How effective is local action?
Skip to content

While the initiatives in Hangzhou and Hefei are an extensive front of subnational action in China’s AI development, it is important not to overstate the scale and success of local activity. Consider the over-hyping of Tianjin’s $15 billion AI fund, highlighted in a 2020 report by the Institute for Defense Analyses. Upon deeper investigation, the unfavourable comparisons to US federal government expenditures on AI R&D were misleading because ‘the announced expenditure from the Tianjin government does not appear to be annualised, focused on R&D, come from the central government, or consist of an actual outlay of money.’

Sometimes local government targets for AI development are simply infeasible. For instance, Sichuan province’s target for the scale of its core AI industry is 1000, double that of Beijing’s target even though Beijing is home to more than 20 times the number of AI firms as Sichuan.

'Global and extra-regional linkages are also necessary to build competitive clusters in a world of globalised innovation'

Similar tendencies have informed other Chinese S&T policies. As one analysis of financing schemes to implement ‘Made in China 2025’ concludes, ‘local authorities clearly tend to overstate the size of collected funds in order to signal compliance with central government policies, and funds pledged are often much higher than those eventually deployed.’ Another danger is that overeager intervention by local governments distorts the market, as was the case with the overinvestment in the solar photovoltaics industry.

As the brief studies above of Hangzhou’s AI Town and Hefei’s China Speech Valley demonstrate, a better understanding of local industrial ecosystems is a first step to getting empirical evidence on the effect of local implementation of China’s AI strategy.


Jeffrey Ding

DPhil researcher and China lead at the Centre for the Governance of AI, University of Oxford’s Future of Humanity Institute