The rapid success of DeepSeek, a rising star in China’s artificial intelligence sector, has led to a significant shift in the strategies of the country’s top AI start-ups. As DeepSeek’s technology gains widespread adoption, particularly after the launch of its R1 model in late January, many of China’s leading generative AI companies are rethinking their business models in an effort to stay competitive. According to a Financial Times report, the company’s breakthrough has left its rivals, which had previously attracted significant investment during the AI boom, struggling to replicate its success.
Among the companies pivoting is Zhipu, once a frontrunner in China’s large language model (LLM) scene. Zhipu, which has been burning through cash while trying to build its enterprise sales business, is now looking toward an initial public offering (IPO) to support its growth. The company’s efforts to sell tailored AI solutions to local governments and businesses have led to significant financial losses, prompting a reassessment of its future direction. Despite these challenges, Zhipu is continuing to pursue multiple business lines, including consumer applications, while hoping for the cash boost an IPO could provide.
Other start-ups are making more radical changes to their business models. Beijing-based 01.ai, led by prominent venture capitalist and former Google China head Kai-Fu Lee, has ceased pre-training large language models in favor of selling tailored AI solutions based on DeepSeek’s models. This shift reflects how companies are increasingly focusing on applying AI models to specific industries rather than developing their own foundational models. Similarly, Baichuan, which initially focused on consumer-facing AI chatbots, has refocused its efforts on the healthcare sector, including AI tools to assist with medical diagnoses.
DeepSeek’s success has forced companies like Moonshot to reevaluate their marketing strategies. After its AI chatbot, Kimi, gained significant attention last year, Moonshot has struggled with frequent outages and a decline in popularity as competitors launched their own products. The start-up has slashed its marketing budget in favor of concentrating on model training, though its uncertain future raises concerns about its long-term sustainability.
These shifts in strategy highlight the impact of DeepSeek on China’s rapidly evolving AI industry. As DeepSeek continues to dominate with its breakthrough technology, many start-ups are considering whether to compete head-on with the company or adopt its open-source models to target niche markets, reducing the need for costly in-house model development. As Wang Tiezhen, an engineer at AI research hub Hugging Face, observed, “The Chinese LLM market is rapidly consolidating around a handful of leaders,” with DeepSeek at the forefront.
With competition intensifying, some companies, like Zhipu, face mounting pressure to find new ways to remain competitive, including exploring avenues like IPOs or refocusing their product offerings. As AI continues to reshape the technological landscape, these business model overhauls reflect the broader shifts taking place in China’s booming AI sector. The industry’s future will likely be shaped by whether these companies can successfully navigate the post-DeepSeek era, balancing innovation with financial sustainability.
