potential. (The other main offramp for VCs, mergers and acquisitions, is still rare in China). Neither is applicable to DeepSeek. The road to artificial general intelligence is full of uncertainty, and even AI giants like OpenAI have no clear pathway to profitability in the short term. Nor does DeepSeek necessarily need outside funding: Liang’s quant fund is among the biggest in China.
The irony is that, at least in the case of DeepSeek, venture capital and innovation can even be mutually opposed: Had DeepSeek received VC funding, it might not have been as successful as it is today.
DeepSeek has a singular mission: to develop AGI in China. However, this mission is in fundamental tension with traditional venture capital investment models, where the primary goal is financial returns within a specific time window. If DeepSeek had taken VC funding, its mission would likely have been compromised. The pressure to deliver returns would have pushed the company toward user growth and revenue metrics, potentially derailing its core objectives.
DeepSeek’s choice to open-source its technology is also unusual. Within China’s VC scene, the idea of investing in open-source projects is treated with skepticism. Yet, it’s precisely this open-source approach that has established DeepSeek’s credibility and influence in the field, especially abroad. The DeepSeek team has already been hit with numerous unfair criticisms that they used sanctioned GPUs, that they “stole” from OpenAI, or that the whole company was simply a Chinese Trojan Horse — and that’s with anyone being able to check their work. Had Liang chosen a closed-source approach, the project might have been thrown into the dumpster by an uninformed public.
To be clear, I am not suggesting that VCs make no contribution to China’s tech innovation. In a country where the investment culture remains immature, venture capital has served as a critical source of funding for start-ups. What we should do away with is the perception of VCs as omnipotent and the only viable model for innovation, which risks directing financial resources to low-hanging fruit rather than more challenging long-term projects.
I hope DeepSeek’s breakthrough can offer Chinese venture capitalists a much-needed wake-up call. The industry is at a critical juncture. The flood of American money that once drove venture capital in China has dried up, meaning it is increasingly crucial for VCs to carve out a niche for themselves — diving deep and mastering their subject area.
Those who put in this “DeepWork” will gradually find themselves facing less competition and greater potential returns. A friend of mine is a VC investor in consumer markets. For the last few years, he has stayed focused on consumer investments — even as they fell out of favor in VC circles. After organizing industry tours and building a successful podcast program, he’s now one of the biggest names in an often-overlooked sector.
That may not sound as cool as an AI product, but it represents a return to the true spirit of venture capital: Rather than follow the trends, find a niche and build a comparative advantage within it. I am hopeful that this dedication to true innovation and entrepreneurship, rather than hype, will bear fruit, not just for Chinese venture capital, but for the country as a whole.
Editor: Cai Yineng; portrait artist: Wang Zhenhao.
(Header image: Visuals from VCG and Brandlogos, reedited by Sixth Tone)
