WiseLaw, a legal-technology startup based in the Hong Kong Special Administrative Region, has begun testing a cross-border computing model this year, shifting some processing for its digital employee, Lawrence, to Guangdong province to cut costs as artificial intelligence-driven demand for computing power grows.
The startup's cross-border setup is made possible by a regional initiative known as the "processing with supplied data" pilot, which allows users in Hong Kong to send data over dedicated network connections to servers in Guangdong for computing-intensive tasks that do not require ultra-low latency. Hong Kong, in turn, serves as an international gateway for Chinese mainland computing services going global.
Arthur Lei, WiseLaw's chief operating officer, told China Daily that token calls under the arrangement are approximately 20 percent cheaper than those offered by Hong Kong service providers.
ALSO READ: Sun: HKSAR in 'best period' for inno-tech development
The arrangement operates in a dedicated zone within the Shaoguan computing cluster, part of China's national integrated computing network serving the Guangdong-Hong Kong-Macao Greater Bay Area. The zone's servers are physically located in Shaoguan, a city in Guangdong, but remain technically isolated from its overall network.
Feng Lanxiao, chairman and CEO of China Unicom Global, said raw data remains strictly within the zone throughout processing and is neither transferred outside nor mixed with other data. "Once processing is completed, the results are returned directly to Hong Kong, with no retention on the mainland side," he added.
Beyond using mainland computing resources, Hong Kong can also help connect these services with overseas customers. According to Feng, Hong Kong, with its abundant submarine cables and land-based cross-border links, is regarded as one of the most connected cities in the Asia-Pacific region.
The city's legal, compliance and business environment — aligned with international standards — also enables it to provide computing-related business services, compliance support and cross-border settlement for overseas clients, he added.
The pilot comes as the shift from chatbots to AI agents fuels computing demand. While a conventional chatbot is capable of answering a question and stopping, an agent can divide a task into steps, use external tools, assess interim results and repeat until it reaches a satisfactory answer.
This loop of planning, action and self-correction can consume far more tokens than a chatbot.
Hong Kong's financial services industry and growing technology sector need computing power, but high land and electricity costs make data infrastructure costly to build and operate. Feng said computing costs can account for over 30 percent of revenue at some small and medium-sized technology firms.
Lao Yuting, deputy general manager of Huashao Data Valley, the Shaoguan computing cluster's first operational hyperscale data center, said inquiries from Hong Kong telecom operators, universities and research teams have risen significantly since 2025, as the mainland's computing network expands and its cost advantages become clearer.
The data center aims to complete the second phase of construction in the third quarter of 2027, said Lao, adding that its design has undergone several revisions to meet the changing demands of rapidly advancing large AI models.
While strengthening cooperation with the mainland, Hong Kong will continue expanding its own capacity, said the Hong Kong SAR's Innovation, Technology and Industry Bureau.
"Local computing infrastructure remains essential for supporting research, sensitive workloads, low-latency applications and the development of a vibrant AI ecosystem," the bureau added.
ALSO READ: HK eyes bigger role in financing APAC’s green, digital transition
Core systems for high-frequency trading, which require ultra-low latency, are expected to remain in Hong Kong data centers to maintain speed and stability, Feng said, while computing-intensive tasks that can tolerate millisecond-level latency could move to the Shaoguan zone, including certain AI inference workloads, large-scale analysis of historical data and public-opinion monitoring.
The Shaoguan model is still in its early stages. "What I worry about most is still compliance," Lei said. Looking beyond the pilot phase, he said he hopes formal laws and regulations will be introduced to provide a stable, predictable long-term policy framework, so that users in Hong Kong and overseas "can feel more confident".
Contact the writers at irisli@chinadailyhk.com
