MARA Holdings, Inc. (MARA) Stock Analysis and Metrics

MARA Holdings

NASDAQ: MARA

$12.14 ▼ -0.05 (-0.41%)

At close: Jul 14, 2026 · 3:59 PM UTC

Financial Ratios

Market Cap5.08 Bn

P/E-2.49

P/S5.86

Div. Yield0.00

ROIC (Qtr)-0.01

Total Debt (Qtr)2.37 Bn

Revenue Growth (1y) (Qtr)-18.36

Add ratio to table…

About

MARA Holdings, Inc. is an energy and digital infrastructure company that acquires, manages, and allocates energy to its highest value uses, using Bitcoin mining as a flexible energy responsive workload to monetize excess and underutilized power while developing artificial intelligence inference and high performance computing capabilities across its global portfolio of approximately 1.9 gigawatts of capacity located in 18 data centers on four continents.

The company primarily generates revenue from Bitcoin mining, earning block subsidies and transaction fees by operating roughly 490,000 mining rigs that produced 8,799 bitcoin in 2025 and delivering a hashrate of 66.4 exahashes per second. It also monetizes its bitcoin holdings through lending, structured trading, collateralized financing and other bitcoin denominated transactions that generated about $32.1 million of interest income in 2025. It is building AI inference and high performance computing services to offer compute capacity and private cloud solutions to third party customers, though these activities remain early stage and have not yet produced material revenue.

The company operates through the following segments:

• Bitcoin Mining: This segment encompasses the ownership and operation of a global fleet of mining rigs, the management of approximately 53,822 bitcoin holdings as of year end 2025, and the provision of mining services that earn block rewards and transaction fees, while also optimizing power usage across owned and hosted sites to maximize the value of each megawatt managed.

• AI and High Performance Computing: This segment focuses on developing and operating AI inference and high performance computing infrastructure, leveraging the acquisition of a majority interest in Exaion, the strategic agreement with Starwood Digital Ventures to create a joint venture capable of delivering more than 1 GW of initial IT capacity (with potential to scale to 2.5 GW), and the integration of inference capabilities into existing data centers such as the Granbury, Texas site to offer dedicated compute and cloud solutions to enterprise and public sector clients.

MARA positions itself as one of the largest publicly traded bitcoin holders, with a bitcoin treasury valued at approximately $4.7 billion based on a spot price of $87,498 per bitcoin on December 31, 2025. It also describes itself as a leading energy responsive miner that controls roughly 70% of its total capacity, giving it greater operating control, margin durability and capital efficiency compared with peers that rely more heavily on third party hosting. Its competitive advantages include direct ownership of power assets, a flexible load management platform that can shift workloads between bitcoin mining, AI/HPC and grid participation, and a growing footprint in low cost power regions such as Central Texas and Nebraska.

The company serves institutional investors, trading counterparties and lenders that participate in its bitcoin lending and financing activities. It also serves enterprise and public sector customers seeking AI inference and high performance computing compute resources. Specific customer names are not disclosed in the filing, so the base is described by type rather than by individual names.

