MARA Holdings
Owns and operates energized data-center capacity to mine bitcoin (block subsidy + fees), holds a large BTC treasury marked at fair value, earns interest by lending/pledging BTC, and is building AI-ready data-center capacity (up to 2.5 GW JV) to lease compute/power to AI tenants. Capital-intensive; funded through equity issuance, convertible notes, and BTC monetization.
Earnings, margins, COGS & capex
FY2025 revenue grew 38% to a record $907.1M on a 20% larger BTC treasury (53,822 BTC) and rising energized hashrate, yet the company swung to a ~$1.31B GAAP net loss (vs +$541.0M in 2024) as fair-value accounting marked the bitcoin treasury down with BTC's price decline and D&A rose to $772.8M; the Q4 2025 loss alone was ~$1.7B. Q1 2026 continued the pattern: revenue -18% to $174.6M (below the ~$192.7M consensus) and a ~$1.26-1.3B net loss as BTC fell from ~$87.5k (Dec 31 2025) to ~$68.2k (Mar 31 2026). Management used the quarter to aggressively monetize BTC, cut debt, and fund an AI/HPC data-center pivot.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~55¢ is cost of goods and ~0¢ operating expense, leaving ~45¢ of operating profit (~1¢ net).
Revenue trend
Margins
Compresses when BTC price falls or energy cost/BTC rises
Dominated by non-cash BTC fair-value swings — not representative of cash economics
Rising (up from $35,728 in Q1 2025) as network difficulty grows post-halving
COGS structure
Cost of revenue is primarily electricity/energy plus third-party hosting, plus large non-cash depreciation of mining rigs and facilities. Nine-month 2025: operating & maintenance $68.5M and third-party hosting/other energy $212.9M; Q4 2025 cost of sales alone ~$157.6M. FY2025 D&A $772.8M is the single largest expense line. Owned-site power cost ~$0.04/kWh (Q1 2026) is a structural cost advantage.
Capex
FY2025 capex $407.1M (rigs, infrastructure, and increasingly AI/HPC-ready build-out). Ongoing intensity is high and shifting mix toward AI data-center construction, partly financed externally through the Starwood joint venture.
Latest earnings
Revenue and EPS missed (revenue ~$174.6M vs ~$192.7M consensus); a ~$1.26-1.3B net loss driven by unrealized BTC losses
No hard numeric guidance; management framed the quarter as a deliberate balance-sheet reset — monetize BTC, retire high-cost debt, and redeploy capacity toward AI-ready data centers via the up-to-2.5 GW joint venture with Starwood Capital Group.
- BTC produced Q1 2026
- 2,247 BTC (~25.0 BTC/day)
- BTC held (Mar 31 2026)
- 35,303 BTC (~$2.4B at $68,222); ~28% pledged/loaned
- Energized hashrate
- 72.2 EH/s (+33% YoY)
- Energy cost/BTC
- $40,047; owned-site power ~$0.04/kWh
- Debt action
- >$1.0B face of 2030/2031 converts repurchased; $200M LOC repaid; $150M refinanced 10.5%->7%
Growth drivers
- Bitcoin price (largest single driver of both revenue and mark-to-market results)
- Energized hashrate expansion (66.4 EH/s year-end 2025 -> 72.2 EH/s Q1 2026, +33% YoY) and fleet efficiency
- Low owned-power cost (~$0.04/kWh) and vertical integration (in-house Auradine-linked chips, firmware, immersion cooling)
- Bitcoin lending/pledging income (~$32.1M FY2025 from activating 15,315 BTC; higher run-rate cited for 2026)
- AI/HPC data-center pivot — up-to-2.5 GW JV to lease power/compute to AI tenants, a potential new non-BTC revenue stream
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-03-02. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
A scaled, low-cost power operator with a large corporate BTC treasury and real optionality to convert cheap gigawatts into contracted AI/HPC revenue — leverage to both a bitcoin bull market and the AI-infrastructure buildout, on a freshly de-levered balance sheet.
- Owns/controls large low-cost (~$0.04/kWh) energized capacity — the scarce input both bitcoin mining and AI compute compete for.
- 35,303 BTC treasury (~$2.4B at Q1) gives direct BTC upside plus a monetization/collateral lever, as demonstrated by the ~$1.5B Q1 sale funding debt paydown.
