The shift of mining power to AI: potential risks for the crypto market in 2026+

[Smartprofit]

Here’s a simple question: in your opinion, how much does it actually cost to mine 1 Bitcoin, including all direct and indirect expenses?
This isn’t a complaint; it’s a simple practical question: how do you mine if you don’t believe these figures?

The way I know humans the use their household Electricity bill and divide it by the coins produced.
You could set up a farm, use Eton's Starlink and use wind power, solar panels to provide for electric. Depending on the miner what generation those miner costs.

Antminer         S19 Pro       110 Th/s SHA-256   $2,860   
AvalonMiner A1166 Pro   81TH/s   SHA-256   $1,550   
AvalonMiner 1246           90Th/s   SHA-256   $3,890   
WhatsMiner M30S++   112TH/s±5%   SHA-256
WhatsMiner M32-62T  62TH/s +/- 5   SHA-256   $1,100   

What would you go for?

Similar to the noise coming out of the EU for the strait of Hormuz blockade, about 25% of the oil consumed comes thru there.
25% reduction should not create an implosion of the EU.
 

I would be talking about the total cost of Bitcoin mining. And that includes:
- the cost of electricity consumed
- facility rental
- staff salaries
- taxes
- the cost of purchasing mining equipment
-...

The cost of electricity alone is just a PART of the actual costs

I believe halvings have the biggest impact on mining costs. Each halving essentially doubles the cost of mining 1 BTC. This puts enormous pressure on miners. That's why it's crucial that the Bitcoin price rises by 2028 (ideally to $300,000). Otherwise, most miners will simply go bankrupt! 🙋

Electricity plays a significant role in the cost structure of Bitcoin mining. Currently, it's the most scarce resource. It's no coincidence that mining is banned or restricted in some regions. Obviously, if the choice were between having electricity in a kindergarten and mining Bitcoin, the choice would be to have electricity in the kindergarten. It's precisely these circumstances that I attribute such restrictions and bans to...

It's also important that miners periodically upgrade to newer, more powerful ASIC chips.  There are also costs for cooling the equipment... In my opinion, wages make up an insignificant part of the cost structure (this is manufacturing, after all, and not the service sector).🤷

[DrBeer]

....

I believe halvings have the biggest impact on mining costs. Each halving essentially doubles the cost of mining 1 BTC. This puts enormous pressure on miners. That's why it's crucial that the Bitcoin price rises by 2028 (ideally to $300,000). Otherwise, most miners will simply go bankrupt! 🙋

Electricity plays a significant role in the cost structure of Bitcoin mining. Currently, it's the most scarce resource. It's no coincidence that mining is banned or restricted in some regions. Obviously, if the choice were between having electricity in a kindergarten and mining Bitcoin, the choice would be to have electricity in the kindergarten. It's precisely these circumstances that I attribute such restrictions and bans to...

It's also important that miners periodically upgrade to newer, more powerful ASIC chips.  There are also costs for cooling the equipment... In my opinion, wages make up an insignificant part of the cost structure (this is manufacturing, after all, and not the service sector).🤷

I’ll start from the end I agree that, compared to other expenses, wages are unlikely to exceed the costs of electricity and equipment, but all expenses must be taken into account... even the cost of maintaining the employee parking lot
As for mining costs, yes, that’s correct-mining costs are rising, for both objective and subjective reasons, and if the price of the product is "below cost," then the economic viability of such a business disappears. And yes-this will most likely lead to bankruptcy. Although, logically speaking, the decline in mining will, in the long run, lead to a reduction in network difficulty, and thus a reduction in mining costs.

[WillyAp]

I believe halvings have the biggest impact on mining costs. Each halving essentially doubles the cost of mining 1 BTC. This puts enormous pressure on miners. That's why it's crucial that the Bitcoin price rises by 2028 (ideally to $300,000). Otherwise, most miners will simply go bankrupt! 🙋

They are best placed to change to AI farming:
https://www.forbes.com/sites/greatspeculations/2026/02/02/why-bitcoin-miners-are-winning-the-ai-data-center-arms-race/

Hopefully they leave some as miners for the network.

