The Great Migration of Mining Companies: Some Already Hold $12.8 Billion in AI Orders

Odaily星球日报Published on 2026-03-06Last updated on 2026-03-06

Abstract

Driven by rising Bitcoin mining difficulty, shrinking profits, and extreme capital pursuit of AI, major crypto mining firms are collectively pivoting towards AI and high-performance computing (HPC). They leverage existing assets—power, land, cooling systems, and data centers—to secure massive AI compute orders. Key catalysts include Bitcoin’s hashprice dropping to multi-year lows, pushing firms like Bitdeer to sell all BTC holdings and NFN8 into bankruptcy. Conversely, companies are securing enormous contracts: TeraWulf holds a $12.8B HPC deal, IrisEnergy a $9.7B order with Microsoft, and Hut8 a $7B agreement. Others like CipherDigital ($5.5B with AWS) and CoreScientific ($10B+ with CoreWeave) are rebranding and expanding data centers. Q1 2026 earnings show AI/HPC revenue now constitutes 40%-60% for some. While firms like Marathon Digital remain committed to Bitcoin long-term, others like BitDigital have fully exited BTC mining for ETH and AI. Market sentiment is volatile but selective, with capital backing execution-ready firms with strong power contracts and scalability. The migration is structural: adapt to AI or risk obsolescence.

Original|Odaily Planet Daily(@OdailyChina)

Author|Wenser(@wenser 2010)

Over the past decade, Bitcoin mining companies were once the most stable foundation of PoW networks and the cost anchor of the BTC "Tier 0 market." But now, these industry cornerstones are collectively pivoting, either actively or passively, towards AI.

On the surface, the direct incentive for mining companies to transform is the continuous increase in mining difficulty and the compression of profit margins by a sluggish market; but the deeper driving force is the extreme pursuit of the AI narrative by capital markets—and mining companies happen to possess the most easily convertible real-world assets: electricity, land, cooling systems, data centers, and ready-made data infrastructure, which can be exchanged for AI computing power orders worth tens of billions of dollars.

Amid the clamor of multi-model competition, mining companies standing at the intersection of energy, electricity, computing power, and crypto assets are undergoing an unprecedented yet almost inevitable industry migration.

Some are steady and cautious, waiting and watching; others are forced to turn around, making desperate gambles. But one thing is certain: the wind has risen—this is a structural shift blowing from the crypto market towards the AI world.

A Hard Battle That Must Be Fought, and a Cake Too Tempting to Refuse

Entering 2026, for mining companies, the real pressure has never come solely from price fluctuations, but from structural squeeze: continuously rising difficulty, continuously declining unit revenue, and continuously increasing operating costs.

In the Winter: Selling Coins to Survive and Bankruptcy Liquidations

On February 20, Bitcoin mining difficulty once increased by 15% to 144.4T, the largest increase since 2021. During the same period, the network hash rate rebounded from 826 EH/s to 1 ZH/s, but the hashprice fell to a multi-year low of only about $23.9/PH/s. Under the continued profit compression brought by the 2024 halving, mining companies were forced into a cash flow defense mode.

The most symbolic event came from Bitdeer. On February 20, it disclosed that its own BTC holdings had dropped to 0, with that week's production and sales being exactly equal. Although founder Jihan Wu later explained that "being 0 now does not mean being 0 in the future," the market still viewed it as a microcosm of the pressure on mining companies.

The困境 isn't limited to one company. In early February, NFN8 Group filed for Chapter 11 bankruptcy protection in Texas, USA, planning to sell all its assets. Documents showed that a core mine fire, the lease burden from sale-leaseback models, and the cliff-like drop in hashprice after the halving directly crushed its cash flow. Despite owning multiple mining facilities, NFN8's own assets of 5000 mining machines were valued at less than $50,000, while its debts reached the million-dollar range.

As the environment continues to deteriorate, mining companies' reaction has been unusually consistent—heading towards AI.

Second Spring: Astonishing Profits Behind AI/HPC Mega Orders

For AI giants, computing power data centers are always scarce: traditional construction cycles take 3-5 years, with high costs for land, electricity, and cooling. Mining companies already have power contracts, infrastructure, and operational experience, making them the most realistic承接方 (contractors) in the AI expansion cycle.

Since last year, mining companies have seen a集中爆发 (concentrated explosion) of orders. According to public data statistics, as of the time of writing, 6 mining companies including IREN, CIFR, and HUT have accumulated AI/HPC orders worth approximately $38.5 billion. Among them, TeraWulf's $12.8 billion order contract with Fluidstack and IREN's 5-year $9.7 billion contract with Microsoft are惊人 (astounding) figures and have become important supports for their stock prices. From financial reports, the proportion of AI/HPC revenue for many mining companies has increased from less than 15% to 40%–60%.