- Q1 2026 de-levering (>$1.0B converts retired, LOC cut, remaining debt refinanced 10.5%->7%) materially lowers financial risk and interest burden.
- The up-to-2.5 GW AI-ready data-center JV with Starwood Capital Group could add a higher-multiple, non-BTC recurring revenue stream that re-rates the equity away from pure miner comps.
- Largest US-listed miner by hashrate with vertical integration (chips/firmware/cooling) — scale and cost edge in a consolidating industry.
A capital-hungry, dilutive, commodity bitcoin miner whose reported earnings are a leveraged bet on BTC price, now chasing an AI pivot where peers are further along and hyperscalers are far better capitalized — with no AI revenue yet to show for it.
- GAAP results are effectively a BTC-price derivative: a ~$1.31B FY2025 loss and ~$1.26-1.3B Q1 2026 loss came from treasury markdowns, not operations.
- Structural margin compression — every halving and difficulty rise lifts cost/BTC ($40,047 energy cost/BTC in Q1 2026) against a fixed block subsidy.
- Persistent dilution and ~45%-of-revenue capex mean shareholders fund growth that a BTC drawdown can erase overnight.
- AI pivot is unproven vintage-2026: no material AI tenant revenue disclosed, and rivals (Core Scientific/CoreWeave, IREN, Cipher, TeraWulf) plus neoclouds have a head start and deeper capital.
- Market cap (~$4.7-5.1B) sits well above Q1 BTC treasury value (~$2.4B), so the equity already prices in mining + AI optionality that must be delivered.
What it is worth
Sum-of-parts / mNAV lens: BTC treasury value + mining operating business + AI/HPC optionality, sanity-checked against miner peers (RIOT, CLSK, CORZ, IREN) on EV/hashrate and EV/EH plus BTC-NAV premium.
BTC drawdown + rising difficulty compress mining margins and mark down the treasury while AI revenue fails to materialize on schedule; the premium to treasury NAV collapses and dilution deepens losses.
MARA remains scaled BTC-beta with a de-levered balance sheet and early AI optionality; trades in a wide band tethered to BTC price around a premium to treasury NAV.
BTC re-rates higher and the up-to-2.5 GW AI JV signs creditworthy tenants -> a non-BTC recurring revenue stream re-rates MARA off pure-miner multiples; equity worth a large premium to treasury NAV.
At ~$4.7-5.1B market cap vs a Q1 2026 BTC treasury of ~$2.4B, the equity trades at a premium to treasury value, implying the market ascribes meaningful value to mining operations plus AI-pivot optionality. GAAP P/E is meaningless given BTC-markdown losses; the stock behaves as leveraged BTC beta with an embedded AI call option. Valuation is supported only if BTC holds/appreciates AND the AI JV converts power into contracted tenant revenue.
SWOT
Strengths
- Largest US-listed bitcoin miner by energized hashrate (72.2 EH/s Q1 2026); among the largest corporate BTC holders (35,303 BTC).
- Low owned-site power cost (~$0.04/kWh) and vertical integration across chips, firmware, and immersion cooling.
- Large, liquid BTC treasury usable as collateral and monetization lever; generated ~$32M+ lending income in FY2025.
- Access to capital markets (equity + convertibles) enabling rapid scale and de-leveraging optionality.
Weaknesses
- Earnings dominated by non-cash BTC fair-value swings — GAAP results are highly volatile and loss-heavy when BTC falls.
- Heavy capex intensity (~45% of revenue) and a history of significant shareholder dilution.
- Commodity economics — hashrate is undifferentiated and margins compress with every halving and difficulty increase.
- AI/HPC pivot is early-stage and unproven — no established AI tenant revenue yet; execution risk is high.
Opportunities
- Repurpose low-cost power + land into AI/HPC data centers (up-to-2.5 GW JV) to capture non-BTC, contracted revenue.
- BTC treasury upside if bitcoin appreciates — direct leverage to a crypto bull market.
- Bitcoin lending/collateral strategy to earn yield on otherwise idle treasury.
- Consolidation of the fragmented mining sector using scale and balance-sheet strength.
Threats
- Bitcoin price crashes drive both revenue and treasury markdowns — existential to the model.