[jaberwock]

Well... the thing is, GPU power was something that the crypto world really wanted, and desired, and even those bitcoin mining specific items were all made in companies that would focus on that. But these days, AI power is more important, the more powerful machine you have, the better AI servers you could fill it with, and many companies are spending insane amount of money to do that.

Nvidia literally has hard time selling stuff to regular people like us, could they sell more by producing more? hey could, but they are literally selling to companies, so we are no longer their customers.

[DrBeer]

Well... the thing is, GPU power was something that the crypto world really wanted, and desired, and even those bitcoin mining specific items were all made in companies that would focus on that. But these days, AI power is more important, the more powerful machine you have, the better AI servers you could fill it with, and many companies are spending insane amount of money to do that.

Nvidia literally has hard time selling stuff to regular people like us, could they sell more by producing more? hey could, but they are literally selling to companies, so we are no longer their customers.

It’s not all bad here. NVIDIA hasn’t shut down its consumer devices division, though it’s worth noting that priorities have shifted-solutions for AI and data centers are now the top priority. But this makes sense-it’s now more profitable for the company to establish itself in the AI market, earn more revenue, and develop new technologies in order to maintain or gain new market positions. However, the production of “home graphics cards” hasn’t been halted:

As of now (May 2026), the main consumer lineup is the GeForce RTX 50 Series based on the Blackwell architecture:
Flagships: GeForce RTX 5090 and RTX 5080 - the most powerful solutions for 4K gaming and AI workloads.
Mid-range segment: GeForce RTX 5070 Ti and RTX 5070 (including the updated mobile version of the RTX 5070 with 12 GB of memory, introduced in late April 2026).

Budget solutions: GeForce RTX 5060 and RTX 5050, which remain popular for Full HD resolution

[WillyAp]

But this makes sense-it’s now more profitable for the company to establish itself in the AI market, earn more revenue, and develop new technologies in order to maintain or gain new market positions.

To me Nvidea is a snake oil sales platform now.
They know that once the AI industry is forced to collect money from every prompt, which comes rather earlier (Many Data centres won't be realized) the party ends.
The time people should remember email went thru a similar decision. The only companies selling email services are Google, MS, Zoho. Protonmail among others. 

[DrBeer]

But this makes sense-it’s now more profitable for the company to establish itself in the AI market, earn more revenue, and develop new technologies in order to maintain or gain new market positions.

To me Nvidea is a snake oil sales platform now.
They know that once the AI industry is forced to collect money from every prompt, which comes rather earlier (Many Data centres won't be realized) the party ends.
The time people should remember email went thru a similar decision. The only companies selling email services are Google, MS, Zoho. Protonmail among others. 

The example of email suggests that there will be a division between free and paid services, with the latter typically catering to corporate users. AI is following the same path:
- Ordinary “home users” can use a basic set of services for free
- if they need a little more convenience, there are simple and affordable subscriptions.
But home users don’t need email systems for 1,000+ users or  petabyte-scale storage-corporate customers come for that and pay for it.
AI will follow the same path, which is basically what we’re seeing now

[STT]

That's why it's crucial that the Bitcoin price rises by 2028 (ideally to $300,000). Otherwise, most miners will simply go bankrupt! 🙋

Im not sure its quite that simple that price is dictated by the profitability, its all subject to adjustment in the workload not just the price.
  The other scenario is that not enough mining power is available so difficulty adjusts, its not a one way street.  Competition itself is what has made the blocks harder to solve, we could all be doing it via cpu if there were some simple agreement to never use more power perhaps.  That would be really unrealistic of course, its a free market and people will compete but I dont think the price and all the differing factors simply dictates where price can or should go.

[WillyAp]

The example of email suggests that there will be a division between free and paid services, with the latter typically catering to corporate users. AI is following the same path:
- Ordinary “home users” can use a basic set of services for free

Funny enough it is mostly home users who use Google Services.
Enterprises prefer MS, upcoming Enterprises Zoho.
All have issues compared to a server-based Email solution, MS does not handle alle server incoming emails  from all severs.
Gmail has issues when sharing information over email. Those emails are in the sent mail folder, but they never reach anyone.

Ai, Crypto, Enterprises most use blending as a marketing sales strategy.   