Investment Thesis

▲ Bull case

Marathon Digital Holdings (MARA) possesses a unique structural advantage in the AI infrastructure race through its ownership of already-energized, low-cost power assets, particularly highlighted by the Long Ridge acquisition. The 505-megawatt combined-cycle gas turbine at Long Ridge generated $144 million in annualized adjusted EBITDA in 2025 with 76% contracted capacity, providing immediate, visible cash flow upon closing—unlike greenfield developments requiring years of permitting and construction. This asset alone offers over 1 gigawatt of total potential capacity, scalable to 2.4 gigawatts, positioning MARA to capture AI/HPC demand in the PJM Interconnection, a premier North American data center and power market where power scarcity, not capital, is the binding constraint on AI growth. The company’s ability to deliver power at approximately $15 per megawatt-hour—among the lowest in the sector—creates a durable cost advantage that hyperscalers and enterprise tenants cannot replicate, turning what was a Bitcoin mining cost center into a premium infrastructure monetization engine.
The Starwood joint venture represents a capital-efficient, scalable monetization model for MARA’s powered land portfolio that the market is underestimating as a one-off partnership. By contributing sites with pre-agreed equity credit based on power, land, and interconnection value, MARA can generate $50 million to $100 million of net annualized stabilized cash flow per 200-megawatt project at a 9% to 15% yield-on-cost with little to no incremental equity required. Starwood brings captive EPC expertise, trusted tenant relationships (having built over 7 gigawatts globally), and development certainty—enabling MARA to scale rapidly across 90% of its owned and operated sites without draining its balance sheet. Unlike peers who must sequentially deploy capital site-by-site, MARA can acquire multiple sites while Starwood handles execution, creating a pipeline effect that accelerates tenant leasing and revenue recognition. This model preserves long-term upside while managing capital exposure, effectively transforming MARA from a Bitcoin miner into a recurring revenue infrastructure owner with institutional-grade tenants.
MARA’s balance sheet de-risking via Bitcoin monetization—selling $1.5 billion of BTC to retire ~30% of convertible debt at a discount—has reduced future dilution by approximately 46 million shares (9% on a fully diluted basis) while avoiding ATM equity issuance since end-2025. This strategic use of Bitcoin as a reserve asset for balance sheet strengthening, rather than speculative holding, demonstrates disciplined capital allocation that enhances financial flexibility for AI/HPC investments. The refinancing of $150 million of its line of credit from 10.5% to 7% further lowers interest expense, improving pro forma profitability. With Bitcoin’s post-quarter rebound of ~20% from March 31 lows and the company’s proven ability to monetize BTC without dilution, MARA retains strategic optionality: if Bitcoin resumes its institutional uptrend, its remaining 35,303 BTC holdings (28% pledged for $66.4 million in 2026 interest income) can again serve as a low-cost funding source for accretive infrastructure growth, creating a dual-engine value model where Bitcoin volatility funds the transition to stable AI infrastructure cash flows.

Marathon Digital Holdings (MARA) possesses a unique structural advantage in the AI infrastructure race through its ownership of already-energized, low-cost power assets, particularly highlighted by the Long Ridge acquisition. The 505-megawatt combined-cycle gas turbine at Long Ridge generated $144 million in annualized adjusted EBITDA in 2025 with 76% contracted capacity, providing immediate, visible cash flow upon closing—unlike greenfield developments requiring years of permitting and construction. This asset alone offers over 1 gigawatt of total potential capacity, scalable to 2.4 gigawatts, positioning MARA to capture AI/HPC demand in the PJM Interconnection, a premier North American data center and power market where power scarcity, not capital, is the binding constraint on AI growth. The company’s ability to deliver power at approximately $15 per megawatt-hour—among the lowest in the sector—creates a durable cost advantage that hyperscalers and enterprise tenants cannot replicate, turning what was a Bitcoin mining cost center into a premium infrastructure monetization engine.
The Starwood joint venture represents a capital-efficient, scalable monetization model for MARA’s powered land portfolio that the market is underestimating as a one-off partnership. By contributing sites with pre-agreed equity credit based on power, land, and interconnection value, MARA can generate $50 million to $100 million of net annualized stabilized cash flow per 200-megawatt project at a 9% to 15% yield-on-cost with little to no incremental equity required. Starwood brings captive EPC expertise, trusted tenant relationships (having built over 7 gigawatts globally), and development certainty—enabling MARA to scale rapidly across 90% of its owned and operated sites without draining its balance sheet. Unlike peers who must sequentially deploy capital site-by-site, MARA can acquire multiple sites while Starwood handles execution, creating a pipeline effect that accelerates tenant leasing and revenue recognition. This model preserves long-term upside while managing capital exposure, effectively transforming MARA from a Bitcoin miner into a recurring revenue infrastructure owner with institutional-grade tenants.
MARA’s balance sheet de-risking via Bitcoin monetization—selling $1.5 billion of BTC to retire ~30% of convertible debt at a discount—has reduced future dilution by approximately 46 million shares (9% on a fully diluted basis) while avoiding ATM equity issuance since end-2025. This strategic use of Bitcoin as a reserve asset for balance sheet strengthening, rather than speculative holding, demonstrates disciplined capital allocation that enhances financial flexibility for AI/HPC investments. The refinancing of $150 million of its line of credit from 10.5% to 7% further lowers interest expense, improving pro forma profitability. With Bitcoin’s post-quarter rebound of ~20% from March 31 lows and the company’s proven ability to monetize BTC without dilution, MARA retains strategic optionality: if Bitcoin resumes its institutional uptrend, its remaining 35,303 BTC holdings (28% pledged for $66.4 million in 2026 interest income) can again serve as a low-cost funding source for accretive infrastructure growth, creating a dual-engine value model where Bitcoin volatility funds the transition to stable AI infrastructure cash flows.