- Post-halving reward compression + rising global network difficulty squeeze mining margins.
- AI-pivot competition from better-capitalized neoclouds/hyperscalers and from peers (CORZ, IREN, CIFR) already further along.
- Regulatory/energy-policy risk (crypto rules, power access, grid interconnect queues) and refinancing risk if capital markets tighten.
Moats, dependencies & bottlenecks
Moats
72.2 EH/s is industry-leading but hashrate is a commodity; scale helps unit cost, not pricing power.
The most durable edge — cheap, controllable power is scarce and underpins both mining and the AI pivot.
In-house tech (Auradine-linked silicon, 2-phase immersion) can improve efficiency vs peers reliant on Bitmain/MicroBT.
A balance-sheet asset and yield source, not a competitive barrier; fully exposed to BTC price.
Dependencies
Drives revenue and the fair-value marks that dominate net income.
3.125 BTC subsidy post-2024 halving; rising difficulty steadily raises cost/BTC.
Input/Infrastructure Margins and the AI buildout both hinge on cheap power and available capacity.
Concentrated rig supply; vertical integration only partially mitigates.
AI pivot needs GPU access and creditworthy tenants that do not yet exist in the numbers.
Growth and refinancing depend on continued access; source of chronic dilution.
Advantages
- Largest US-listed miner by energized hashrate with a low owned-power cost base.
- Vertical integration across silicon, firmware, and cooling.
- Deep, liquid BTC treasury usable as a monetization and collateral lever.
- Freshly de-levered balance sheet after the Q1 2026 debt paydown/refinancing.
Weaknesses
- Earnings are a volatile, non-cash-driven function of BTC price.
- Chronic dilution and high capex intensity.
- Undifferentiated commodity mining economics with post-halving margin compression.
- AI pivot is early, capital-hungry, and behind faster-moving peers.
Bottlenecks
- Power capacity and grid interconnect timelines gate both hashrate growth and AI data-center buildout.
- GPU/AI-hardware supply and the need for creditworthy AI tenants to underwrite the up-to-2.5 GW JV.
- Capital intensity and dilution — funding capex without further eroding per-share value.
- Rising network difficulty raising cost per mined BTC faster than efficiency gains.
Top signals & trends
Top signals
De-risks the balance sheet and cuts interest cost (10.5%->7% on refinanced tranche).
Strategic optionality and external capital, but unproven and a distraction from focus until tenants sign.
Operational scale-up continuing despite the pivot narrative.
Less BTC upside beta, but a more resilient balance sheet.
Highlights earnings fragility to crypto price, even as cash operations continue.
Trends
Cheap power + data-center land repurposed for AI compute (CORZ/CoreWeave, IREN, CIFR, WULF) — MARA following, later than some.
Makes reported earnings swing with BTC price; investors must look through to cash economics.
Secular squeeze on pure-mining margins, pushing diversification.
MARA is among the largest corporate holders; treasury value is a meaningful swing factor in the equity.
Demand for gigawatt-scale power/compute is the thesis underpinning the pivot's optionality.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Private (China-based); dominant Antminer ASIC supplier to the industry (context, not a buy call).
Private (China-based); Whatsminer ASIC supplier (context, not a buy call).
Private US chip/infrastructure startup MARA has backed; supports vertical-integration efforts.
GPU supplier for the AI/HPC data-center pivot.
Block subsidy + transaction fees are the primary 'customer' of mining output.
Future neocloud/hyperscaler/enterprise tenants for the up-to-2.5 GW data-center JV — not yet material revenue.
Institutions that borrow/hold pledged BTC, generating interest income.
Large US bitcoin miner also converting Texas power capacity toward AI/HPC hosting.
Efficient pure-play US bitcoin miner focused on low-cost operations.
Miner that pivoted hard to HPC/AI hosting (major CoreWeave contracts) — furthest along on the AI thesis.
Formerly Iris Energy; miner plus fast-growing AI cloud/GPU business — a direct pivot comp.
US miner developing large-scale HPC/AI data-center capacity.
Low-carbon miner expanding into AI/HPC hosting.
Vertically integrated miner + self-designed ASICs; competes on efficiency and scale.
Diversified miner/energy-infrastructure operator with AI ambitions.