*
Tiger Brokers integrates DeepSeek-R1 model into its AI
chatbot
*
Chinese brokers, funds rush to capitalize on DeepSeek
breakthrough
*
AI to reshape China's financial industry, boosts IT
spending,
UBS says
By Samuel Shen and Tom Westbrook
SHANGHAI/SINGAPORE Feb 18 (Reuters) - Tiger Brokers said
on Tuesday it embedded DeepSeek's model into its AI-powered
chatbot, as brokerages and money managers race to capitalise on
the Chinese start-up's artificial intelligence breakthrough, and
develop use cases for the financial industry.
The integration of the DeepSeek-R1 model into TigerGPT
follows DeepSeek's meteoric rise which stunned Silicon Valley
and triggered a rally in Chinese tech shares on bets of an AI
revolution sweeping across sectors from education to finance.
Tiger Brokers TIGR.O , an online broker backed by
shareholders including Chinese electronics maker Xiaomi and
renowned U.S. investor Jim Rogers, is the latest to embrace
DeepSeek.
At least 20 Chinese brokers and fund managers, including
Sinolink Securities, CICC Wealth Management and China Universal
Asset Management, have already started to integrate DeepSeek
models into their businesses, potentially changing the way they
conduct research, manage risks, make investment decisions and
interact with clients.
DeepSeek will be able to tap Tiger Brokers' financial data
and help customers analyse valuations, make trading decisions
and "feel the beauty of investment," said Wu Tianhua, Tiger
Brokers' founder and CEO.
"Its impact is real. It's no longer a concept, or a
marketing trick," he told Reuters in an interview.
UBS expects rapid AI adoption will boost financial IT
spending by 24%, or 69 billion yuan ($9.49 billion), in five
years, benefiting vendors including Hundsun Technologies Inc
600570.SS , Northking Information Technology Co 002987.SZ and
iSoftStone Information Technology (Group) Co 301236.SZ .
"We expect the launch of DeepSeek R1 to drive faster GenAI
adoption in the financial industry in 2025," UBS analyst Haifeng
Cao said, referring to DeepSeek's recently released model that
was developed at a fraction of the cost of Western rivals.
"We think the industry is likely to be more profoundly
reshaped by GenAI than others, given its data-heavy and high
labour intensive characteristics, and high mix of
language-related tasks."
An index tracking China's Fintech companies .CSI930986 has
jumped 17% this month, flirting with record highs.
INSPIRING
Tiger Brokers said the upgraded version of its investment
assistant TigerGPT will initially be available for free to users
in mainland China and Singapore.
"The integration of DeepSeek has enhanced TigerGPT's logical
reasoning abilities, enabling it to analyse market shifts more
clearly and interpret investment opportunities more
effectively," CEO Wu said.
The chain of thoughts generated by the DeepSeek model "is
often inspiring to even most seasoned traders."
Sinolink Securities 600109.SS , which sees AI as a key
engine of growth, said early this month it would use the
DeepSeek model in a range of scenarios including information
search, market analysis and industry research. In the future,
the application will be expanded to core business areas such as
risks management and investing.
CICC Wealth Management said its has integrated DeepSeek's R1
model into its investment advisory services, potentially
boosting information processing efficiency by 90%.
Zhongou Fund Management said the immediate challenge for the
financial industry is to standardize internal database to feed
AI models.
"AI has given financial companies a tool to make better use
of troves of data" which was previously sitting idle, said
Zhongou's head of tech research Du Houliang.
"Now, many financial institutions, including us, are
hurrying into meetings" to discuss how local adoption of
DeepSeek's models can empower internal management, marketing,
and investing.
($1 = 7.2730 Chinese yuan renminbi)
(Reporting by Samuel Shen in Shanghai and Tom Westbrook in
Singapore
Editing by Shri Navaratnam)
((samuel.shen@thomsonreuters.com; +86 21 20830018; Reuters
Messaging: samuel.shen.thomsonreuters.com@reuters.net))