▼ Bear case

Marathon Digital Holdings (MARA) faces imminent margin compression and cash flow volatility as its Bitcoin mining legacy continues to drag on profitability during the transition to AI infrastructure, despite management’s optimistic narrative. The Q1 2026 net loss of $1.3 billion—driven by $1.0 billion in unrealized mark-to-market losses from Bitcoin’s 18% average price decline—exposes the company’s persistent susceptibility to cryptocurrency volatility, even as it pivots toward data centers. While Long Ridge provides immediate cash flow, the Hannibal mining site remains operational and will continue to consume capital and management focus until repurposed for AI load, creating a dual-track burden: sustaining legacy mining operations (with $0.04/kWh energy costs still vulnerable to gas price spikes and network difficulty increases) while simultaneously funding AI buildout. General and administrative expenses excluding stock-based comp rose 56% year-over-year to $57.7 million, fueled by $11.0 million in acquisition/integration costs and restructuring, indicating that the organizational realignment is currently increasing, not reducing, overhead—a trend that could persist if integration of Long Ridge, Aegion, and Starwood sites proves more complex than anticipated, delaying the expected G&A run rate decline.
The Long Ridge acquisition’s financial upside is overstated and contingent on uncertain timelines and execution risks that MARA has not adequately addressed. Although the 505 MW gas plant generated $144 million in annualized adjusted EBITDA in 2025, 76% of its capacity is already contracted, leaving only 24% (≈121 MW) available for new AI/HPC tenants—far less than the implied multi-gigawatt upside. The path to 600 gross megawatts of AI/IT load requires behind-the-meter expansion (18–24 months) and grid interconnection upgrades, both subject to PJM approvals, environmental reviews, and potential grid congestion delays in a region already strained by data center demand. Moreover, MARA’s claim of expanding capacity from 1.3 GW to 2.4 GW relies on optimistic assumptions about permitting speed and fuel supply continuity; natural gas price volatility or PJM interconnection queue backlogs could easily delay or scale back these projections. Crucially, the $900 million pro forma debt post-acquisition (down from $1.1 billion) still leaves meaningful leverage on a campus where only a fraction of power is immediately monetizable for AI, and the $785 million Barclays bridge loan commitment is a contingency—not a funded facility—introducing refinancing risk if Bitcoin monetization windows close during market stress.
The Starwood joint venture’s economics are highly sensitive to tenant credit quality and lease duration, with MARA overestimating the scalability of its 9%-15% yield-on-cost model in a competitive AI infrastructure leasing market. While Starwood brings EPC expertise, the JV structure requires MARA to contribute sites without incremental equity, but the projected $50 million–$100 million annualized cash flow per 200 MW project assumes long-term, triple-net leases with investment-grade tenants—yet MARA has disclosed no signed leases to date, only “advanced conversations” and expectations of signing “multiple tenant leases by year-end.” In a market where hyperscalers are securing power directly through long-term PPAs or owning generation (e.g., Google’s geothermal investments, Microsoft’s nuclear deals), MARA may struggle to command premium yields without owning the generation asset outright or offering superior contract terms. Furthermore, as the JV scales, funding shifts toward traditional construction financing (80% LTV), increasing MARA’s effective capital commitment over time and diluting the initial capital-efficiency advantage—potentially forcing the company to either accept lower returns or divert Bitcoin holdings to fund growth, reintroducing balance sheet volatility exactly as it seeks to escape it.

