Lattice Hamiltonians allow one to find non-perturbative solutions of quantum field theories without incurring the usual problems associated with Monte Carlo sampling. At present, the main limitation is memory usage, due to the exponential scaling of the Hilbert space size with volume. In this talk, I present a novel approach that has the potential to overcome this scaling problem using Physics-Informed Neural Networks (PINNs). I discuss how PINNs can solve the eigenvalue equation for the Kogut–Susskind Hamiltonian, learning how the eigenfunctions and eigenvalues evolve along the renormalization flow. By exploiting the adiabatic theorem of quantum mechanics, we can smoothly move away from the analytically known strong-coupling region toward the continuum limit.