As AI companies work furiously to improve the intelligence and usefulness of their products, their demand for cheap, plentiful energy has skyrocketed. This gold rush has been extremely profitable for an unlikely beneficiary: Bitcoin miners.
In recent months, major Bitcoin mining companies have started to swap out some of their mining equipment in favor of rigs used to run and train AI systems. These companies believe that AI training could provide a safer and more consistent source of revenue than the volatile crypto industry. And so far, these pivots have been warmly received by investors, leading to the market cap of 14 major bitcoin mining companies jumping in value by 22%, or $4 billion, since the beginning of June, J.P. Morgan reported on June 24.
This transition reflects several trends of the moment: the roaring hype cycle of AI; the dwindling access to power, and a tenuous bitcoin mining landscape following the bitcoin halving.
Read More: What’s the Deal With the Bitcoin Halving?
The AI boom has led to an enormous demand for energy
Generative AI models like ChatGPT improve through the brute computational might of data centers, which process massive data sets to find patterns and improve responses. But computing power is expensive, and for years, wasn’t a worthwhile investment for many data center operators. When IREN, a data center and bitcoin mining company, looked into using their spaces for machine learning four years ago, “there just wasn't enough volume from a commercial perspective for it to make sense,” says Kent Draper, IREN’s chief commercial officer.
But the gargantuan success of ChatGPT beginning in late 2022 changed the calculus, and other AI companies raced to train and run their own models in the hopes of outpacing OpenAI’s flagship model. This requires a stupendous amount of energy: A ChatGPT query, for example, uses 10 times more energy than a standard Google query.
This leaves AI companies on the hunt for direct access to inexpensive power sources, large tracts of land to hold warehouses filled with thousands of computers, and resources like water or giant fans to cool their machines. Their ravenous activity means it’s becoming increasingly competitive to find sites that meet those criteria, especially in North America. Some jurisdictions have implemented long waitlists for large data centers to connect to the grid. And once companies get initial approval, building a data center from scratch can take years, millions of dollars, and necessitate lengthy slogs through regulation and bureaucracy.
Read More: How AI Is Fueling a Boom in Data Centers and Energy Demand
“If you go back five or 10 years, 80% of the data center loads were located in six or seven primary markets,” says Nazar Khan, the COO and CTO of the bitcoin mining company Terawulf. “Those markets are filled up, and a couple of them have already issued moratoriums on further data center construction. So those data center loads are now looking for new homes.”
Bitcoin miners face headwinds
Some of those homes, it turns out, lie within the existing facilities of bitcoin miners. Bitcoin miners uphold and safeguard the bitcoin network through a complex computational process, and earn bitcoin for doing so. In the early years of bitcoin, miners found that increasing the size of their computer rigs vastly increased their profits, so they created enormous server farms which tapped into cheap sources of energy and ran day and night.
Large-scale bitcoin mining has historically been an immensely profitable business. But it is also subject to the whims of the volatile crypto market. Following the 2022 crypto crash—which was precipitated by the risky endeavors of entrepreneurs like Sam Bankman-Fried and Do Kwon—many miners were thrust into bankruptcy or shuttered their doors altogether.
Mining companies who survived the crash reaped profits in 2023 and early 2024. But a new challenge emerged this April: a technical update to Bitcoin called the halving, which slashed miners’ rewards in half. Bitcoin miners hoped that the halving would lead to a dramatic increase in the price of bitcoin, as has happened in previous crypto cycles, to offset this reward decrease. But bitcoin’s price has stayed more or less even since April, squeezing bottom lines, and forcing some miners to seek ways to diversify their business models. AI training is at the top of the list.
“You've seen a number of crypto miners that were sort of struggling that have actually made a full pivot away, and that may have been a function of necessity,” Draper says.
The partnership between the AI and bitcoin mining industries is a logical one, given the needs from both sides. AI companies need the space, access to cheap energy, and infrastructure that bitcoin miners already have. And bitcoin miners seek the stability of AI compute revenue, and the enormous potential profits flowing from the current AI hype cycle.
Some bitcoin mining companies are leasing their space to AI clients. In June, Core Scientific—which recently emerged out of bankruptcy stemming from the 2022 crypto crash—announced it would host over 200 megawatts of GPUs (graphic processing units, which power AI training and operation) for the AI startup CoreWeave. Core Scientific CEO Adam Sullivan told TIME in April that AI companies were making aggressive offers for the use of bitcoin mining facilities: “They have started to buy up mining sites for greater prices than what Bitcoin miners are willing to pay,” he said. He added that the number of requests from AI companies was “extraordinarily high on our side, and we’re evaluating our best go-to-market here.”
Other bitcoin mining companies are operating the GPUs themselves. On June 24, bitcoin miner Hut 8 received a $150 million investment from Coatue Management to build artificial intelligence infrastructure. And in some IREN facilities, GPUs, for AI, and ASICs (application-specific integrated circuits which power bitcoin mining), share the same walls. “We view them as mutually complementary: They’re quite different business profiles,” Draper says. “Bitcoin is instant revenue but somewhat more volatile. AI is customer-dependent—but once you have customers, it’s contracted and more stable.”
This increase in demand has climate repercussions
With bitcoin miners operating in both industries, an enormous amount of energy is being used. Data centers use 10 to 50 times the energy of a typical commercial office building, the U.S. Department of Energy says. A recent Goldman Sachs report predicted that data centers will use 8% of total U.S. power by 2030, up from 3% in 2022. This level of electricity growth “hasn’t been seen in a generation,” the report read.
Some bitcoin companies, like Terawulf, say they’re focused on using green energy. But many of the new data centers overall are being powered by fossil fuels. “Some of the smaller renewables don’t meet the demand for consistent and high quality energy that some of the high speed compute folks require,” Khan says. “You’re seeing utilities proposing to add more large-scale gas-fired power plants, which we haven’t seen for a number of years. It will take a portfolio of facilities: gas, nuclear, renewables to meet this need.”
All of this activity is concerning climate activists. “Bitcoin miners are diversifying into traditional data centers and AI—and obviously they use different machines, but they still use voracious amounts of energy,” says Mandy DeRoche, a deputy managing attorney at Earthjustice’s clean energy program. "That tremendous increase in energy demand has consequences for the grid, for the cost of electricity, and the climate."
Andrew R. Chow’s book about crypto, Cryptomania, will be published in August and is available for preorder.
More Must-Reads from TIME
- Introducing the 2024 TIME100 Next
- Sabrina Carpenter Has Waited Her Whole Life for This
- What Lies Ahead for the Middle East
- Why It's So Hard to Quit Vaping
- Jeremy Strong on Taking a Risk With a New Film About Trump
- Our Guide to Voting in the 2024 Election
- The 10 Races That Will Determine Control of the Senate
- Column: How My Shame Became My Strength
Contact us at letters@time.com