One important fact about the eigenvalues of the stiffness-matrix may be the following: every system which reaches an instability-point (e.g. a buckling point) has at least one zero-eigenvalue.
Eigen values of a stiffness matrix tell you about the stability of a system and also the amount of stretching in each (eigen) direction corresponding to that eigen value.
Any square matrix of order n can be 'decomposed' into n parts using the system of spectral decomposition. This is what explains the physical significance of characteristic roots of a matrix. Now for your matrix, you can have explanations accordingly.
If your stiffness matrix of size n comes from the discretisation of a partial differential equation, then the eigenvalues of this matrix approximate first n eigenvalues of the continuous operator. These are the eigenfrequencies (free vibration modes) of the continuum system.
Very simple manner we may think if we apply a dynamic force or displacement to system of a frequency same as Eigen frequency of a system then it shows maximum response.
The eigenvalues of element stiffness matrices K and the eigenvalues of the generalized problem Kx = λMx, where M is the element's mass matrix, are of fundamental importance in finite element analysis. For instance, they may indicate the presence of ‘zero energy modes’, or control the critical timestep applicable in temporal integration of dynamic problems