HK start-ups call for talent and infrastructure to be included in first 5-year plan

The government is drawing up Hong Kong's first-ever five-year plan -- with innovation and technology among its key priorities.

Some local start-ups hope talent development and supporting infrastructure should be part of the blueprint.

This Hong Kong start-up Votee AI develops AI products focused on Cantonese services.

Among its products is a large language model developed in partnership with the Chinese University of Hong Kong to train nursing and social work students.

The company has a team of more than 30 people. Over 80 percent are local hires.

Its management says the company plans to expand into Africa and Southeast Asia over the next five years, but one of its biggest challenges is finding suitable technical talent at different stages of growth.

Its co-founder Ting Pak-sun says large language model training starts with data, so understanding Cantonese -- especially spoken Cantonese is important.

According to a 2023 government report, Hong Kong is expected to face a shortage of between 18,000 and 23,000 IT professionals by 2028.

Hong Kong Cyberport Chairman Simon Chan believes AI will influence the labour market.

He says the objectives outlined in China's 15th Five-Year Plan are very clear, particularly the promotion of "AI+" initiatives in sectors.

He added Hong Kong needs a clear roadmap for human capital development.

Meanwhile, this Hong Kong-founded company SenseTime has been focusing on generative AI in recent years, including multimodel large language models and an AI video creation platform for short dramas and animation.

It says additional infrastructure support, particularly electricity supply, will be essential.

Lewis Fung, Managing Director of SenseTime Hong Kong and Macau, Asia Pacific Business Group says Hong Kong's current computing capacity, is far from enough to support widespread AI adoption across industries.

He says when everyone starts using AI, the bottleneck will ultimately be computing power, and the bottleneck of computing power is electricity.