[dezoel]

There is no mining power shifting to AI, people think there must be in idea, but there is none. The difficulty rises even during hardest times of price falls, so there is none I can assure you that.

It has become a legitimate business in many places. The difference was, it was hobbyist thing when it first started, and then it became a thing where some small companies formed, spending millions on it. Nowadays there are ones who have billion dollar marketcaps, so it has grown a lot.

[Etranger]

There is no mining power shifting to AI, people think there must be in idea, but there is none.

Yes, that is not entirely the case. The audience of miners and the audience of people working with AI sometimes overlap, but they are not identical. And it is not as if the people who are now working with AI are simply those who quit mining and moved into a new field. New infrastructure is being built for AI, and it is undoubtedly in very high demand. But at the same time, this infrastructure is not “consuming” the potential of mining. Both are being developed simultaneously, these are parallel processes

[DrBeer]

The example of email suggests that there will be a division between free and paid services, with the latter typically catering to corporate users. AI is following the same path:
- Ordinary “home users” can use a basic set of services for free

Funny enough it is mostly home users who use Google Services.
Enterprises prefer MS, upcoming Enterprises Zoho.
All have issues compared to a server-based Email solution, MS does not handle alle server incoming emails  from all severs.
Gmail has issues when sharing information over email. Those emails are in the sent mail folder, but they never reach anyone.

Ai, Crypto, Enterprises most use blending as a marketing sales strategy.   

Yes, that’s exactly what I was writing about: free services are funded, for example, by the same ads that users see through Google services.
Yes, and you’re right about the issues; but here we need to remember that any UNIVERSAL system or service always has its drawbacks; it’s like an amphibious vehicle-it doesn’t drive very well (compared to regular cars) and doesn’t float very well on water (compared to a yacht), but its versatility makes it convenient

[DrBeer]

I analyzed the electricity market and made some interesting observations:

Mining.
- Electricity contracts are mostly short- and medium-term, with high price sensitivity (spot + fixed PPAs in those regions).
- The main model is flexible load: we shut down during peak demand and turn on during surplus.
- In many regions, mining has effectively become a “buffer for the grid” rather than a priority consumer.

AI data centers.
Almost always:
- long-term PPAs (10–20 years)
-or direct “energy+capacity reservation” contracts

Often built:
-alongside generation facilities (co-location)
-or with guaranteed grid connection (grid priority access)

What is happening in the market?
Hyperscalers (roughly: AWS / Google / Microsoft) are already:
-reserving capacity years in advance
-competing with industry for grid connections
-Energy companies have begun building new capacity for specific AI clusters, rather than "for the market"

[henmark]

I analyzed the electricity market and made some interesting observations:

Mining.
- Electricity contracts are mostly short- and medium-term, with high price sensitivity (spot + fixed PPAs in those regions).
- The main model is flexible load: we shut down during peak demand and turn on during surplus.
- In many regions, mining has effectively become a “buffer for the grid” rather than a priority consumer.

AI data centers.
Almost always:
- long-term PPAs (10–20 years)
-or direct “energy+capacity reservation” contracts

Often built:
-alongside generation facilities (co-location)
-or with guaranteed grid connection (grid priority access)

What is happening in the market?
Hyperscalers (roughly: AWS / Google / Microsoft) are already:
-reserving capacity years in advance
-competing with industry for grid connections
-Energy companies have begun building new capacity for specific AI clusters, rather than "for the market"

One thing about getting a contract from an electricity company that is that big, is the fact that you have to convince them that you are not going to bankrupt or stop. You can be google, one of the biggest companies in the world, and you still have to end up providing proof that you will continue.

Sure google won't bankrupt, but what if they decide to stop? Overall, even with a contract, it's hard to get 100% recouped, so there will be legal war that will continue for a long time and google has army of lawyers that will make sure they pay as little penalty as possible or at least pay it very late when it's all done.

So, these energy companies want to make sure that if you have a 20 year lease, then you will continue for 20 years, and convincing of them is the hardest part.

[WillyAp]

Sure google won't bankrupt, but what if they decide to stop?

What do think will happen if Alphabet's shares slide towards cero?