I disagree. By no means is there a scarcity of quantum algorithms. Consider for example this review on Quantum Machine Learning. Therein the term qBLAS is contained for Quantum Linear Algebra Subroutines. This term describes all the quantum algorithms that exist for basic linear algebra tasks. Together with the (in)famous Grover Algorithm that gives a quadratic speedup for unstructured search problems and other quantum algorithms one has a large toolbox of generic quantum speedups at one's disposal.
As linear algebra tasks and search are basically used everywhere, quantum algorithms can yield generic speedups in nearly any field and will therefore be extremely important in the future.
But in the present NISQ era, we are indeed in a place where one can say that quantum computers are a "solution looking for a problem". Why is this the case? State-of-the-art quantum computers have few and noisy qubits, which means that none of the algorithms outlined above can be faithfully implemented.
Today's quantum computers can already outperform classical computers in very artifical and useless tasks. But people still hope that we can use those devices to do useful stuff. One major candidate are hybrid approaches like the Variational Quantum Eigensolver that can possibly aid quantum chemistry -- but while many such methods are proposed, no evidence exists that they are actually better than classical methods yet.
So while quantum computers in general are not a solution looking for a problem, today's NISQ devices, in a way, are.