By Jiaxing Li

HONG KONG, Feb 21 (Reuters) - Chinese tech stocks listed in Hong Kong advanced on Friday, heading toward their best weekly winning streak since 2020, powered by earnings surprises and continued investor interest in the sector.

The Hang Seng Tech Index .HSTECH climbed 4.7% to a new three-year high, bringing the week's gain to 4.2%. That marks the sixth straight week of advances, the best winning streak since May 2020.

Hong Kong's benchmark Hang Seng Index .HSI advanced 2.9% to a three-year high.

Alibaba 9988.HK surged 12.7% to the highest level since late 2021, after the Chinese e-commerce giant reported better-than expected revenue and said it plans to invest more in e-commerce and AI.

Lenovo 0992.HK advanced over 12% to near a ten-year high after quarterly results smash estimates, while Xiaomi 1810.HK added 4.8% to fresh record high.

The gains add to a huge rally in Chinese tech stocks sparked by the launch of DeepSeek, a low-cost Chinese artificial intelligence model, which reignited global investors' interest in China and spurred funds to rotate into the sector with its relatively cheap valuations.

"DeepSeek has been a powerful catalyst for China and there's more to go with the valuation discount to emerging markets to narrow more," HSBC analysts said in a note, also citing the return of foreign investment inflows during the last two weeks.

Onshore stocks also edged up, with the bluechip CSI 300 Index .CSI000300 adding 1.2% and the Shanghai Composite Index .SSEC climbing 0.8%. Both benchmarks are at the highest level since late December.

The DeepSeek-triggered rally has helped add more than $1.3 trillion to China's onshore and offshore equity markets in just a month, driving investment flows away from India and into China, said James Cook, investment director of emerging markets at Federated Hermes.

"Chinese companies, including Tencent, Alibaba and Baidu are fast catching up with the global AI frenzy after missing out in the past few years," he said.

(Reporting by Jiaxing Li in Hong Kong; Editing by Sonali Paul)

((jiaxing.li@tr.com))