If mining is a cyclical business, AI is like a long-term cash flow pipeline.

Earnings Consensus: AI Becomes the Keyword

The Q1 2026 earnings season almost gave a unanimous signal: mining companies are systematically transforming.

“HPC Contract Giant” WULF: Holds Over $12.8 Billion in Contracts

Mining company TeraWulf's full-year 2025 revenue reached $168.5 million, a year-on-year increase of 20.3%, of which $16.9 million came from the newly launched high-performance computing (HPC) leasing business.

TeraWulf currently holds over $12.8 billion in HPC contracts, with 522MW of capacity already signed, and has obtained $6.5 billion in financing to support data center expansion.

“AI Mining Mini-Giant” IREN: Holds $9.7 Billion Microsoft Order

Thanks to previous huge orders and rapid transformation, IREN has隐隐 (seemingly) become a new generation of "AI mining mini-giant".

According to the financial report of mining company IrisEnergy (IREN), as of January 31, 2026, it held $2.8 billion in cash and cash equivalents. So far this fiscal year, it has obtained over $9.2 billion in funds through customer prepayments, convertible bonds, GPU leasing, and GPU financing. Subsequent plans include adding 140,000 GPUs, expecting to achieve $3.4 billion in annual recurring revenue by the end of 2026.

“The Trump Family's” HUT: Holds $7 Billion in Orders

Mining company Hut8's 2025 fiscal year generated $9.6 million in revenue from hosting services and held approximately $1.4 billion in cash and Bitcoin reserves.

Additionally, Hut8's spun-off mining subsidiary AmericanBitcoin (ABTC) achieved full-year 2025 revenue of $185.2 million, deployed算力 (hash rate) of about 25 EH/s, and owns about 78,000 ASIC miners. Furthermore, its BTC reserves have exceeded 6,000 coins.

The company is also a major crypto mining company supported by the Trump family, thus receiving high market attention.

“Brand Transformation Complete” CIFR: Holds $5.5 Billion in Orders

Mining company CipherDigital disclosed in its 2025 fiscal year performance report that it officially changed its name from "CipherMining" to "CipherDigital" to complete its brand transformation.

Last November, CIFR reached a leasing agreement with Amazon Web Services worth up to $5.5 billion; additionally, it exchanged 5.4% equity for Google's agreement to guarantee $1.4 billion in the contract signed with Fluidstack.

“Selling Coins to Buy Land and Build Data Centers” RIOT: Reaches Leasing Cooperation with AMD

Mining company RiotPlatforms announced its full-year 2025 performance, with annual revenue reaching $647.4 million, a significant increase from $376.7 million in 2024; its Bitcoin holdings exceed 18,000 coins.

In January of this year, RIOT sold 1,080 Bitcoins and used the proceeds (about $96 million) to purchase the Rockdale plot for data center project development. Furthermore, the company signed a data center leasing and service agreement with AMD, which will deploy 25 megawatts of critical IT load capacity at the Rockdale campus. The aggressive investment firm StarboardValue stated that Riot's potential valuation from transitioning towards AI and HPC could be as high as $21 billion.

“BTC Stalwart” MARA: Partners with Capital Institutions to Layout AI Data Centers

MARA's financial report data shows that affected by a roughly 14% drop in the average Bitcoin mining price, MARA's Q4 2025 revenue was $202.3 million, a year-on-year decrease of about 6%. At the end of February, MARA announced a cooperation with investment institution StarwoodCapitalGroup to build large-scale data centers for artificial intelligence and cloud computing customers on the basis of existing mines in the United States. After the news was announced, its stock price rose about 17% in after-hours trading.

It is worth mentioning that unlike other mining companies firmly transforming into the AI field, MARA's management emphasized that despite short-term price fluctuations, its long-term confidence in the Bitcoin asset class has not changed, Bitcoin will still be the long-term strategic core

“Data Center Revenue Soars” CORZ: Holds Over $10 Billion CoreWeave Order

CoreScientific (CORZ) announced its Q4 2025 financial report. Total revenue for Q4 2025 was $79.8 million, down from $94.9 million in the same period last year. Among them, Bitcoin mining revenue dropped to $42.2 million; data center hosting service revenue大幅增长 (increased significantly) to $31.3 million, higher than $8.5 million in 2024. Q4 gross profit rose to $20.8 million, higher than $4.8 million in the same period in 2024.

CoreScientific CEO Adam Sullivan stated that the company's existing construction projects are more than halfway completed and are expanding the hosting platform to a 1.5 gigawatt leaseable capacity pipeline. Last October, AI company CoreWeave planned to acquire CoreScientific at a valuation of about $9 billion, but it最终作罢 (ultimately fell through) due to lack of shareholder approval; in January of this year, CoreScientific sold 1900 BTC (approximately $175 million) for business transformation.

