Bitcoin Mining Is So Back (Except It’s AI Now)

Bitcoin is up 7% over the last five days. You know what that means? Bitcoin mining is so back (until bitcoin’s price falls 5% over a five-day stretch again, after which it’ll be so over again).

With bitcoin’s price surge, the stock prices of four of the five largest publicly traded miners (measured by total hashrate, or computing power spent securing the Bitcoin network) are up double-digit percentage points.

The one laggard, Iris Energy Ltd (IREN), the fifth largest in this quintet, is down 15% following a report published last week by Culper Research in which the firm disclosed a short position in IREN. The reason Culper’s taking a bearish bet: the unsuitability, in the researchers’ view, of Iris’ Childress, Texas site for artificial intelligence (AI) or high-performance computing (HPC).

AI and HPC might seem unrelated to bitcoin mining, but such diversification has become a way for bitcoin miners to make money, as evidenced by Core Scientific’s (CORZ) 200 megawatt (MW) AI deal with CoreWeave last month, which pumped CORZ’s share price by 40%.

(Perhaps if the price of bitcoin continues its upward creep, the unsuitability of IREN’s sites for non-bitcoin mining revenue-generating activities can fall to the wayside as the company shifts resources back to bitcoin mining.)

In any event, what “bitcoin mining is so back” really means is “bitcoin mining stocks are back,” because on a pure “are there more miners now?” basis, known pool hashrate has only increased slightly over the last five days (from 663.618 exahashes per second to 668.659 Eh/s) rather than increase by 7% as one might expect. (Note: there is no “perfect” data point for hashrate.) Of course, hashrate not reacting immediately and proportionally to a bitcoin price increase is good for the public companies.

But then if you look into the narrative around bitcoin mining and check out what the mining companies are saying, in interviews or public filings, you find that, while they’re still focused on bitcoin mining, there’s much ado about other seemingly unrelated or tangentially related things.

Here’s a splashy headline: Private Equity Giants Are Circling Bitcoin Miners on AI Allure

And another: Core Scientific Upgraded to Buy From Neutral to Reflect HPC Expansion: B Riley

And one more, as good things come in threes: Bitcoin Mining Sector Is Attracting Growing Investor Interest Following Core Scientific Deal: JPMorgan

Last week, I wrote about how both AI and Bitcoin use a lot of energy and, not only that, it seems that it is straightforward for Bitcoin mining facilities to be retrofitted for the next hot thing: AI (or HPC, if you want to avoid the backlash against AI hype).

Investors like this adaptability. From CoinDesk’s Will Canny and Aoyon Ashraf, “Private equity (PE) firms are finally seeing value in bitcoin (BTC) miners, thanks to the rising demand for data centers that can power artificial intelligence-related (AI) machines.”

Research from JPMorgan suggests the same thing and, funnily enough, the investment bank’s research says that IREN (the company Culper deems “not ready for AI”) is best positioned to capitalize on this resource-shifting trend.

Will Foxley, co-founder of Blockspace Media and host of the The Mining Pod, expressed skepticism about claims that Bitcoin mining facilities are suitable to transition to supporting AI computing.

“A lot of these bitcoin miners are just talking about how they can do AI when in reality they aren’t able to do it,” Foxley told CoinDesk.

I’ve contended before that going public is dumb. One of the reasons is that it requires a company to shift to a short-term, quarterly earnings-focused mindset when long-term goals (such as growth into perpetuity or existing next decade) should be the focus. It also makes it such that if a company is struggling, everyone knows, which can make a company vulnerable.

Mining companies were struggling in 2022. Core Scientific (CORZ) even declared bankruptcy. And this was all before the Bitcoin halving in April 2024 cut deeply into miners’ revenue prospects. It was tough for miners in general and, because there are a bunch of public mining companies, competitors could pinpoint exactly who was struggling. Riot Platforms (RIOT) tried to take advantage of this situation and made a takeover bid for a smaller mining company, Bitfarms (BITF). Because BITF is public, RIOT didn’t need to call on BITF leadership and ask politely. Instead RIOT bought a lot of BITF stock in a hostile takeover attempt. This could have worked out well if RIOT was correct in assuming that its operation was better and more efficient than BITF’s, but we won’t ever know as the takeover attempt ultimately failed.

There are other financial tricks out there that can pad shareholder returns (or tank them if unsuccessful; RIOT’s stock is down 25% this year). One example is being purchased by mutual agreement, which is what Coreweave tried to do after it made its AI deal with Core Scientific. The offer was rejected, but it is telling that an AI company with growth aspirations looked at a bitcoin mining company and thought: “hold on a second, we need to grow our operations quick before the AI boat sails past us, and bitcoin miners have warehouses that we could retrofit for our use, so we should buy them.”

“I think some of these bitcoin companies are sitting on attractive power contracts and if you’re a huge data center hyperscaler like Coreweave, what’s a few billion dollars to go level a bitcoin mining site and throw up a new AI data center?” Foxley said. “Of course the takeover would be expensive, but you’re betting that the longevity of the power contract pays you back based both on the multiple you’re going to get being a public AI company and on the revenue of just being an AI company.”

Coreweave surely cannot be the only AI company thinking this.

Mining companies used to mine ether before Ethereum shifted from proof-of-work to proof-of-stake and so now these companies only mine bitcoin.

At least that was what most thought until Marathon (MARA) revealed it had been mining a relatively obscure cryptocurrency called Kaspa since September 2023. Kaspa is, by most measures, a completely random crypto that just so happens to be mineable. Marathon had access to space and electricity to throw at it, it seemed profitable, and so the company did it because profitable activity is good.

“By mining Kaspa, we are able to create a stream of revenue that is diversified from Bitcoin, and that is directly tied to our core competencies in digital asset compute,” said Adam Swick, Marathon’s chief growth officer, in a statement.

I think the mining of Kaspa, and potentially other coins, is more a novelty than a concrete industry shift, because I doubt another proof-of-work cryptocurrency will ever rise to prominence.

But Marathon’s move further highlights the broader point: Bitcoin miners are hurting for revenue and profitability, and they are looking in places besides mining bitcoin to make up the difference.

Note: The views expressed in this column are those of the author and do not necessarily reflect those of CoinDesk, Inc. or its owners and affiliates.