{
  "@context": "https://orbyd.app/schemas/dossier.v1.json",
  "ticker": "MARA",
  "name": "MARA Holdings, Inc.",
  "url": "https://orbyd.app/dossiers/MARA/",
  "json_url": "https://orbyd.app/dossiers/MARA.json",
  "status": "DORMANT",
  "current_conviction": "LOW",
  "graded_conviction": "MEDIUM",
  "archetype": {
    "code": "a4",
    "n": 4
  },
  "current_thesis": "AI/HPC power-buildout re-rate is the leg being bought — the 2026-07-09 HIF deal adds up to 2 GW in Texas, lifting potential capacity toward ~4.8 GW — but the golden-cross base fully round-tripped to ~$11.4 as analysts cut across the board (MS $5.50, Piper $13) and MARA still trades as leveraged Bitcoin beta into a 21-month-low tape. Narrative real, structure broken; a low-conviction probe at most.",
  "invalidation_trigger": "A weekly close below $10 (loses the July shelf and the round number under the sub-$13 analyst cluster, confirming the AI-power-buildout bid is not being paid for and MARA is tracking the sub-$60K Bitcoin tape toward the $5.50 bear target); reinforced by a BTC weekly close under $60K.",
  "catalyst_date": null,
  "outcome": "INVALIDATED",
  "outcome_date": "2026-07-16",
  "invalidation_fired": true,
  "themes": [
    "bitcoin-miners",
    "ai-datacenter-infrastructure",
    "industrial-power-grid",
    "gpu-cloud-neoclouds"
  ],
  "tags": [],
  "sources": [],
  "notes": [
    "Effectively leveraged BTC beta until AI/HPC revenue actually prints — correlates to Bitcoin, not the AI-capex cycle. Don't double-count vs MSTR/RIOT/CLSK/IREN/HIVE.",
    "Q1 2026 reported 2026-06-02 (double-miss: rev $174.6M, EPS -$3.31 vs ~-$1.41 est, ~$1.3B net loss, ~$1.0B BTC writedown). Q2 2026 ~mid-Aug — avoid fresh exposure into the print.",
    "Analyst dispersion extreme: MS $5.50 UW (7/8), Piper $13 OW (7/17), Bernstein $17 MP (7/2), Cantor $10, Citizens $24 MO (6/24). $5.50–$24 spread = battleground, no consensus edge.",
    "Holds 35,303 BTC (Q1 2026), record hashrate 72.2 EH/s (+33% YoY) — every BTC leg lower compounds fair-value writedowns and decouples GAAP EPS from operations.",
    "Power pipeline: HIF/Matagorda Texas (up to 2 GW by Apr 2028) + Long Ridge 505MW → ~4.8 GW potential capacity. Capacity energizes 2027–2028, not 2026.",
    "The trade only becomes interesting on a contracted AI/HPC revenue print (Matagorda/Long Ridge/Starwood go-live) — unscheduled; converts pivot from promise to fact. Watch for it.",
    "Prior golden-cross base (6/23) fully round-tripped by 7/16 and the June higher-low is broken; a fresh momentum setup must be rebuilt (reclaim ~$13/20-EMA) before an entry.",
    "BTC context: 21-month low ~$58K late June, first weekly close below $60K / 200-week MA since 2023, reclaimed $65K on 2026-07-15 on softer CPI + ETF-flow reversal. FOMC ~7/29 is the swing factor."
  ],
  "body_markdown": "## Current Thesis\nThree weeks ago the golden cross (6/23) and Citizens' $24 initiation (6/24) had MARA near $14.85, outperforming a falling Bitcoin — the AI-decoupling read. That leg has fully unwound: MARA sat ~$11.41–$11.56 on 2026-07-16, back below the ~$12.30 June swing low, having round-tripped the entire golden-cross advance. What is genuinely new and constructive is the power evidence: the 2026-07-09 HIF agreement adds a 1,200-acre Matagorda County, Texas site with up to 1 GW of grid access by Oct 2027 and 2 GW by April 2028, lifting potential portfolio capacity toward ~4.8 GW alongside Long Ridge. But the tape votes the other way — Morgan Stanley cut to $5.50 Underweight (7/8), Piper trimmed its Overweight to $13 (7/17), Bernstein to $17 (7/2) — and MARA remains leveraged Bitcoin beta after BTC printed a 21-month low near $58K in late June. Real narrative, broken structure, no contracted AI revenue until ~2028: the setup for a fresh long is a pass here, better to wait for a reclaim of structure or an actual revenue print.\n\n## Bull Case\n- **Power capacity is stacking through signed deals, not slideware**: the 2026-07-09 HIF/Matagorda agreement adds up to 2 GW (1 GW by Oct 2027, 2 GW by April 2028), pushing potential portfolio capacity to ~4.8 GW with Long Ridge; the site already reports HPC tenant interest.\n- **Owned megawatts are the scarce input for AI**: Long Ridge contributes a 505MW gas plant plus 1,600 Ohio acres — power ownership is the binding constraint asset-light GPU renters cannot clear.\n- **A bull still pays up**: Citizens initiated Market Outperform $24 (6/24) explicitly on the HPC/power pipeline, and even after the 7/17 cut Piper keeps an Overweight at $13.\n- **BTC bounced off the low**: after a 21-month low near $58K in late June, Bitcoin reclaimed $65K on 2026-07-15 on softer US inflation and an ETF-flow reversal, easing fair-value-writedown pressure on the 35,303-BTC treasury.