Bitcoin Mining Market Review 2025: Four Survival Models and Undervalued Arbitrage Opportunities

In 2026, the evaluation system for the Bitcoin mining industry is undergoing a dramatic transformation. Merely "hash rate scale" is no longer the sole determinant of market value. The capital market is now seeking two types of targets: one is giants like Core Scientific that have successfully transitioned into AI; the other is miners like CleanSpark that relentlessly pursue operational efficiency.

Beyond these dominant, well-known players, Cang represents a third, atypical model of survival. This article will incorporate the latest report from Digital Mining Solutions, the "2025 Bitcoin Mining Market Review" (hereinafter referred to as the "2025 Annual Report"), and conduct an in-depth horizontal comparison across the same dimensions among three representative companies—MARA, CLSK, and CORZ—alongside CANG. This analysis will dissect the differences in their underlying business logic and the significant discrepancies in their valuation.

In terms of the cost of acquiring computing power (CAPEX), the industry has diverged into two distinctly different approaches.

1. Heavy Asset School: CleanSpark & Riot Platforms

• Strategy Core: "Buy new instead of old, build instead of lease." CLSK and Riot prefer to invest heavily in constructing large-scale mining farms and purchasing the latest S21 or XP series mining machines.
• Advantages: Extremely high energy efficiency (low J/TH), long-term operational stability, and widely favored by institutional investors.
• Disadvantages: Extremely high capital expenditures (CAPEX). The cost of new machines is typically $15–$25 per TH, which prolongs the return on investment (ROI). Once cryptocurrency prices stagnate, the significant depreciation burden can erode profits.

2. Value Arbitrage School: Cangoo

• Strategy Core: "Supply Chain Leverage, Ultra-Low Cost Expansion."

Low-cost inventory building: Canoo did not blindly chase highly premium new mining machines. The 2025 annual report specifically highlighted that Canoo quickly entered the first tier of the industry by acquiring used in-service mining machines, acquiring computing power at an extremely low cost (approximately $8/TH), thereby creating a natural safety cushion compared to its peers.

• Dynamic Upgrade: Canoe does not always use old machines, but adopts a "Refresh" strategy. For example, in Q4, it upgraded 6 EH/s of computing power to S21 machines by paying only the price difference.
• Comparative conclusion: CLSK excels in efficiency, while Cangu stands out in terms of cost-effectiveness per unit of computing power. In the current competitive environment, Cangu's unique "light-asset" model provides higher risk resilience.

As mining rewards are halved, AI has become a fiercely contested arena for mining companies. Data from 2025 annual reports shows that companies with clear AI/HPC revenue significantly outperformed pure-play mining firms in stock price. However, in this sector, Cangoo has chosen a completely different approach compared to the giants.

1. Core Scientific: International Freight Hub

● Business Model: CORZ has secured a major contract with CoreWeave and is committed to building Tier 3/4 ultra-large data centers.

This is similar to building a massive "international airport cargo terminal." It is specifically designed to serve "mega containers" (large-scale model training tasks), with an astonishing throughput. However, such infrastructure requires extremely high standards for runways and control towers (dual power supply, ultra-high redundancy), with a construction period spanning several years and a funding threshold in the billions. Once built, it becomes an irreplaceable core hub.

● Valuation Logic: The market views it as "core infrastructure" and has given it a very high premium because it controls irreplaceable scarce resources.

2. Cangoo: Same-city Express Delivery Network

● Business Model: Canoo avoids the "airport" red ocean of infrastructure development and instead focuses on the long tail of "AI inference." By implementing low-cost, ultra-fast AI transformation, it converts scattered mining sites around the globe into a decentralized AI network.

It's like building a dense "local pre-storage warehouse" or "ultra-fast delivery network."

○ Flexible Retrofit: There is no need to build expensive airport runways. Instead, simple power and network upgrades can be made using existing community facilities (mining sites) to achieve plug-and-play functionality.

• Handling small packages: It does not transport large containers, but instead specializes in processing a massive volume of "small packages" (inference requests from small and medium-sized enterprises, real-time response tasks) with high frequency and speed.

● Smart Scheduling: Through a middleware platform, scattered computing power around the globe can be scheduled like dispatching couriers, forming a distributed network.

● Comparative Conclusion: CORZ focuses on "heavy infrastructure," earning high toll fees, but it is too heavy and slow. In contrast, Canguo focuses on "last-mile logistics," earning profits through high-frequency turnover and flexibility. On the eve of the explosive growth in AI inference demand, Canguo's "express delivery"-style model, which can rapidly respond to the needs of small and medium-sized enterprises, has seized the advantages of "speed" and "breadth."

If fundamentals determine the long-term floor, then market perception biases determine short-term explosive power. The institutional holdings data disclosed in the 2025 annual reports reveal the most astonishing contrasts.

1. The Darling of Wall Street: MARA & IREN

● Current situation: "Congested trading."

○ Data: MARA has 484 institutional holders, IREN has 433.

○ Logic: Almost all funds that focus on crypto or AI have already allocated significant positions in these leading stocks. Their information is fully transparent, and expectations have already been largely reflected in the stock prices (priced-in). To drive stock prices higher, exceptionally impressive performance that exceeds expectations is required.

2. The Forgotten Corner: Can Valley

● Current situation: "A desert of cognition."

○ Data: Although its computing power has ranked among the top five globally (50 EH/s), Cangoo has only 35 institutional holders.

○ Logic: This means that the majority of institutional funds have not yet invested in this company. Such extreme information asymmetry is often the source of excess returns.

○ Catalyst: As Cangull AI business begins to be disclosed, or its Bitcoin holdings are reevaluated, even a small amount of institutional capital starting to "fill the gap" could have a significant marginal pricing impact on such a low-liquidity, undervalued asset.

○ Conclusion: Investing in MARA follows the consensus and captures the industry's beta return; investing in Cangull anticipates a recovery in perception and seeks alpha returns driven by institutional inflows.

Finally, we return to the most fundamental valuation logic.

In-depth Review:

The current pricing of CORZ and MARA includes a very high "expectation premium." However, Canguru's market capitalization is even lower than its "hard assets" on the balance sheet (BTC holdings + mining machines + cash - debt).

This means that the market has not yet assigned any positive valuation to Cangull's 50 EH/s computing power or its Tier 2 AI transformation.

Investment Advice:

● If you believe in the capital concentration effect, pursue absolute liquidity and industry beta, MARA is still the top choice.

● If you are bullish on the deterministic surge in the AI training segment and are not concerned about high valuation, CORZ is the purest play.

● However, if you are a value investor seeking high-reward opportunities with downside asset support and upside AI transformation potential, Cangku clearly offers an extremely attractive entry point under the current pricing system.