Bitcoin Miners Go All In on AI: What’s the Cost?

Bitcoin miners fully transitioning to AI computing are ignoring an important lesson.

By: Prathik Desai

Compiled by: Luffy, Foresight News

In the past year, every major publicly listed Bitcoin miner has faced a survival crisis. A look at their financial reports reveals that they are shedding their identity as "miners."

Now, they are rebranding themselves as "energy infrastructure platforms," "vertically integrated AI cloud service providers," or "digital infrastructure companies built on power, land, and computing resources."

Behind this brand transformation is a shift in revenue structure, with their market valuations now surpassing those from their days solely as Bitcoin mining companies.

On the surface, the story looks promising. Struggling miners are seizing the opportunity to embrace the surging demand for AI inference computing, finding a path to self-rescue. However, a broader perspective reveals that they are also abandoning some critically important aspects.

This article will explain why miners' 180-degree turn away from Bitcoin mining could backfire in the future.

What prompted this transformation?

With the Bitcoin block reward halving in April 2024, miners are forced to make a choice: either mine Bitcoin and hold it for a higher price later, or upgrade their hardware to pivot towards high-performance computing (HPC) for business diversification. Bitcoin's price surged from under $70,000 in October 2024 to $124,000 in October 2025, prompting some miners to continue mining and hoarding their Bitcoin output.

However, the market liquidation event on October 10 saw $19 billion in market value evaporate from the crypto market within 24 hours, triggering a downward cycle that has yet to fully recover.

Currently, Bitcoin's price has nearly halved from last October's peak, sitting around $63,000. At this price point, selling mined Bitcoin is nearly unprofitable. The cost of Bitcoin mining is influenced by both electricity prices and network difficulty. The overall mining cost rose from about $40,000 in February 2024 to between $90,000 and $110,000 from October 2025 to July 2026; by August this year, the comprehensive mining cost hit a new high of $140,000.

Source: MacroMicro

At the same time that the Bitcoin mining environment continues to worsen, a flood of capital is pouring into the AI inference sector.

Bitcoin miners use ASIC chips to run complex algorithms to earn block rewards. This hardware infrastructure is highly similar to the underlying conditions required for high-performance computing and AI inference. This similarity allows miners to transition to AI inference without starting from scratch, significantly lowering the barrier to entry. For miners seeking profitable avenues, pivoting to AI seems like a natural choice.

Almost the entire publicly listed mining sector has collectively shifted direction. The changes in revenue structure over the past few quarters visually reflect the scale of this transformation.

Take Core Scientific as an example: In Q2 2025, its data center hosting business for AI and high-performance computing generated only $10.6 million in revenue; during the same period, revenue from mining and selling Bitcoin was $62.4 million. Twelve months later, the positions of the two have almost completely reversed. In Q2 2026, hosting business revenue reached $136.7 million, while mining revenue shrank by 65% to $21.5 million.

The share of hosting business in the company's total revenue skyrocketed from 14% last year to 83%.

TeraWulf's high-performance computing rental revenue now accounts for 71% of total revenue. This business had no revenue in the same period last year but reached $32 million in Q2 2026; during the same period, cryptocurrency mining revenue shrank by about 75%, leaving only $13 million.

Many miners are also repurposing their existing mining facilities and hardware to accommodate AI and high-performance computing businesses.

A Cautionary Land Grab

The shift from Bitcoin mining to AI computing has evolved into a collective sprint. In addition to reallocating existing capacity to AI, former Bitcoin miners are aggressively securing electricity and land, signing numerous demand contracts. The total value of contracts already signed has reached hundreds of billions of dollars.

Core Scientific has committed to AMD and CoreWeave for up to 2.5GW of available computing capacity, with total potential revenue exceeding $24 billion over the contract period. Hut 8 holds 949 megawatts of contracted capacity, with a base contract value of $26.6 billion, alongside securing $7.5 billion in new project financing.

TeraWulf signed a 20-year lease agreement worth about $19 billion with Anthropic and acquired a gigawatt-scale facility in Kentucky to meet computing demand. Over the past six months, Riot Platforms has completed multiple leases totaling 241 megawatts, corresponding to a contract value of about $10 billion. Just last week, IREN delivered its first batch of AI cloud deployment nodes to Microsoft, with a five-year contract valued at $9.7 billion; it also reached a 5GW partnership with NVIDIA, continuously acquiring power stations in Texas and Spain to meet computing needs.

This series of large contracts has ignited market enthusiasm. Compared to miners' past business, signing 20-year leases with leading cloud providers means more stable cash flow. After a year of continuous losses in mining, locking in contracts lasting up to fifteen years seems like a solid reality.

