The first miner has already run towards AI

[joker_josue]

https://www.corexhosting.io/

is my hosting company.

Interesting.
I would risk it, but nowadays, even for a hobby and experience, it would be "expensive". 

[FP91G]

8.75c / kWh

https://www.asicminervalue.com/ru/miners/bitmain/antmine-s21-xp-270th

The Bitmain Antminer S21 XP will never pay for itself with this kind of electricity price.
The situation is no better for other Bitcoin mining markets, with the halving approaching every month. After the halving, the entire 21st series will be unprofitable.

[dansus021]

—snip—

The hashrate won't decline rapidly because the centers for these projects will be built within 2-3 years. Companies are buying equipment to pay for itself before the halving. A sharp drop in mining hashrate isn't expected in the next two years unless the price drops to 40,000.
I've noticed that the newer the equipment, the longer its payback period. Or perhaps manufacturers are publishing prices out of tradition, as they will always offer good discounts to their major customers.

Yes I also hear the news, according to coindesk A massive structural shift is underway as major public Bitcoin mining companies are aggressively liquidating their Bitcoin reserves to fund transitions into High-Performance Computing (HPC) and Artificial Intelligence (AI) data centers. Public miners sold a record 32,000+ BTC in Q1 2026 alone, driven by unsustainable post-halving mining economics and skyrocketing Wall Street valuations for AI infrastructure.

But I actually curious, did this are making the mining world end, I also remember when Bitmain created Sophon as AI chip back couple years ago

[FP91G]

—snip—

The hashrate won't decline rapidly because the centers for these projects will be built within 2-3 years. Companies are buying equipment to pay for itself before the halving. A sharp drop in mining hashrate isn't expected in the next two years unless the price drops to 40,000.
I've noticed that the newer the equipment, the longer its payback period. Or perhaps manufacturers are publishing prices out of tradition, as they will always offer good discounts to their major customers.

Yes I also hear the news, according to coindesk A massive structural shift is underway as major public Bitcoin mining companies are aggressively liquidating their Bitcoin reserves to fund transitions into High-Performance Computing (HPC) and Artificial Intelligence (AI) data centers. Public miners sold a record 32,000+ BTC in Q1 2026 alone, driven by unsustainable post-halving mining economics and skyrocketing Wall Street valuations for AI infrastructure.

But I actually curious, did this are making the mining world end, I also remember when Bitmain created Sophon as AI chip back couple years ago

The only working solutions I've seen are from Huawei (the Ascend series), which is used by DeepSeek in China.
But I'm sure there are plenty of ways to make chips cheaper and more accessible in this area, unlike Nvidia's very expensive solutions. It is quite possible that Bitmain will make ASICs for AI.

[joniboini]

But I actually curious, did this are making the mining world end, I also remember when Bitmain created Sophon as AI chip back couple years ago

I find it unlikely. Even if the pie gets smaller there'll be someone who have access to cheap electricity and whatnot. Hard to believe they will leave soon when the mining reward hasn't been emptied yet. Not to mention bigger players are also in the market now. I doubt they'll let the opportunity slip if a company exit the market and makes mining profitable for them.

[FP91G]

MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction
link

While this isn't new research, I can predict that many AI-related companies will go bankrupt in the coming years, making expensive data centers unnecessary. Many clients prefer to run models locally for security reasons and don't need expensive data centers. Newer models will require more and more computing resources, which will become very expensive.

[joker_josue]

MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction
link

While this isn't new research, I can predict that many AI-related companies will go bankrupt in the coming years, making expensive data centers unnecessary. Many clients prefer to run models locally for security reasons and don't need expensive data centers. Newer models will require more and more computing resources, which will become very expensive.

This will be a new bubble, similar to what happened in the 2000s with the dot-com bubble.

The market neither has the capacity nor the need to support so many companies involved in AI. Even companies that previously had nothing to do with it are now dedicating a significant amount of their time to it.

