Cloud / data center GPU consumption has also gone up in the past couple years.
If you mention anything about the cloud, deep learning training on GPUs and the impact on the environment that has and the increasing costs of training and fine tuning your neural net, they won't care at all, they will still do it after all these years. As you know it is still a valid point.
There is a recent trend to discuss environmental impacts of training deep learning models in papers. See e.g. Latent Diffusion Models (Heidelberg), RETRO (Deepmind).
Further, I would posit that machine learning has a much clearer value proposition than cryptocurrencies which are a highly speculative.
Finally, your argument is an example of https://en.wikipedia.org/wiki/Whataboutism. Rather than arguing against facts presented you are changing the subject to say "look here though! this happens here and you're all hypocrites for not noticing it!" This makes your argument seem defensive, biased and unconvincing.