Ask HN: What general problem would benefit from ASIC speedup?

What would be in your opinion a computing problem for what it would make an economical sense (that is, would be there a market for it) to produce a specific solution in silicon (ASIC) i.e. an accelerator?

By that I mean a specific computing problem and not a general one like floating point operations or physics simulations etc.

For example there are ASIC solutions for bitcoin mining that can be about 10^6 faster than single CPU core calculations.

There are also special accelerators for cryptographic operations in today CPUs etc.

Would it make sense for example if we had SAT solvers in silicon?

(Related question: What could be the most popular applications for an integrated FPGA? I know that Microsoft is flirting with this idea in data centers and Intel bought a FPGA producer, so it is not an ungrounded question in my opinion).