ALIBABA Group Holding’s stock surged the most in weeks after taking the wraps off a model that it claims performs as well as DeepSeek with just a fraction of the data required.
China’s e-commerce leader, which in 2025 jumped to the forefront of the domestic artificial intelligence (AI) sector with highly benchmarked models, on Thursday (Mar 6) open-sourced the QwQ-32B model. The platform marked a big leap over the previous version using roughly 5 per cent of the parameters, or data units, that DeepSeek’s R1 employs.
Alibaba’s stock gained as much as 7.5 per cent in Hong Kong, the most intraday in nearly two weeks. The tech company has gone on a tear this year, gaining some US$135 billion of market value. Investors have warmed to the company founded by Jack Ma as it stabilises a business sideswiped by a years-long government crackdown. Its growing prowess in AI and mounting signs of Beijing’s support helped galvanise its comeback.
“There’s quite a few positive drivers for Alibaba with their open-source reasoning model the latest catalyst,” said Vey-Sern Ling, a managing director at Union Bancaire Privee. “Their core business is improving and clearly will benefit from China’s push to drive consumption. Investors now also recognise the value that AI will bring to their cloud computing business.”
Alibaba has pledged to invest more than 380 billion yuan (S$70 billion) on AI infrastructure such as data centres over the next three years, a major commitment that underscores the e-commerce pioneer’s ambitions of becoming a leader in AI.
That target marks one of China’s biggest AI infrastructure budgets, underscoring Alibaba’s growing ambitions in the field. But it comes at a time investors are pondering whether big tech firms are overestimating future demand for AI services, or the capital needed to create them. BLOOMBERG
Share with us your feedback on BT’s products and services