The company estimates that AI business will drive revenue compound growth of 60.9% from 2026-2028, reaching $1.5 billion by 2028.

Other Mining Company Representatives: Bitfarms Renames, BitDigital Switches to ETH Camp

In February, Bitfarms (BITF) announced that it would relocate its headquarters from Canada to the United States and plans to change its name to KeelInfrastructure (pending shareholder, exchange, and court approval), accelerating its transformation into infrastructure. Previously, the company had converted $300 million in debt financing into project financing in October last year for the construction of a data center in Pennsylvania, and sold the Paso Pe mine for $30 million in January of this year, officially exiting the Latin American market.

On the other hand, BitDigital's turn is more thorough. As early as last July when the DAT (Odaily Note: Digital Asset Treasury company)热潮 (boom) emerged, it率先宣布 (was the first to announce) switching from BTC to ETH treasury listed companies; in January this year, it further clarified that it would completely stop Bitcoin mining and instead increase its efforts in Ethereum infrastructure, staking, and HPC/AI strategy, marking the official completion of the阵营切换 (camp switch) for this mining company that has been deeply involved in mining for five years. Currently, its AI subsidiary WhiteFiber has completed an IPO, and BitDigital holds about 27 million shares, worth over $457 million at the current market value.

In addition to the above two, Galaxy, Bitdeer, Cleanspark, Cango Cango, etc., are still in the stage of promoting AI transformation, and the proportion of revenue contribution remains to be increased. Among them, Cango Cango completed a $10.5 million equity financing in February this year and received an additional $65 million investment commitment, which may accelerate the layout of AI/HPC data center business.

The following is a brief comparison based on public information for reference.

Capital Attitude: Choosing Winners, Not Narratives

The market does not fully accept the "AI transformation" but is quickly differentiating.

In early February, JPMorgan pointed out in a report that Bitcoin mining companies performed strongly at the beginning of the year, mainly driven by the phased缓解 (easing) of network competition and the升温 (heating up) of the HPC narrative. At that time, the total market capitalization of the 14 US-listed mining companies and data center operators it tracked rose to about $60 billion at the end of January, a month-on-month increase of 23%, far exceeding the S&P 500's gain of about 1% during the same period.

But soon, with the密集发布 (dense release) of a new round of AI models and the impact of OpenClaw on the valuation system of software stocks, market sentiment quickly turned. Capital began to worry about the structural disruption brought by AI, and the stock prices of mining companies related to AI infrastructure subsequently corrected, with CIFR, IREN, and Hut8's intraday declines once exceeding 10%.

On February 10, Morgan Stanley released a research report, giving CIFR and WULF an overweight rating, while downgrading MARA to underweight.

By the end of February, with order fulfillment and stock price recovery, the market风向 (wind direction) reversed again. Some analysts believe that against the background of high short-selling ratios by hedge funds, coupled with mining companies locking in long-term low-cost power contracts, their strategic value is no longer limited to traditional mining but is closer to that of AI infrastructure suppliers.

With order fulfillment and stock price recovery, the market logic has gradually become clear: capital only bets on structural winners.

Therefore, the future of mining companies largely depends on three things:

Execution: Whether the migration of computing power form can be completed quickly;

Resource Endowment: Whether electricity and land have scale advantages;

Narrative Ability: Whether it can be embedded in the upstream AI supply chain.

In fact, the company's transformation decision is not important; what is more important is capital screening.

The tide has come. Mining companies only have two choices: either migrate with the trend or become history.

Trending Cryptos

Related Questions

QWhat is the main reason for Bitcoin mining companies to shift towards AI, according to the article?

AThe main reason is the structural squeeze from increasing mining difficulty, declining unit revenue, and rising operational costs, combined with the extreme pursuit of AI narratives by capital markets. Mining companies possess easily convertible real assets like electricity, land, cooling systems, and data infrastructure, which can secure AI computing orders worth billions of dollars.

QWhich mining company holds the largest AI/HPC order mentioned in the article, and what is its value?

ATeraWulf (WULF) holds the largest mentioned AI/HPC order, valued at over $12.8 billion, signed with Fluidstack.

QHow did the market react to Hut8's (HUT) association with a prominent political family?

AThe article states that Hut8's spun-off mining subsidiary, American Bitcoin (ABTC), received high market attention because it is a crypto mining company supported by the Trump family.

QWhat significant action did Riot Platforms (RIOT) take to fund its data center development?

ARiot Platforms sold 1,080 Bitcoin (worth approximately $96 million) in January and used the proceeds to purchase the Rockdale land plot for developing its data center project.

QAccording to the article, what are the three key factors that will determine the future of mining companies in their AI transition?

