Why Are Mining Companies Pivoting to AI? Why MARA and Bitdeer Are Moving Away from Bitcoin Mining

Markets
Updated: 2026-03-30 05:38

In the first quarter of 2026, the core narrative of the crypto industry is quietly but profoundly shifting. Publicly traded mining companies—once hailed as the "bedrock of the Bitcoin ecosystem"—are now collectively steering their capital and computing power away from the singular track of Bitcoin (BTC) mining and making large-scale moves into the field of artificial intelligence (AI) cloud computing. Leading players such as MARA Holdings (formerly Marathon Digital), Riot Platforms, Bitdeer Technologies Group, and Iris Energy (IREN) are restructuring their business portfolios through two main strategies: "GPU computing power substitution" and "selling BTC reserves." This is not only a strategic self-rescue at the corporate level but may also signal a reshuffling of power across crypto mining and the broader digital asset industry. This article dissects the phenomenon using facts and data, offering a structured analysis and trend projection.

A Collective Shift Driven by Cost and Profitability

Starting in the second half of 2025, several North American publicly listed Bitcoin mining companies—led by MARA Holdings, Riot Platforms, Bitdeer, and IREN—began systematically pivoting their business focus from traditional Bitcoin mining to high-performance computing (HPC) and AI cloud services. This transformation is evident in two key actions: first, large-scale procurement of GPU chips from vendors like NVIDIA to deploy AI computing centers, gradually replacing the original ASIC miner infrastructure; second, selling Bitcoin reserves from their balance sheets to raise capital for this transition. Notably, MARA Holdings publicly disclosed in the first quarter of 2026 that it had sold 15,133 BTC to support its data center’s strategic shift toward AI operations.

From Hydropower Windfalls to Two Years of "Inverted Cost Structures"

2024: The Halving and the Cost Divide

After Bitcoin’s fourth halving, the block reward dropped to 3.125 BTC. At the same time, total network hash rate continued to climb, and with fluctuating energy costs, the average cost of Bitcoin mining began to approach—and at times exceed—the market price. Industry data shows that by the end of 2024, some miners’ "all-in costs" (including electricity, equipment depreciation, and operational management) had reached nearly $40,000.

2025: The AI Boom and Mining Facility Revaluation

With the explosive growth of generative AI, demand for computing power soared exponentially. Mining companies’ existing data center infrastructure—often with access to cheap electricity—became highly attractive assets for AI companies. Miners realized that their core advantages—location, power resources, and operational expertise—were equally applicable to AI computing. That year, some miners began pilot GPU computing rental businesses, testing both technology and market demand in preparation for a larger-scale pivot.

2026 and Beyond: Accelerated Strategy and Capital Restructuring

By 2026, the "inverted cost structure" in Bitcoin mining had intensified. As of March 30, Bitcoin traded around $67,534.5, but according to operational data disclosed by several mining firms, their average mining costs (including electricity, equipment, and new capacity investments) were estimated between $75,000 and $80,000. This made selling BTC for cash and investing in high-margin AI businesses a rational business decision. In February 2026, MARA announced that its annualized revenue from AI operations had surpassed $200 million and stated its continued commitment to an "HPC-first" strategy, sparking a chain reaction across the industry.

Computing Power Migration and Capital Reallocation

Based on publicly available financial reports and strategic documents, we can quantify miners’ transformation along two core dimensions:

Transformation Dimension Specific Action Representative Cases & Data (as of March 2026)
Computing Power Structure Shift from ASIC miner hash power to GPU/HPC computing power MARA: Advancing AI/HPC infrastructure, emphasizing expansion from Bitcoin mining to AI computing; IREN: Launched GPU expansion for AI/HPC infrastructure.
Asset Structure Selling BTC reserves to raise cash or invest in AI infrastructure MARA: Sold approximately 15,133 BTC to improve its balance sheet and invest in AI infrastructure; currently holds over 38,000 BTC. Riot: Evaluating use of remaining power capacity for AI/HPC data center construction.

This data clearly reveals the logic behind miners’ asset restructuring. On one hand, computing assets are migrating from single-purpose ASIC miners (used solely for SHA-256 mining) to general-purpose GPU clusters serving AI training and inference. This shift significantly increases asset liquidity and diversifies application scenarios. On the other hand, financial assets in the form of BTC reserves are being actively reduced. Traditionally, miners hoarded BTC as a "store of value" to hedge against inflation; now, BTC is seen as a "financing tool"—sold at market highs to fund AI ventures offering higher capital returns.

