Chinese AI startup DeepSeek is considering outside funding for the first time, attracting interest from Alibaba and Chinese state funds. The Information reported this news on Wednesday, citing sources familiar with the internal discussions. DeepSeek has gained recognition for its low-cost AI models that rival or outperform Western counterparts, challenging the perceived U.S. dominance in AI technology. Executives at DeepSeek and its parent hedge fund, High-Flyer Capital Management, are discussing a potential shift towards generating revenue and profit from their research. DeepSeek’s emergence has spurred an AI race among China’s internet giants, with companies like Alibaba, Tencent, Kuaishou, Baidu, and ByteDance increasing their AI investments.

DeepSeek’s R1 model has disrupted the AI industry due to its cost-efficiency, performance, and open-source approach. It was developed at a fraction of the cost of Western models like ChatGPT or Google’s Gemini, with approximately $6 million spent on computing power. Despite lower costs, DeepSeek R1 has demonstrated impressive capabilities, rivaling larger competitors in key benchmarks6. The company offers lower costs per token than OpenAI’s models, making it an attractive solution for developers and enterprises. DeepSeek has released its R1 model in a relatively open-source manner, allowing researchers and developers to access and modify the code. The model employs a Mixture of Experts (MoE) architecture, activating only 37 billion of its 671 billion parameters per forward pass, ensuring scalability without proportional increases in computational cost.

Alibaba’s latest large language model (LLM), Qwen 2.5-Max, released in January 2025, utilizes a MoE architecture similar to DeepSeek and has been trained on over 20 trillion tokens. Qwen 2.5-Max has demonstrated strong performance across multiple benchmarks, outperforming models like DeepSeek-V3, GPT-4o, and Llama-3.1-405B in various tests6. The model is available in different sizes, ranging from 3 billion to 72 billion parameters, and includes base and instruction-tuned versions. Qwen 2.5-Max excels in language understanding, coding, mathematics, and reasoning, and features multimodal capabilities, allowing it to process text and visual inputs. Alibaba has made the model accessible through APIs on its generative AI development platform, Model Studio, enabling developers worldwide to leverage its capabilities.

However, an Alibaba executive has denied reports that the company plans to invest in DeepSeek. Yan Qiao, a vice president at Alibaba, stated on her WeChat moments feed that “news circulating that Alibaba will invest in DeepSeek is fake news”. Earlier reports had suggested that Alibaba planned to invest $1 billion into DeepSeek. DeepSeek’s efficient releases have challenged the narrative that China was years behind America in AI development. The company’s success has prompted a reevaluation of AI development strategies globally and highlighted the potential for innovation under constraints. DeepSeek’s technological tricks are poised to significantly reduce the cost of building AI.