AThe three key factors are: Execution capability (the ability to quickly complete the computing power migration), Resource endowment (whether their electricity and land provide a scale advantage), and Narrative ability (the ability to embed themselves into the upstream AI supply chain).

Related Reads

Dan Koe: The Counterintuitive Truth—You Don't Need to Remember Everything You Read

Dan Koe: The Counterintuitive Truth — You Don't Need to Remember Everything You Read The central idea is that deliberately trying to remember information is often misguided. True learning isn't about memorizing facts but about having important knowledge surface naturally when needed through use. Most forgetting is normal, not a failure. The article reframes learning using a control theory framework—a four-step feedback loop: having a clear Goal, accurately Sensing your current state, Comparing the gap, and Acting to close it. Most learning stalls because people only do step 2 (blind input) without a goal to create the necessary "error signal" for focused action. The most effective method is to start with output, not input. Begin a meaningful personal project first, and learn only what's necessary to complete it. This project-driven, "just-in-time" learning ensures knowledge is contextual and retained. The concept of a "Second Brain" often fails because it becomes a digital graveyard—over-collected and under-utilized. The goal should be building a "Second Subconscious"—a dynamic system that proactively surfaces relevant ideas during creation, not a static storage vault. Tools like Obsidian+Claude or Eden can help by automating organization and enabling semantic search, but their value depends on linking knowledge to active projects. Ultimately, what matters is not what you store, but what you filter and internalize. Focus on ideas that shape your worldview, use projects as filters, and transform collected material through writing and sharing. AI should be used to reduce friction in research and editing, not to formulate your core views. In conclusion, remembering is a byproduct, not the goal. Knowledge that sticks comes from pursuing personal goals, applying it in real projects, and digesting it through creation. The tools are merely aids; the crucial step is to start doing meaningful work and let the necessary knowledge find you.

marsbit52m ago

Dan Koe: The Counterintuitive Truth—You Don't Need to Remember Everything You Read

marsbit52m ago

A New Era: The Fundamental Transformation of China's Entrepreneurs

A profound generational shift is underway among Chinese entrepreneurs. The wealth and influence once dominated by real estate and internet giants is now being claimed by a new wave of founders driving breakthroughs in AI, semiconductors, and robotics. This change is vividly reflected in 2026's wealth rankings. Figures like Zhang Yiming (ByteDance), Liang Wenfeng (DeepSeek), Chen Tianshi (Cambricon), and Wang Xingxing (Unitree Robotics) are ascending. Their wealth stems not from traditional business models but from market expectations for future technological competitiveness, with AI, chips, and smart hardware becoming the primary engines of wealth creation. Their common trait is a foundational focus on technology, often starting from the laboratory rather than a business plan. Examples include Chen Tianshi's decade-long push in AI chips, Liang Wenfeng's core algorithmic innovations at DeepSeek with a compact team, and Zhu Yiming's "no salary until profitable" 9-year journey to build Changxin Memory into a global DRAM player. This transition marks a fundamental shift in China's economic imperative: from commercial expansion and learning to indigenous innovation and deep industrial capability. While the previous generation built the foundational market and infrastructure, this new cohort is tasked with achieving global leadership in core technologies, moving China from "keeping pace" to pioneering original, breakthrough innovations that are industrialized at scale. The baton is being passed to those competing on the world stage through technological originality.

marsbit52m ago

A New Era: The Fundamental Transformation of China's Entrepreneurs

marsbit52m ago

Once-Popular Web3 Enters Wave of Layoffs

The once-hot Web3 industry is experiencing a severe wave of layoffs. While many companies attribute job cuts to AI-driven restructuring, the primary reason is often financial pressure. The Web3 sector, at the intersection of tech and finance, has been hit particularly hard. Employees at major cryptocurrency exchanges report sudden, impersonal layoffs—often with system access revoked overnight—and minimal or no severance. Common tactics include setting impossible performance targets or terminating employees for minor policy violations. The working atmosphere has become toxic, marked by intense monitoring, excessive meetings, and management obsessed with control and internal politics rather than product innovation. The industry's core business model is collapsing. Exchange revenue from trading fees and listing charges has plummeted due to a decline in quality projects and retail investor exodus. Events like the massive forced liquidation on October 10th further shattered confidence. Competition from on-chain derivatives platforms and prediction markets is intensifying the downturn. As layoffs continue, displaced workers struggle to find new opportunities. Many transition to the AI sector, but face significant bias from traditional finance and even some AI firms, which view crypto industry experience with suspicion. The current downturn appears more structural than cyclical, driven by unsustainable practices, internal strife, and a failure to innovate, raising questions about the industry's future trajectory.

marsbit1h ago

Once-Popular Web3 Enters Wave of Layoffs

marsbit1h ago

Trading

Spot

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片