They have been getting closer to ignition in the last four years. But its still a physics problem. Meaning they havn’t figured out the physics yet. Meaning that its still 20 years off.
However, if the whole process of experimentation could be accelerated byo machine learning as is done with materials research and dna research—then the time to net positive energy output will be shrunk.
Of course, we don’t know if the machine learning actually works or whether it solves for the right things.
So we’ll see.
I'm sure it'll prove to be a big help in the continued development of virtual fusion reactors. :^)