\n- **Policy tailwind building**: the CLARITY Act reframed crypto as an AI-race-vs-China priority (2026-07-14), a structural positive for domestic mining and HPC power ownership.\n\n## Bear Case\n- **The whole June rally round-tripped**: from ~$14.85 (6/22) to ~$11.4 (7/16); the golden-cross base and June higher-low are both gone — the cheap-and-broken structure that traps dip-buyers.\n- **Analysts are cutting the bulls too**: MS $5.50 Underweight (7/8), Bernstein $17 (7/2 from $23), Piper $13 (7/17 from higher); a $5.50–$24 target spread is a battleground, not a consensus.\n- **AI revenue is a 2028 event**: no contracted HPC revenue line exists and the Texas capacity energizes 2027–2028; the re-rate is being paid ahead of any cash flow, and the 7/9 deal pop faded within days.\n- **Impaired core P&L**: Q1 2026 (reported 6/2) printed a ~$1.3B net loss, EPS -$3.31 vs ~-$1.41 est on revenue $174.6M; holding 35,303 BTC means every leg down compounds the markdown.\n- **Still a BTC proxy in a downtrend**: Bitcoin halved from its $126K Oct-2025 ATH and closed a full week below $60K (first sub-200-week-MA close since 2023); miners \"slid as AI pivot faces fresh headwinds\" (7/4).\n\n## Setup & Price Structure\n- Last ~$11.41–$11.56 (2026-07-16), intraday range $11.25–$12.24; the stock has surrendered the entire 6/22–6/24 golden-cross/Citizens pop and trades below the ~$12.30 June swing low.\n- The 20-EMA (~$13) and the round-tripped ~$14.8 June high sit overhead; there is no higher-low to lean on — price is making lower lows into a falling tape.\n- $10 is the next visible shelf and the psychological line under the sub-$13 analyst cluster (Cantor $10, Piper $13); losing it opens the MS $5.50 zone.\n- Market cap ~$4.36B. This is a falling-knife structure, not an accelerating breakout — strength has to be rebuilt (a reclaim of ~$13/20-EMA on volume) before the momentum setup this playbook wants even exists.\n\n## Catalyst Calendar (next 30 days)\n\n- **~2026-08-12 (est.)** — Q2 2026 earnings, likely just outside 30 days; a binary print on cash burn, hashrate, and any AI/HPC revenue commentary. Avoid fresh exposure into it if it pulls forward.\n- **Unscheduled** — a contracted AI/HPC revenue announcement (Matagorda / Long Ridge / Starwood go-live) converts the pivot from promise to fact and is the datapoint that would re-rate the name off Bitcoin beta.\n\n## Elapsed catalysts\n\n- **~2026-07-29** — FOMC rate decision. The dominant near-term driver for a BTC-beta miner; Bitcoin was rangebound $56–62K into the meeting, and a hawkish hold versus a dovish cut swings MARA hard. *(passed 11d ago)*\n- **2026-07-09 (elapsed, tracking)** — HIF/Texas closing conditions and regulatory approvals; phased construction \"expected to begin 2026.\" Any go-live or named HPC tenant is an unscheduled upside catalyst. *(passed 31d ago)*\n\n## What Would Change Our Mind\n- **Structure reclaim**: a weekly close back above the ~$13 shelf / 20-EMA on expanding volume would re-establish the higher-low the golden-cross thesis needs, turning a stand-aside into a chase-able setup.\n- **A contracted AI/HPC revenue print**: any signed HPC tenant or dollar figure on the Texas/Long Ridge capacity would decouple the multiple from BTC and validate the ~4.8 GW story ahead of the 2028 timeline.\n- **BTC regime flip**: a sustained Bitcoin weekly close back above ~$65–68K removes the fair-value-writedown overhang and restores the leveraged-beta bid.\n- **Downside confirmation**: continued lower lows with analysts migrating toward the $5.50 MS target confirms the AI narrative is not being paid for and MARA is pure declining-BTC beta.\n\n## Correlation Notes\n- **Bitcoin is the master variable** until AI revenue prints — MARA is leveraged spot-BTC beta through both the 35,303-BTC treasury and mining margin. A BTC weekly close under $60K reasserts the miner-economics bleed.\n- **Miner cohort** (RIOT, CLSK, IREN, HIVE, CIFR) moves together — the 7/4 \"AI pivot faces fresh headwinds\" slide hit the whole group; do not double-count MARA against them, nor against MSTR, which is a levered-BTC treasury proxy rather than an AI-capex name.\n- **AI-datacenter / power complex** (neoclouds, industrial power) is the aspirational comp. MARA wants to trade like a power-to-AI developer — cf. Galaxy's $4.5B AI-infra deal (6/24) — but the market will not grant that multiple until megawatts convert into contracted HPC revenue.",
  "first_seen": "2026-04-24",
  "last_analyzed": "2026-07-18T08:13:51+00:00",
  "last_synthesized": "2026-07-18",
  "last_update_source": "watchlist_research",
  "license": "Content © orbyd. Cite the canonical URL."
}