I believe that only 2 or 3 large models will remain, and then some more specialized models in certain areas. The rest will collapse...

[FP91G]

MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction
link

While this isn't new research, I can predict that many AI-related companies will go bankrupt in the coming years, making expensive data centers unnecessary. Many clients prefer to run models locally for security reasons and don't need expensive data centers. Newer models will require more and more computing resources, which will become very expensive.

This will be a new bubble, similar to what happened in the 2000s with the dot-com bubble.

The market neither has the capacity nor the need to support so many companies involved in AI. Even companies that previously had nothing to do with it are now dedicating a significant amount of their time to it.

I believe that only 2 or 3 large models will remain, and then some more specialized models in certain areas. The rest will collapse...

Companies with government and military contracts will remain in the market.
It's possible that these areas will be separated, with a separate AI for medicine that will address only medical needs. Developing this technology requires significant investment, which only large global companies can afford.

[FP91G]

Bitdeer Cranks up Bitcoin Mining Power While Rivals Stampede Into AI
Bitdeer Technologies Group is dropping roughly 1.93 exahash per second (EH/s) of bitcoin mining capacity into a 28-megawatt co-mining deal with Soluna Holdings at a wind-powered data center in South Texas.
link
Bitdeer deploys 28 MW and 1.93 EH/s at Soluna’s Kati 1 beginning in September 2026.
Soluna keeps 83 MW Kati 1 focused on bitcoin mining while Kati 2 and Dorothy 3 chase artificial intelligence (AI) demand.
Bitdeer targets 350 MW of AI capacity by Q1 2028 while continuing to add BTC hashpower.

[Outhue]

MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction
link

While this isn't new research, I can predict that many AI-related companies will go bankrupt in the coming years, making expensive data centers unnecessary. Many clients prefer to run models locally for security reasons and don't need expensive data centers. Newer models will require more and more computing resources, which will become very expensive.

The few who can still get their feets back will have to go back to Bitcoin mining then, but I am thinking what will happen when we don't have alot of Bitcoin left to mine, new Bitcoin miners won't be created anymore and everything will start to change at that moment, any one of you guys thinking about that already?

Someone whom asked me to teach them about Bitcoin mining changed his mind and when I asked why he said there isn't much Bitcoin left to mine, he is already scared about that, what do you guys think even though this isn't happening anytime sooner.

[FP91G]

MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction
link

While this isn't new research, I can predict that many AI-related companies will go bankrupt in the coming years, making expensive data centers unnecessary. Many clients prefer to run models locally for security reasons and don't need expensive data centers. Newer models will require more and more computing resources, which will become very expensive.

The few who can still get their feets back will have to go back to Bitcoin mining then, but I am thinking what will happen when we don't have alot of Bitcoin left to mine, new Bitcoin miners won't be created anymore and everything will start to change at that moment, any one of you guys thinking about that already?

Someone whom asked me to teach them about Bitcoin mining changed his mind and when I asked why he said there isn't much Bitcoin left to mine, he is already scared about that, what do you guys think even though this isn't happening anytime sooner.

I can't influence this, so I don't want to rack my brains. Mining equipment today becomes obsolete within 2-3 years, so miners have a short planning horizon. As long as mining is profitable, miners will buy new equipment. But if the Bitcoin price doesn't rise, the network hashrate will fall.

[joniboini]

Someone whom asked me to teach them about Bitcoin mining changed his mind and when I asked why he said there isn't much Bitcoin left to mine, he is already scared about that, what do you guys think even though this isn't happening anytime sooner.

We've had many discussions about this in this forum too. I think you can search a bit and see how the community react around that scenario. That being said, the barrier to mining is quite high even now. If your friend has a lot of money to burn, I think he can get a few rewards in a year or two and see whether he recoup his cost or not. I dunno, I think it makes more sense not to mine Bitcoin because it's too expensive instead of worrying they won't get any new block rewards decades later.