DeepSeek's launch of its latest open-source models has heralded China's own "ChatGPT moment" by igniting a frenzy around artificial intelligence (AI) adoption, while raising questions about the viability of developing pre-trained in-house models, according to Lee Kai-fu, founder and CEO of Chinese start-up 01.AI.
In an interview with the South China Morning Post on Friday, Lee, former head of Google China, said 01.AI was leveraging DeepSeek's popularity to sell AI solutions to corporate clients, particularly in the finance, video gaming and legal sectors.
"It becomes imperative for us to ... embrace [DeepSeek] as our primary bet," Lee said. That "became as clear as the writing on the wall" after the surge in demand for the company's models from Chinese CEOs after late January, he said.
Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.
Established in mid-2023, 01.AI is one of China's pioneering unicorns specialising in large language models (LLMs). Its transition from a developer of LLMs to a solutions provider began last year but has accelerated thanks to what Lee called the "godsend" opportunity presented by DeepSeek. Many Chinese hardware and software vendors have since been aligning their services with DeepSeek's models.
The Chinese version of the DeepSeek app seen on a smartphone. Photo: Xinhua alt=The Chinese version of the DeepSeek app seen on a smartphone. Photo: Xinhua>
Lee said 01.AI's experience in LLM development set it apart, as only a dozen Chinese companies possessed similar capabilities. 01.AI has prioritised DeepSeek, a move that distinguishes Lee's venture from competitors that favour in-house models.
While many hardware manufacturers can readily offer on-site plug-and-play solutions with comprehensive business functions, 01.AI's advantage lies in its ability to fine-tune DeepSeek models for clients, according to Lee.
"How do you train it, how do you tune it, how do you do reinforcement learning and how do you do fast inference? That last part can only be done by companies that have LLM capabilities," he said.
While 01.AI, which employs about 200 employees, has begun producing its own servers, Lee said it was not venturing into the market for "all-in-one" boxes, which combine AI hardware and software. The technology is championed by Chinese tech companies like Lenovo Group, Huawei Technologies and Inspur Group.