Author: Nancy, PANews

As crypto assets continue to weaken and decline, crypto mining companies are facing increasingly severe survival pressure. In search of new growth curves, more and more miners are accelerating their pivot into the AI sector. This transformation narrative has quickly gained favor in the capital markets, with many miners' stock prices surging significantly, even hitting record highs.

However, while AI businesses inject new growth imagination into miners, the massive capital expenditures, continuous funding requirements, and long payback cycles behind them are pushing miners into another war of capital attrition. At a time when the profitability of mining operations remains under pressure, this high-stakes bet on AI transformation is testing the financial strength and execution capabilities of mining companies.

Stock Prices Significantly Outperform Bitcoin, Miner Valuations Enter a Phase of Divergence

Miners are transforming into the compute landlords of the AI era.

As Bitcoin mining profit margins continue to narrow, with some miners even falling into losses, the AI explosion is driving a sharp increase in global demand for data centers, power resources, and GPU computing power. More and more miners are accelerating their transformation into the AI infrastructure sector, seeking new growth curves.

For miners, this transformation has inherent advantages. Over the long term, to meet large-scale mining demands, miners have already secured key assets such as abundant power resources, land reserves, substation access capabilities, and mature cooling and heat dissipation systems. Compared to data center operators starting from scratch, miners only need to upgrade and retrofit existing facilities to quickly enter the AI infrastructure market, capturing AI computing demand at lower costs and with shorter cycles.

Since last year, the pace of miners' transformation to AI has accelerated significantly. Some miners have decisively scaled down or even exited traditional mining operations to fully pivot to AI computing and data center operations; others have retained some mining machine businesses but are gradually shifting resource allocation and capital expenditure focus toward the AI sector. Today, several miners have grown into important participants in AI infrastructure construction.

In terms of transformation timing, CoreWeave, Applied Digital, and Bitdeer began laying out AI computing and data center businesses as early as 2022 to 2023, making them early movers in the industry; while miners like Iris Energy, Terawulf, Hut 8, Riot Platforms, and Bitfarms started comprehensively ramping up AI infrastructure construction in 2025, coinciding with the AI industry's entry into a rapid expansion cycle.

From a stock price performance perspective, the market has given high recognition to miners' AI transformation narrative. The average year-to-date increase for 11 miners reached 75.97%, significantly outperforming Bitcoin over the same period, with most stock prices even hitting new highs after the transformation. Among them, Bitfarms (129.62%), Hut 8 (131.87%), Terawulf (118.68%), and Riot Platforms (93.71%) performed particularly prominently, being beneficiaries of this round of AI infrastructure revaluation.

In terms of market capitalization, miners have already shown clear divergence. As a successful example of transformation, CoreWeave's market cap has reached $62.855 billion, far exceeding other miners and becoming a new valuation benchmark for the industry; Iris Energy, Terawulf, Hut 8, Applied Digital, and Riot Platforms form a market cap tier of $10 billion to $20 billion; companies like MARA Holdings, Core Scientific, Bitdeer, CleanSpark, and Bitfarms remain in the sub-$5 billion range. This divergence stems not only from first-mover advantages but also from the market's differentiated pricing of each miner's AI strategy execution capability, customer resources, and data center deployment progress.

However, from a fundamental perspective, most miners are still in the heavy investment phase of AI transformation. Although the latest quarterly reports for many miners show revenue growth, overall profitability remains under pressure. On one hand, value fluctuations in crypto asset portfolios drag down profit performance; on the other hand, AI data center construction requires massive capital expenditures, with increasing investments in power capacity expansion, infrastructure construction, and equipment procurement like GPUs, driving operating costs continuously higher and leaving most miners yet to escape losses.

It is worth noting that despite generally pressured performance, the stock prices of these miners have still achieved substantial gains, meaning the current market focus is not on short-term profitability, but on the growth potential of miners as next-generation computing infrastructure operators.

Miners' Survival Battle Upgrades, AI Transformation Still Faces Multiple Hurdles

The downturn in the Bitcoin market is making the survival environment for miners increasingly severe.

