Computing power infrastructure cuts corporate financialization, study says

A new study of Chinese listed firms finds that National Supercomputing Centers reduce corporate financialization and push more capital back into fixed assets, R&D and other real-economy investment. The results suggest computing infrastructure can change how companies allocate capital, with effects that spill beyond host cities.

Why it matters: - Corporate financialization has long pulled capital away from productive investment and can weaken innovation, R&D and long-term competitiveness. - The study suggests computing power infrastructure can reverse part of that drift by making real investment more attractive than financial asset hoarding. - The findings also matter for policy because the benefits appear to extend beyond the cities that host supercomputing centers.

What happened: - Researchers from Central University of Finance and Economics in Beijing studied Chinese A-share listed companies from 2012 to 2023. - The paper, published in Financial Innovation in June 2026, uses the staggered rollout of 14 National Supercomputing Centers as a quasi-natural experiment. - The study finds that city-level computing power deployment reduces corporate financialization. - Firms in National Supercomputing Center host cities cut financialization, measured as financial assets divided by total assets, by an average of 1.1 percentage points. - That decline equals roughly 17.5% of the sample mean. - The paper is indexed under DOI 10.1186/s40854-026-00946-5.

The details: - The researchers used a staggered difference-in-differences model to estimate causality. - The first mechanism is data factor capitalization capability. - Computing power helps firms turn raw operational data into commercially valuable assets. - That strengthens core business returns and raises the opportunity cost of shifting money into financial markets. - The second mechanism is intelligent decision-making efficiency. - Big-data analytics and AI-powered systems reduce operational uncertainty, improve capital allocation and lower management costs. - The study finds the inhibitory effect is concentrated in speculative financial assets. - The coefficient for precautionary financial assets is statistically insignificant, which suggests normal liquidity management is not materially affected. - The reduction in financialization is larger in computing-intensive industries, and the between-group difference is statistically significant. - The effect is also larger among firms with low analyst coverage, although that between-group difference does not reach conventional significance levels. - After National Supercomputing Center establishment, fixed asset investment intensity rises by 0.8 percentage points. - R&D investment intensity rises by 0.5 percentage points. - Capital expenditure intensity rises by 0.7 percentage points. - Spatial analysis shows the effect spills into neighboring regions through technology diffusion, network connectivity and competitive demonstration.

Between the lines: - The findings suggest computing infrastructure can act like an industrial policy tool, not just a technology upgrade. - The stronger effects in computing-intensive firms imply that the same infrastructure does not help all companies equally. - The low-coverage result points to bigger gains where outside information is thinner and internal data capabilities matter more. - The spillover effect implies public investment in supercomputing can create regional benefits, not just single-city gains.

What's next: - The authors say future research should build continuous measures of city-level computing power access as data systems improve. - The paper also notes that the current treatment variable is binary, so it captures the presence of a supercomputing center but not the scale of computing supply. - Policymakers could use the findings to justify intercity computing networks and more targeted digital transformation support. - Managers may use the results to reassess whether financial assets are crowding out higher-return real investment.

The bottom line: - Computing power infrastructure appears to pull corporate capital out of financial assets and back into productive investment, with effects that extend beyond the cities that host the hardware.