China’s New AI Tools Ignite Stock Market Frenzy as Alibaba Soars.

Link copied

Alibaba stock jumps after it unveils latest DeepSeek rival.

stock :: 2025-03-06 :: source - bloomberg

By Charlotte Yang and Luz Ding

Image illustration by Markus Winkler / Pexels

(Bloomberg) -- A pair of new artificial intelligence tools from China sparked frenzied trading in the nation’s stock market, propelling an index of technology companies to a multi-year high.

Most Read from Bloomberg

Alibaba Group Holding Ltd. on Thursday unveiled its latest open-sourced the QwQ-32B model. The platform marked a big leap over the previous version using just a fraction of the data DeepSeek’s R1 employs. Also this week, Manus AI launched what it called a general AI agent, or a bot that can perform tasks, saying its model is performing better than OpenAI’s DeepResearch on at least some fronts.

Alibaba’s stock surged as much as 8.2% in Hong Kong, the most intraday in nearly two weeks, helping propel an index of Chinese tech shares about 5% higher, on track for the strongest close since 2021. In mainland China, Focus Technology Co., which has AI agent products, soared by the 10% daily limit. Related stocks including software maker Client Service International Inc. also jumped.

The gains follow DeepSeek’s AI breakthrough earlier this year, which ignited a bull run for Chinese stocks and sent shockwaves across global markets. The tech sector got another boost this week after China said at the National People’s Congress it would support the extensive application of large-scale AI models and development of new-generation intelligent terminals and manufacturing equipment.

Alibaba in particular has gone on a tear, gaining some $135 billion of market value this year. Investors have warmed to the company founded by Jack Ma as it stabilizes a business sideswiped by a years-long government crackdown. Its growing prowess in AI and mounting signs of Beijing’s support helped galvanize 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 recognize the value that AI will bring to their cloud computing business.”

In the wake of DeepSeek’s emergence, multiple companies have trotted out models and services they claim match the Chinese startup’s or OpenAI, whose ChatGPT is credited with igniting the generative AI boom. The litany of rollouts has captivated both investors and the Chinese public.

Even with the recent surge in tech shares, valuations are still not demanding. The Hang Seng Tech index is trading at around 19 times forward earnings, compared to 45 times four years ago.

“We see a further re-rating happening in the market given how cheap a lot of the China tech stocks look compared to their US peers. Showcasing the enthusiasm is that even some hardware tech names which haven’t really performed earlier are going up,” said Ken Wong, an Asian equity portfolio specialist at Eastspring Investments.

Alibaba’s latest model is intended to take on both DeepSeek’s R1 and OpenAI’s o1. The new reasoning model is considered efficient at 32 billion parameters, as demand grows for AI that requires minimal data and consumes less computational resources. Parameters are the parts of the AI model learned from the training data, and help adjust its behavior to make more accurate predictions or generate meaningful outputs.

Alibaba has pledged to invest more than 380 billion yuan ($52 billion) on AI infrastructure such as data centers over the next three years, a major commitment that broadcasts the e-commerce pioneer’s ambitions of becoming a leader in artificial intelligence.

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.

Meanwhile, Manus AI said its AI agent did better than OpenAI’s DeepResearch on some metrics using the GAIA benchmark, designed for assessing an AI model’s ability to handle real-world scenarios.

--With assistance from Saritha Rai, Debby Wu, Jing Jin and Shikhar Balwani.

(Adds a quote.)