Marathon Digital Holdings (MARA) faces imminent margin compression and cash flow volatility as its Bitcoin mining legacy continues to drag on profitability during the transition to AI infrastructure, despite management’s optimistic narrative. The Q1 2026 net loss of $1.3 billion—driven by $1.0 billion in unrealized mark-to-market losses from Bitcoin’s 18% average price decline—exposes the company’s persistent susceptibility to cryptocurrency volatility, even as it pivots toward data centers. While Long Ridge provides immediate cash flow, the Hannibal mining site remains operational and will continue to consume capital and management focus until repurposed for AI load, creating a dual-track burden: sustaining legacy mining operations (with $0.04/kWh energy costs still vulnerable to gas price spikes and network difficulty increases) while simultaneously funding AI buildout. General and administrative expenses excluding stock-based comp rose 56% year-over-year to $57.7 million, fueled by $11.0 million in acquisition/integration costs and restructuring, indicating that the organizational realignment is currently increasing, not reducing, overhead—a trend that could persist if integration of Long Ridge, Aegion, and Starwood sites proves more complex than anticipated, delaying the expected G&A run rate decline.
The Long Ridge acquisition’s financial upside is overstated and contingent on uncertain timelines and execution risks that MARA has not adequately addressed. Although the 505 MW gas plant generated $144 million in annualized adjusted EBITDA in 2025, 76% of its capacity is already contracted, leaving only 24% (≈121 MW) available for new AI/HPC tenants—far less than the implied multi-gigawatt upside. The path to 600 gross megawatts of AI/IT load requires behind-the-meter expansion (18–24 months) and grid interconnection upgrades, both subject to PJM approvals, environmental reviews, and potential grid congestion delays in a region already strained by data center demand. Moreover, MARA’s claim of expanding capacity from 1.3 GW to 2.4 GW relies on optimistic assumptions about permitting speed and fuel supply continuity; natural gas price volatility or PJM interconnection queue backlogs could easily delay or scale back these projections. Crucially, the $900 million pro forma debt post-acquisition (down from $1.1 billion) still leaves meaningful leverage on a campus where only a fraction of power is immediately monetizable for AI, and the $785 million Barclays bridge loan commitment is a contingency—not a funded facility—introducing refinancing risk if Bitcoin monetization windows close during market stress.
The Starwood joint venture’s economics are highly sensitive to tenant credit quality and lease duration, with MARA overestimating the scalability of its 9%-15% yield-on-cost model in a competitive AI infrastructure leasing market. While Starwood brings EPC expertise, the JV structure requires MARA to contribute sites without incremental equity, but the projected $50 million–$100 million annualized cash flow per 200 MW project assumes long-term, triple-net leases with investment-grade tenants—yet MARA has disclosed no signed leases to date, only “advanced conversations” and expectations of signing “multiple tenant leases by year-end.” In a market where hyperscalers are securing power directly through long-term PPAs or owning generation (e.g., Google’s geothermal investments, Microsoft’s nuclear deals), MARA may struggle to command premium yields without owning the generation asset outright or offering superior contract terms. Furthermore, as the JV scales, funding shifts toward traditional construction financing (80% LTV), increasing MARA’s effective capital commitment over time and diluting the initial capital-efficiency advantage—potentially forcing the company to either accept lower returns or divert Bitcoin holdings to fund growth, reintroducing balance sheet volatility exactly as it seeks to escape it.

Peer Comparison

Companies in the Capital Markets

S.No.	Ticker	Company	Market Cap	P/E	P/S	Total Debt (Qtr)
1	MS	Morgan Stanley	330.70 Bn	0.00 Bn	4.50	119.83 Bn
2	GS	Goldman Sachs Group Inc	309.79 Bn	0.00 Bn	5.12	259.45 Bn
3	SCHW	Schwab Charles Corp	167.21 Bn	0.00 Bn	6.74	-
4	FUTU	Futu Holdings Ltd	111.36 Bn	85.66 Bn	82.13	0.01 Bn
5	HOOD	Robinhood Markets, Inc.	97.69 Bn	0.00 Bn	21.18	-
6	LPLA	LPL Financial Holdings Inc.	23.49 Bn	0.00 Bn	1.29	-
7	TW	Tradeweb Markets Inc.	21.59 Bn	0.00 Bn	9.99	-
8	CRCL	Circle Internet Group, Inc.	15.14 Bn	0.00 Bn	6.85	-

Research Tools

Developers

About

Investment Thesis

Cumulative Effect, Period of Adoption Breakdown of Revenue (2023)

Peer Comparison

Research Tools

Developers

About

Investment Thesis

Cumulative Effect, Period of Adoption Breakdown of Revenue (2023)

Peer Comparison

Create an account

Sign In

Reset Password