Or: How the “Future of Finance” Got Outbid by ChatGPT’s Electric Bill
Here’s an irony for you: Bitcoin miners spent a decade finding the cheapest electricity on Earth—stranded hydro in Iceland, surplus wind in Texas, natural gas flares in North Dakota. They built an entire industry around being the buyer of last resort for power nobody else wanted.
Then AI showed up with deeper pockets and a willingness to pay more. And suddenly, Bitcoin mining’s economic model has a problem.
This isn’t a “Bitcoin is dying” story. It’s a story about what happens when two energy-intensive technologies compete for the same resource, and one of them can afford to pay premiums while the other is built on razor-thin margins.
Let’s talk about what’s actually happening, why it matters, and what it means if you own Bitcoin, mining stocks, or are just trying to understand why your electricity bill went up 10% last year.
What’s Actually Happening (The Short Version)
The Setup: Bitcoin mining has always been about cheap electricity. Miners don’t care where the power comes from—they just need it to be cheaper than their competitors’ power. That’s the whole game.
The Problem: AI data centers need electricity too. Lots of it. And unlike Bitcoin miners, they can’t just shut down when the grid gets stressed. Training large language models and serving ChatGPT queries requires 24/7 uptime.
The Numbers: In Texas alone, large-scale power requests hit 226 gigawatts in 2025. About 73% came from AI companies, not miners. AI is now the biggest driver of new electricity demand growth in the United States.
The Result: Electricity prices are rising. Bitcoin miners, who built their business on cheap power, are getting priced out. Some are pivoting their entire operations to serve AI customers instead of mining Bitcoin.
And then there’s the political angle: President Trump is reportedly pushing for an “emergency power auction” that would force tech companies to bankroll $15 billion in new power plants. The justification? Household electricity bills jumped 10.5% between January and August 2025—one of the largest increases in over a decade.
Why Bitcoin Mining Needs Cheap Electricity (A Quick Primer)
If you’re not familiar with how Bitcoin mining works, here’s the simplified version:
Bitcoin’s network is secured by miners who compete to solve complex mathematical puzzles. The winner gets newly created Bitcoin plus transaction fees. But here’s the catch: the difficulty adjusts every two weeks to keep blocks coming roughly every 10 minutes, no matter how many miners are competing.
What that means in practice: Mining Bitcoin is a race to see who can do math problems the cheapest.
If you’re paying $0.03 per kilowatt-hour for electricity and your competitor is paying $0.05, you win. If electricity costs rise to $0.08, suddenly you’re not profitable anymore. Your competitor already shut down. And now you’re thinking about shutting down too.
Bitcoin miners survive by finding electricity that nobody else wants:
- Excess hydro production in the Pacific Northwest during spring runoff
- Wind farms producing more power than the grid can use at 2 AM
- Natural gas that would otherwise be flared (burned off as waste) at oil wells
- Industrial sites with contracted power they’re not fully using
The entire Bitcoin mining industry is built on being flexible about when and where they use power, in exchange for getting it cheap.
How AI Changed Everything
AI data centers are different in three critical ways:
1. They can’t be flexible. If you’re training GPT-5 or running inference for millions of ChatGPT queries, you can’t just shut down when electricity gets expensive. Your customers expect 24/7 uptime. That means you need firm, reliable power—not the “we’ll turn off when you need it” arrangement Bitcoin miners offer.
2. They can afford to pay more. Nvidia’s gross margins on AI chips are massive. Microsoft, Google, and Meta are spending tens of billions building AI infrastructure. When you’re making that kind of money, you can afford to pay premium electricity rates. Bitcoin miners operating on 10-20% margins cannot.
3. Utilities prefer them as customers. From a power company’s perspective, AI data centers are ideal: they sign long-term contracts, they pay reliably, they don’t curtail during peak demand. Bitcoin miners? They’re the interruptible load that gets cut when the grid gets stressed. That’s valuable for grid stability, but it’s not the kind of customer that gets priority when new capacity comes online.
The result: Bitcoin miners are losing the bidding war for electricity. And some of them are drawing a logical conclusion: if we can’t beat AI, we should join them.
