The way some companies rushed to joined mining and later fold up is same way people that rush AI will later stop because some people change means of income once they see bubbles and they don't last. If any mining company leaves, onew miners are going to join. If old ones leave, the blocks will only take an average of 14 days and if difficulty is adjusted, the rest will be doing the work. It will even open opportunity for more miners that want to mine Bitcoin.

It's even norma when miners leave during the bear market. Some unplug their miners during the bear market due to cost and price of Bitcoin and resume when the price is fair to cover the cost of production. The AI might not be a reason to leave a profitable business to start over another bubble you are not sure about.

Yes you are completely right, there is no ASIC to GPU conversion going on. Nobody is going to use a SHA256 ASIC for AI. I do not think that is what Core Scientific is doing (or able to do for that matter).

However I disagree that there is little "insurance" open to larger miners. Core Scientific for instance has signed 12year dollar based contracts with CoreWeave for their AI Data Center business. Core Scientific is basically acting as landlord and CoreWeave will be funding the equipment costs required directly (rather than Core Scientific) in return for usage inside of Core Scientific power area. In total these contracts are ~$6.7B USD of estimated total revenue for Core Scientific over the next 12years.

This may or may not be profitable in big picture of AI, but regardless Core Scientific is set to make on average hundreds of millions per year off these contracts. I think this mixed business model is mostly available to larger miners (MARA, Core Scientific, Hut8 tier) who have the skills, spare power capacity and good locations that AI tech giants want. Not sure a smaller miner is in a position to do such a thing.

AI definitely consumes more energy than Bitcoin mining is going right now even if the government and investors don't want to agree to that because of the potential profit margin they stand to make. However, it doesn't mean Bitcoin mining would die off completely because it's still a very much evolving currency of value that is still needed.
AI seems to tackle more real world problems in almost every sector and that's why there a shift into that sector despite the energy cost, but it's more of a profitable investment and less disturbed infrastructure by the government as we see with not as rigid regulatory policies against it.

Many of these mentioned company seems to be able to secure deal or leases before pivoting, not all mining company can do that which maybe the reason why some company decided to decommission their mining farm. The keyword is deal and lease, otherwise anybody else might be better off to keep using their ASIC miners and continue to mine. Frankly, if a mining company get offered tens of billions to create AI data center, hard to say the company wouldn't take it. A good opportunity to secure place in compute intensive world and to scale your company to be bigger.

I believe that there is significant difference between adapting the old mining operations. And building brand new AI equipment. While power and cooling solutions are available for miners. The former can't be easily adapted for AI workloads, giving miners some edge over the ASICs. They require significant investment in GPUs. Networking and other equipment. I think some mining companies will shift gears a bit and not give up mining Bitcoin. Another factor that should be taken into account is if the demand for AI will stay at a level that will make these significant long term investments viable. It's one thing to be profitable when the market is going up. But it's another when it goes down. Will be interesting to see who can do both businesses successfully.

The AI bubble has burst in China and it's possible that in some other countries where it is big like US might also occur. But I don't see it happening soon, the time isn't yet to come for that. Yet once it has come, we'll see that the bitcoin miners have decided well to continue the usual thing they've used to do and if they want to get into AI or make it as a data center, they're going to invest to another things for their facility to be more usable and ai-ready. I have already seen those individuals making their own little data centers in their rooms but for those with existing mining farm, they don't need to change a lot.

I doubt they would even have the money to buy that equipment and even if some mad investors were able to throw cash at them this will be harder to get than it was for ASIC manufacturers to secure an order with TSMC during the previous madness.


The bubble burst in China because:
- a lot of it was hype and gimmicks and not actual utility, no, really, nobody needs humanoid robots running around doing martial arts and stuff
- Unitree was full of vaporware and basically no orders, cause, well, as I said, nobody needed them
- Chinese companies have begun a price war before having a product, almost none of them had an actual plan for making money
- Chinese hardware companies that were supposed to mitigate the US sanctions have turned out to be, well, propaganda

Right now, the AI bubble burst that might or might now happen in the US is not affected by any of this, it has way different problems at the horizon.