A Clash of Optimism and Skepticism

Debate over "miners pivoting to AI" currently centers on these main points:

  • Optimists: The Value Reassessment Argument
    • Core view: Miners’ shift to AI is a "dimensional upgrade." They already possess cost-effective data centers, power resources, and operational know-how. Migrating mining capabilities to AI computing is essentially an upgrade from "energy arbitrage" to "computing power services," promising more stable, higher-margin cash flow and recasting company valuations from "commodity producers" to "technology infrastructure providers."
  • Skeptics: Narrative Bubble and Execution Risk
    • Core view: The pivot is no easy feat. AI cloud services and Bitcoin mining differ significantly in technology stack, customer base, and business model. Miners face talent barriers, technical adaptation challenges, and fierce competition from established cloud giants like Amazon AWS and Microsoft Azure. Large-scale GPU purchases require massive capital outlays, and selling BTC could mean missing out on the next Bitcoin bull market—a risky bet of "trading core assets for uncertain returns."
  • Neutral Observers: Changing Supply and Demand Structure
    • Core view: Regardless of the outcome, miners’ collective pivot is already having a tangible impact on the crypto industry. Some miners exiting or reducing hash power relieves upward pressure on Bitcoin network hash rate, leaving more profit potential for remaining miners. At the same time, miners have become steady BTC market sellers, altering the previous supply-demand dynamics and exerting downward pressure on BTC prices.

Three Layers of Information to Distinguish

To ensure analytical rigor, we must clearly distinguish among three levels:

Facts:

  • MARA has sold 15,133 BTC.
  • Multiple publicly listed mining companies have announced large-scale GPU procurement plans.
  • In Q1 2026, some miners’ AI/HPC revenues have surpassed their mining revenues.

Opinions:

  • "Pivoting to AI is an inevitable trend for miners." — This is not an absolute truth but an industry judgment based on current economic models and cost curves. Miners still have choices; some focus on improving mining efficiency or sourcing cheaper power instead of a full-scale pivot.
  • "This will mark the end of the Bitcoin mining era." — This is an exaggerated claim. The global Bitcoin mining ecosystem remains vast; the pivot is mainly among North American public miners and does not mean all miners worldwide will follow suit.

Speculation:

  • "By 2027, AI will contribute over 80% of miners’ revenues." — This is a linear extrapolation based on current growth rates, without fully accounting for market competition, technological evolution, or potential dramatic swings in Bitcoin prices. If BTC prices rebound sharply, miners’ strategic priorities could shift again.

Industry Impact Assessment: From Miners to the Crypto Ecosystem

  • On Miners Themselves: Diversifying business helps smooth out revenue volatility from single-asset (BTC) price swings but brings new capital expenditure pressures and operational complexity. Successful pivots could earn higher valuation multiples in capital markets, while failed ones may find themselves "losing mining advantages without having built AI capabilities."
  • On the Bitcoin Network: In the short term, some miners’ hash power migration may slow overall network hash rate growth, or even cause minor fluctuations. In the long run, mining difficulty will self-adjust. More profoundly, the network may lose some of its most active "developer-miner" hybrid participants, making future network evolution more reliant on the community and core developers.
  • On the AI Computing Market: Crypto miners’ entry adds significant new supply to the AI computing market, especially in mid- and lower-end computing rentals, which could ease supply-demand tensions. This may lower AI computing rental prices, benefiting small and midsize AI development teams.
  • On Market Supply and Demand: Miners shifting from "BTC hoarders" to "BTC sellers" make them a steady selling force in the market. This challenges BTC’s long-term supply-demand balance—especially when new capital inflows are lacking, miners’ ongoing selling could become a key headwind for price appreciation.

Scenario Analysis: Multiple Evolutionary Paths

Based on current information, we can outline three main evolutionary scenarios:

  • Scenario 1: AI Business Succeeds, Miners Achieve Valuation Rebuild
    • Logic: Leading miners leverage their infrastructure and power cost advantages to successfully enter the AI computing services market and establish stable client relationships. AI business delivers steady, high-margin cash flow, and company stock prices and valuations align with tech sector peers. BTC mining becomes just one part of a diversified business.
  • Scenario 2: AI Bubble Bursts, Miners Return to Mining
    • Logic: AI computing demand grows slower than expected, or market competition drives rental prices down to unprofitable levels. Meanwhile, Bitcoin prices rebound sharply due to the halving cycle, again exceeding mining costs. Some unsuccessful pivots will have to scale back or divest AI operations and return to traditional mining.
  • Scenario 3: Industry Divergence, Dual Pathways Coexist
    • Logic: The industry won’t see a black-and-white outcome. Ultimately, well-capitalized, well-managed miners (like MARA) will achieve dual growth as "digital energy + computing service" platforms. Others will focus on mining efficiency or specialize in certain AI niches. The industry landscape will become more diverse and complex.

Conclusion

The collective pivot of mining companies from Bitcoin mining to AI is not a fleeting trend, but a structural choice shaped by macroeconomic pressures, industry competition, and technological paradigm shifts. MARA’s sale of 15,133 BTC to support its AI strategy is just one footnote in this sweeping narrative. It reflects the crypto industry’s evolution from wild growth to maturity, and from a single storyline to a tapestry of integrated narratives. For market participants, it is crucial to view the opportunities and challenges behind this transformation rationally, to distinguish among facts, opinions, and speculation, and to seek their own certainty amid the industry’s dynamic evolution. This computing power revolution—sparked by miners—may ultimately redefine crypto’s place in the future global tech landscape.

The content herein does not constitute any offer, solicitation, or recommendation. You should always seek independent professional advice before making any investment decisions. Please note that Gate may restrict or prohibit the use of all or a portion of the Services from Restricted Locations. For more information, please read the User Agreement
Like the Content