The Great Mining Pivot: From Bitcoin to AI
Here’s where it gets interesting. Several major Bitcoin mining companies are converting their operations to serve AI customers instead:
Bitfarms announced plans to exit Bitcoin mining entirely by 2027, converting their Washington facility to GPU-as-a-Service for AI workloads. The CEO, Ben Gagnon, was refreshingly honest about why: “You can move over to a business that is not volatile and is predictable, has higher revenues, higher margins, higher cash flows.”
(Translation: Bitcoin mining is exhausting. AI customers pay better and don’t care if Bitcoin dropped 15% last week.)
Galaxy Digital, CleanSpark, and IREN are all converting mining sites into AI facilities. They already have the power infrastructure, the cooling systems, and the data center expertise. They just swap out ASIC miners for GPU clusters and suddenly they’re in a more profitable business.
Riot Platforms sold 1,080 Bitcoin to buy 200 acres at their Texas site outright, then signed a 10-year deal with AMD to provide data center space. They’re hedging: still mining some Bitcoin, but increasingly focused on AI infrastructure.
The pattern is clear: Bitcoin miners built valuable infrastructure in places with cheap power. Now that infrastructure is worth more serving AI than mining Bitcoin.
What This Means for Bitcoin (The Technical Stuff)
Okay, so miners are leaving. Why does that matter for Bitcoin?
Hash Rate and Security: Bitcoin’s security comes from computational power (hash rate). If mining becomes unprofitable and miners shut down, hash rate drops. Lower hash rate means less security (in theory—Bitcoin is still massively secure by any measure, but the direction matters).
Mining Difficulty Adjustment: Here’s the good news—Bitcoin’s difficulty adjusts every 2,016 blocks (roughly two weeks). If miners drop off and blocks slow down, difficulty decreases, making mining easier for those who remain. This is Bitcoin’s built-in stabilizer. It keeps working even when hash rate drops.
Decentralization Concerns: Bitcoin mining was already concentrating in a few regions (Texas, Kazakhstan, parts of China despite the ban). If the only miners who can survive are the ones who can afford premium electricity rates or who have special deals with power companies, that’s further centralization. Not great for a network that’s supposed to be decentralized.
The Long-Term Question: What happens if cheap electricity essentially disappears for Bitcoin miners? Does mining move to places with subsidized power? Does it become a niche activity? Does the block reward (currently 3.125 BTC per block, halving every four years) eventually not be enough to keep miners interested?
We don’t know yet. But the trend is concerning.
The Trump Power Auction: Political Response to an Energy Crisis
Here’s where the political world collides with the technical:
President Trump, along with governors from Pennsylvania, Ohio, Virginia, and other northeastern states, is reportedly pushing PJM (the largest U.S. electricity grid operator) to hold an emergency power auction. The proposal would force tech companies to bid on long-term power contracts, essentially underwriting $15 billion in new power plant construction.
The stated reason: Electricity prices are squeezing households. The 10.5% increase in 2025 hit regular people hard, and nobody wants to explain to voters why their electric bill went up because tech companies are training AI models.
The unstated reason: The grid literally cannot handle current demand growth. If AI buildout continues at this pace, we’re looking at brownouts and blackouts in major metro areas. New capacity takes years to build. Something has to give.
What this means for miners: If tech companies are forced to finance new power plants, that’s new capacity coming online. In theory, that could ease pressure on electricity markets. In practice, a lot depends on who gets priority access to that new capacity—and given that Bitcoin miners are interruptible loads while AI data centers aren’t, you can guess who wins that fight.
Real Talk: What This Means For You
Let’s get practical. What should you take away from this if you’re:
Holding Bitcoin: Your Bitcoin isn’t going anywhere. The network will keep running. But the economics of mining are changing, and that has long-term implications for network security and decentralization. Not a crisis today, but something to watch.
Holding mining stocks (MARA, RIOT, etc.): You’re not investing in Bitcoin miners anymore—you’re investing in energy infrastructure companies that might pivot to AI. That could be good (more stable revenue) or bad (you wanted Bitcoin exposure, not data center exposure). Read the earnings calls. Watch for language about “diversification” and “GPU hosting.”
