Generally yes. I would expect most bitcoin mining operations would continue for a while after they stopped making a return on investment, until they were no longer making a profit on the cost of electricity. However, electricity is a fairly large part of the cost of mining, and there is a treadmill on the hardware side as well as better ASICs are developed (the ASIC developers skim the cream of this as well, units sold to customers have likely already been used, sometimes heavily).

> mining operations would continue for a while after they stopped making a return on investment

Marginal return. The ASICs could have been bought for $10 with the expectation of making $1 a year and paying out 50¢ for electricity (a 5% yield). Now it’s making 51¢ a year (a 0.1% yield). Not enough to make economic sense, but enough to keep running the rig; 0.1% is more than zero.

Broadly, I was pushing back against the author’s inchoate claim that “the continued growth in hash power demonstrates a strong, continued belief in Bitcoin by miners worldwide.”

Profits aren't even required if you're laundering money. You might pay a 30-50% fee for any laundering scheme, so if you can only lose 15% of your money by running in the red rigs you're still accomplishing your goals

That advance investment in infrastructure must be a very important part of this effect. Those sunk costs probably are most important thing here. ASIC lead time is months and not the only advance investment.

But betting on future price, speculation, also is important.

Also there maybe could be influence from changes in energy pricing.

Yep, major ASIC sellers require up front payments in some cryptocurrency, last time I checked.

And the delayed effect of R&D into better ASIC design has an even longer delay than purchasing or manufacturing of existing designs.

An article about an ASIC startup that made HN mentioned an 12-24-month lag, IIRC, between design and mass production at scale.

So we're not even past the point of seeing the effect of the peak bitcoin value.