Data from Capriole Investments shows that as of June 18, the average production cost of Bitcoin was approximately $63,707, with electricity costs around $50,965, leaving a miner profit margin of only 17.45%. Over the past 30 days, miner profit margins have contracted by 47.8%. Meanwhile, data from the Luxor Hashrate Index also shows that as of June 18, the daily return per 1 TH/s of hashrate fell to $0.032, a significant drop from $0.053 in the same period last year.

Amid continuously shrinking mining revenues, many miners are forced to sell Bitcoin to maintain cash flow, further intensifying survival pressure on small and medium-sized miners, with mining resources accelerating concentration towards top players. Currently, the three major mining pools—Foundry USA, AntPool, and F2Pool—collectively hold 59% of the total network hashrate market share. In comparison, the top three Bitcoin mining pools held only 44% of the hashrate market share in 2022.

Although the traditional mining business is sluggish, the explosive growth in demand for AI data centers is also driving the market to reassess the value of miners. VanEck pointed out in its latest research report that the most valuable assets of miners are not the mining machines themselves, but power resources, substation access capabilities, land reserves, and data center infrastructure—precisely the scarcest core resources in the current AI industry. Since AI customers are willing to pay much higher electricity rates and rents than traditional mining operations, AI infrastructure is expected to become the main growth engine for miners over the next decade.

According to a report disclosed by research firm Bernstein, hyperscale cloud providers, AI cloud service providers, and chip companies have announced over $90 billion in AI infrastructure collaborations, involving approximately 3.7GW of power capacity. Currently, the pursuit of power resources is becoming the core of AI infrastructure competition, with Bitcoin miners collectively controlling over 27GW of planned power capacity. In some parts of the United States, the cycle for connecting 1GW of new power can take up to 50 months, making existing mining farms important landing points for AI data center expansion.

However, the AI transformation is far from an easy path. VanEck stated that the current market is still in the early stages of AI transformation, with corporate valuations primarily measured by Gross Energized Power. Miners that have signed AI leases generally receive higher valuation premiums, while projects still in the planning stage struggle to gain market recognition. In the future, the industry's valuation logic will gradually shift from "power capacity" to "project delivery capability," ultimately returning to core metrics such as cash flow, return on capital, and tenant quality. Currently, the industry has only completed about 25% of signed capacity delivery; whether AI data center construction can be completed on time and on budget will become a key factor determining corporate valuations.

VanEck also emphasized that the quality of AI tenants will directly impact the valuation levels of miners. Hyperscale cloud provider customers like Microsoft, Amazon, and Google can bring more stable cash flows and lower financing costs, while smaller GPU cloud service providers correspond to higher operational risks and capital costs.

The massive capital investment required for the transformation is also testing the financial strength of miners. VanEck estimates that miners still face significant capital expenditure needs for the AI infrastructure transformation, with a short-term financing gap of approximately $50 billion, and long-term capital needs potentially reaching $221 billion.

Under immense financial pressure, many miners have already begun raising funds through various means. For instance, miners like Iris Energy, TeraWulf, Bitfarms, and CleanSpark have raised funds by issuing convertible bonds, attracting investors with low coupon rates and future conversion potential; while companies like Core Scientific, Terawulf, MARA, Bitdeer, and Riot Platforms have chosen to sell or even liquidate portions of their Bitcoin reserves to continuously fund the AI transformation.

Additionally, many miners have started locking in future revenue by signing long-term AI or high-performance computing (HPC) contracts, thereby securing project financing support and reducing overall operational risk. For example, CoreWeave reached a $6 billion AI cloud service cooperation agreement with Jane Street; IREN secured a $9.7 billion AI cloud computing contract with Microsoft; Hut 8 signed data center lease agreements totaling $9.8 billion; and Bitdeer partnered with Norway's DCI to build the country's largest AI data center project, among others.

For miners, AI undoubtedly offers a development path with far more imaginative potential than traditional mining operations at this stage. However, this transformation is not simply a switch from mining to selling computing power; it is essentially a long-term competition revolving around capital, resources, and execution capability.