Interested in mining yourself: Unless you have access to very cheap electricity (under $0.03/kWh) or you’re in it for ideological reasons, hobby mining is increasingly uneconomical. The days of profitable garage mining ended years ago, and industrial mining is getting squeezed.
Just trying to understand your electricity bill: AI and Bitcoin both contribute to rising electricity demand. AI is now the bigger factor. Your rates are going up partly because utilities are investing in new capacity to meet data center demand, and they’re passing costs through to everyone.
The Bigger Picture: Energy as the Bottleneck
Here’s what this story reveals about the next decade of tech:
For years, people talked about Bitcoin’s energy consumption as a unique problem. “Bitcoin uses more electricity than Argentina!” Well, guess what—AI might use more than Bitcoin by 2026, according to some estimates.
The real bottleneck isn’t chips or algorithms or innovation. It’s energy. Specifically, it’s grid capacity, transmission infrastructure, and generation that can run 24/7 at scale.
Bitcoin mining could be flexible about energy (operate when cheap, curtail when expensive). AI can’t be. That’s why AI is winning the bidding war.
But this creates a weird dynamic: Bitcoin miners offered grid stability by being the load that could be turned off during stress. If they’re all gone, replaced by AI data centers that run 24/7, the grid becomes less flexible, not more.
BlackRock’s analysts summarized it well: AI data centers could use up to 25% of U.S. electricity by 2030. That’s not a prediction you can hand-wave away.
The Irony Nobody’s Talking About
Bitcoin was supposed to be the future of finance—decentralized, censorship-resistant, outside the control of governments and corporations.
It’s getting displaced by AI, which is arguably more centralized (dominated by a handful of tech companies), more dependent on existing power structures (who else can build $10 billion data centers?), and more integrated with governments (who are now negotiating power access).
The technology that was supposed to disrupt the system is being outcompeted by a technology that works within the system very comfortably.
Maybe there’s a lesson there about what actually wins in the real world. Or maybe it’s just capitalism doing what capitalism does: capital goes where returns are highest, and right now, AI returns beat Bitcoin mining returns.
What Happens Next?
Short term: More miners pivot to AI or shut down. Hash rate continues to fluctuate. Difficulty adjusts. Bitcoin keeps running, just with different economics for miners.
Medium term: New power capacity comes online (if Trump’s auction happens). Maybe electricity prices stabilize. Maybe miners find a new niche with that capacity. Or maybe they don’t, and mining becomes even more concentrated.
Long term: We figure out whether Bitcoin can maintain decentralization and security with a mining industry that looks very different from what Satoshi probably envisioned. Maybe it works fine. Maybe it’s a problem. We’ll find out.
What we know for sure: Bitcoin mining’s “golden age” of finding cheap stranded energy and printing money is over. The business is maturing, margins are compressing, and competition for electricity is fiercer.
That’s not the end of Bitcoin. But it is the end of Bitcoin mining as we’ve known it for the past decade.
The Bottom Line
Bitcoin miners built an industry on being the smartest buyers of electricity. They found power nobody else wanted and turned it into digital gold.
Then AI showed up willing to pay more for the same power. And it turns out, when you’re competing against companies with trillion-dollar market caps who need your electricity to stay online 24/7, your “flexible load” advantage isn’t worth as much as you thought.
Some miners are adapting by switching to AI infrastructure. Some are trying to hold on. Some are shutting down.
Bitcoin will survive this—it’s designed to survive hash rate drops and miner exits. But the economics of the network are changing in ways that matter.
If you own Bitcoin, you should understand that the network securing your coins is undergoing a major shift. If you own mining stocks, you should understand that you might be investing in AI infrastructure companies soon, whether you meant to or not.
And if you’re just watching your electricity bill go up and wondering why, well—now you know. ChatGPT is expensive to run, and somebody’s gotta pay for it.
Turns out that somebody is partially Bitcoin miners, and partially you.
This is not financial, investment, or energy policy advice. This is me explaining a complicated situation where two energy-intensive technologies are competing for the same resource, and one of them is winning. What you do with that information is up to you.
