Guizhou to fund large AI model applications with one-time awards

Southwest China's Guizhou province will offer one-time awards of up to 5 million yuan ($702,870) for qualifying large AI model applications, provincial big data bureau director Zhu Zongyao said at a Tuesday news briefing.

The awards are part of a big-data development fund intended to support industry growth. Zhu outlined this year's funding priorities and eligibility criteria, which include product and technical standards innovation, expansion of application scenarios, provision of high-quality data, and broader access to computing resources.

"For companies whose data workforce reaches a specified scale, tiered awards ranging from 1 million to 10 million yuan will be offered," Zhu said.

According to this year's funding document, each large AI model product filed with national authorities will receive 500,000 yuan. Eligible large-model applications and industry-enabling platforms may be considered for one-time awards of up to 5 million yuan.

Zhu added that Guizhou's digital industry reached 254.94 billion yuan in 2024, and that the province's digital-economy growth rate has ranked among the nation's leaders for nine consecutive years.

The big data bureau and the provincial finance department jointly manage the provincial big-data development fund. It aims to promote high-quality industry development, with support measures adjusted regularly to enterprise needs.

Zhu said data labeling will be a development focus for Guizhou's data industry, leveraging the province's abundant, cost-competitive labor force and more than 53,000 annual graduates in big-data-related majors.

Guiyang and Gui'an New Area will provide employment subsidies for data annotation companies. The new area now offers one year of rent-free housing for new graduates.

"Guizhou will continue to leverage its advantages in data, labor force, and policy to build the data-labeling industry into a new engine of the digital economy," he said, adding that the province aims for roughly 20,000 data-labeling workers by 2026 and about 50,000 by 2028.