JPMorgan downgrades MARA Holdings stock rating on data center strategy

JPMorgan downgraded MARA Holdings to Underweight and cut its December 2027 price target to $11 from $13, below the stock's $11.98 trading price. The bank cited MARA's capital-light AI data-center joint venture structure, which leaves it dependent on Starwood Digital Ventures while capturing only 50% of value creation. MARA has posted negative adjusted EBITDA for three consecutive quarters, with trailing-12-month EBITDA of negative $1.88B, and requires Bitcoin to remain near $80,000 to break even; Q2 2026 EPS was a $1.60 loss versus consensus for a $0.35 profit, while revenue of $174.9M missed the $208.37M estimate.

Analysis

MARA is increasingly exposed to a valuation mismatch: equity holders are underwriting an AI-infrastructure multiple while the company retains mainly land-contribution economics and limited control over construction, financing, utilization, and operating execution. This structure can preserve cash in the near term, but it also caps upside participation and makes any AI-value attribution highly contingent on third-party lease execution. Relative to vertically integrated power-to-compute peers such as IREN and CORZ, MARA should remain vulnerable to multiple compression until contracted capacity—not development plans—becomes visible.

The near-term equity driver remains Bitcoin economics rather than AI optionality. With profitability sensitive to a relatively high BTC threshold, a flat-to-down Bitcoin tape can produce disproportionate downside through reduced treasury accumulation, weaker operating cash flow, and potential dilution or incremental financing needs; that risk matters over the next 1-3 months. Conversely, a sustained BTC rally could temporarily overwhelm the fundamental short thesis, since miners still trade as high-beta crypto proxies regardless of underlying margin quality.

The market may be underestimating the timeline risk around power-market and regulatory approvals. A delayed approval or absence of a creditworthy tenant lease over the next two quarters would convert the AI narrative from a valuation support into a capital-allocation overhang, especially if the company contributes additional capex before project financing is fully committed. The contrarian bull case is that powered land scarcity becomes sufficiently valuable to attract a premium tenant; that thesis is falsifiable by a signed long-duration lease with disclosed economics and non-recourse development funding.

Key Decisions for Investors

• Initiate a 1-3 month relative-value short MARA / long IREN basket, sized beta-neutral to BTC. IREN offers more direct compute-infrastructure upside, while MARA carries greater lease, approval, and margin-execution risk; target 15-25% relative outperformance. Exit if MARA discloses a financed, creditworthy long-term tenant contract with economics that materially exceed land-rent returns.
• For directional crypto exposure, avoid using MARA as the preferred BTC proxy until operating breakeven falls meaningfully below the stated BTC threshold. Prefer liquid BTC exposure or a higher-quality miner basket; MARA's incremental downside is likely greater than its upside in a merely sideways Bitcoin environment.
• Use an alert—not a fresh long catalyst—around ERCOT/FERC decisions and tenant announcements. A regulatory approval without a binding lease should not change the thesis; a lease without disclosed term, power price, capex funding, and MARA's share of EBITDA should be treated as promotional rather than valuation-changing.
• If MARA rallies materially above the current level solely on Bitcoin strength or AI-sector sympathy, add to the short via defined-risk put spreads dated 3-6 months. The key risk is a sharp BTC squeeze; use a thesis stop if Bitcoin sustains well above the operational breakeven level and MARA's quarterly adjusted EBITDA turns positive without material dilution.