# Multiple regression - covariance matrix, law of large numbers

I'm currently reading an article: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4302277/#S6title

And in Section 6 there is a least squre estimator of $$\beta$$ given of course as $$\hat \beta = (X_c'X_c)^{-1}X_c'y$$ with covariance $$Cov(\hat \beta)=\sigma^2 \mathbb E[(X_c'X_c)^{-1}]$$, but then there is a line:

From the law of large numbers, for large n:

$$Cov(\hat \beta)\approx \frac{\sigma^2}{n-1}\Psi_x^{-1}$$

And I have no idea how the law of large numbers is applicated here - I don't even know if we use strong or weak version.