However, many market commentators underestimate the risks lurking within.

Despite the challenges of mining, Bitcoin mining possesses a characteristic that AI hosting businesses do not: a self-correcting mechanism. When mining becomes unprofitable, miners will shut down their machines; when enough miners exit, the overall network difficulty of Bitcoin automatically adjusts downward. The remaining miners, with the same equipment and electricity costs, can achieve higher returns.

As many miners shift to the AI sector, Bitcoin's overall network difficulty has fallen from a peak of about 156 trillion in 2025 to 127.5 trillion. Each company that exits mining is making the mining returns for those who remain higher.

However, the AI computing industry lacks this automatic adjustment mechanism. If a large number of players enter, the supply of computing power may exceed demand, leading to a continuous decrease in prices.

Although the overall demand for AI computing is rising, prices may actually fall—this seems contrary to basic economic principles. In fact, several AI companies have already begun to lower service prices. In the early days of the industry, hardware and supporting resources were scarce, allowing service providers to charge high premiums; now, major cloud providers worldwide are competing for megawatt-level power resources, and this scarcity is attracting many Bitcoin miners and data center companies to expand capacity. Once the new supply fills the demand gap, computing prices will quickly decline.

When prices fall, miners who signed long-term leases at today's high prices will look back and find the cost to be steep.

Another tricky aspect of this transformation is that it is essentially a one-way street. Converting a mining facility into an AI computing park is straightforward, as the power infrastructure is already in place. However, once you replace ASIC miners with H100 GPUs and sign a twenty-year lease, you will be bound by this contract for the next twenty years, unable to easily revert to mining.

When the market cycle reverses, and Bitcoin prices rise above the $60,000 low, combined with a decrease in overall network difficulty, mining will inevitably become profitable again. At that point, former miners who have fully transitioned will be constrained by long-term lease agreements and can only watch opportunities slip away. Worse still, if the actual returns from AI computing fall short of the rosy expectations at the time of signing, a double blow will follow.

So should miners not transition to AI and patiently wait for the mining cycle to reverse? I don’t think so. For most companies, transitioning is a survival-driven choice. With comprehensive mining costs reaching $140,000 while the coin price is only $63,000, continuing to mine is not commercially viable; AI is the only lifeline at present.

However, the desire for survival can drive companies to make extreme choices. The risks come precisely from those players going all in on AI: under survival pressure, they completely abandon Bitcoin mining, destroy or fully replace ASIC miners, and lock themselves into decades-long contracts in an emerging industry that will still experience price corrections.

However, not all miners have chosen to make a complete 180-degree turn; some companies are taking a more cautious approach.

Hedging Options Favor the Cautious

Marathon Digital, which holds the largest Bitcoin treasury among publicly listed mining companies, has taken a cautious route. In Q2 2026, it sold 30% of its Bitcoin holdings to reduce debt, but it did not abandon its Bitcoin mining equipment; instead, it chose another way to fund its AI expansion.

MARA leveraged its remaining Bitcoin as collateral to secure $150 million in funding through a 2026 credit facility to finance its AI business expansion. As of June 30, it still holds over 35,000 Bitcoins on its balance sheet while maintaining a flexible mining operation.

Hut 8 also has not given up mining. It has spun off its Bitcoin mining business into a new subsidiary, American Bitcoin Corp, which is fully controlled by the parent company, while the parent company shifts its focus to high-performance computing and AI infrastructure.

By keeping mining operations running and maintaining Bitcoin inventory, these companies hold a choice that has been completely abandoned by those who have gone all in on transformation. When mining becomes profitable again, they can redirect their power load back to mining; meanwhile, companies deeply bound by long-term AI contracts can only watch from the sidelines.

Treating the old mining business as a burden and fully betting on AI computing is akin to gambling on an unproven future. However, historical patterns are not difficult to predict: with each emerging technology, such as chips, computers, and smartphones, products continuously decrease in price as the industry matures, and signs of price loosening are already appearing in the AI computing sector.

Companies like Marathon and Hut 8 have not missed out on the AI windfall. They have not liquidated everything to purchase GPUs but have retained some mining capacity and Bitcoin inventory, treating AI as a diversification strategy rather than betting all their capital on an immature sector that will still undergo price revaluation. As long as ASIC miners are still operational, when mining returns become higher and AI computing premiums diminish, they can shift their load back to Bitcoin mining.

From a broader perspective, these companies are essentially power operators, with the hardware being either ASIC miners or H100 GPUs. Companies that can flexibly allocate power resources between mining and AI computing based on which business is more profitable will not be completely at the mercy of market cycles and narratives.