Abstract
This study analyzes the sustainability of digital infrastructure and equity gaps in the Peruvian university system amid widespread Artificial Intelligence (AI) adoption, evaluating the “Shadow AI” phenomenon and the causal relationship between technological investment and academic performance. A non-experimental, cross-sectional quantitative design was employed, using microdata from the National Household Survey (ENAHO) of the National Institute of Statistics and Informatics (INEI) and administrative records from the Ministry of Education (ESCALE). For inferential analysis, linear regression models (OLS) were applied. After detecting heteroscedasticity (Breusch-Pagan test, p < 0.05) and endogeneity, a Two-Stage Least Squares (2SLS) model was implemented using mining royalty allocation as an instrumental variable. Additionally, the Compound Annual Growth Rate (CAGR) was calculated to project investment scenarios through 2030. A critical disconnect—termed "Shadow AI"—is observed, where 95.6% of students access AI tools through private mobile networks, contrasting sharply with the obsolescence of institutional networks. The robust 2SLS model confirmed high causal elasticity (β~IV~ = 25.4), indicating that investment in physical infrastructure is a significant determinant of logical-mathematical performance, exceeding simple correlation estimates. Digital infrastructure constitutes a sine qua non for educational equity in the algorithmic era. Without sustained investment increases (required CAGR of 28.5%) and implementation of shared infrastructure strategies (regional clouds), AI adoption will amplify pre-existing structural inequalities between regions with and without extractive resources.