Galaxy Digital

Phase A — Understand the business

Lens 1 · Company Overview

What it actually is. A crypto-native financial-services firm bolting on a data-center landlord. Two reportable operating segments plus a corporate/treasury bucket:

• Digital Assets segment — the real operating business. Two sub-units:
• Global Markets: OTC spot + derivatives trading, lending, structured products, and M&A / ECM / DCM advisory. Galaxy acts as principal in digital-asset sales, so it grosses up turnover (this is why "revenue" is $60B — see Lens 4).
• Asset Management & Infrastructure Solutions: ETF + alternatives strategies, validator staking, tokenization, and the GK8 self-custody tech. ~$12.3B in assets across the platform as of 2025-12-31. (Note: management elsewhere cites "~$17B in digital assets" — a broader AUM definition; treat $12.3B as the filing-grounded figure.)
• Data Centers segment — the Helios campus (West Texas panhandle, Dickens County). Develops HPC/AI infrastructure; no revenue yet, expected to begin H1-2026 as CoreWeave critical-IT load is delivered.
• Treasury & Corporate — Galaxy's own balance-sheet book: proprietary trading, proprietary bitcoin mining (winding down), and a portfolio of digital assets + venture/PE/fund investments. This is the prop book that swings the P&L.

Customers / counterparties. "More than 1,600 trading counterparties spanning both crypto-native and traditional finance". Concentration is real: in FY2025 two digital-asset trading platforms = 38% of revenue (FY24: 40%); three counterparties = 45% of the loan-receivable book; one centralized platform held 14% of the firm's digital assets. The single most important "customer," however, is now CoreWeave (15-year anchor data-center tenant).

Contract structure / payment terms. Mixed: Global Markets earns bid/ask spreads (transactional, volume-driven), net interest on lending (the loan book averaged $1,794.8M in FY25 vs $639.9M in FY24), and event-driven advisory fees. Asset Management earns recurring management + performance fees. The Data Centers leg is the structural change: long-term, take-or-pay-style 15-year leases that management explicitly frames as "long-term, predictable revenue… uncorrelated to the prices of digital assets". $150M of CoreWeave lease prepayment already sits as deferred lease revenue.

New surface (Oct 2025): GalaxyOne — a retail fintech app (relaunch of acquired "Fierce"): FDIC-insured high-yield deposit via Cross River Bank, a GDH-guaranteed debt security for accredited investors, and commission-free equities/crypto via DriveWealth + Paxos. Early-stage; a strategic optionality bet on retail, not yet material.

Lens 2 · Supply Chain

This is a financial + infrastructure firm, so the "supply chain" is two distinct value chains — name the actual nodes:

(A) Digital-asset trading/lending chain. Upstream liquidity/inputs → centralized exchanges & trading platforms (one platform = 14% of Galaxy's held digital assets; two platforms = 38% of revenue ) and DeFi protocols — Aave, Kamino, and Spark = 70% of Galaxy's DeFi-deployed balance at 2025-12-31 (FY24 it was Sky/MakerDAO, Coinbase wrapped tokens, Aave, Pendle = 77%). → Galaxy (principal trader / lender / staker) → end users = the 1,600+ institutional/TradFi counterparties + GalaxyOne retail. Funding inputs: fiat loans-payable (one lender >9.4% of current liabilities), $2.1B exchangeable notes, customer collateral.

• Chokepoint: counterparty + platform concentration (a platform failure or a top-3 borrower default is the single-source risk). Collateral >100% is held against each top-3 borrower, which mitigates but does not eliminate gap risk in a fast crypto drawdown.

(B) Helios AI/HPC data-center chain. Power — ERCOT grid interconnection is the gating input: 1.63 GW gross approved at Helios (doubled in 2026 via a +830 MW interconnection approval). → Land — 1,500+ contiguous acres under Galaxy's control, 3,400+ MW total potential. → Construction/retrofit (converting Argo-legacy mining halls to liquid-cooled AI halls) — $529.8M construction commitment outstanding at YE2025. → Project finance — Deutsche Bank $1.4B senior secured term loan (Galaxy Helios I LLC), $878M drawn at YE2025. → Anchor tenant — CoreWeave (the GPU-cloud operator), 526 MW critical IT load contracted, headed toward the full ~800 MW.

• Chokepoints: (1) ERCOT power timeline; (2) construction execution / liquid-cooling delivery on schedule; (3) CoreWeave's own creditworthiness — a single tenant carrying a >$1B/yr, 15-year obligation. CoreWeave is itself a highly-levered, fast-growing, AI-cycle-dependent name; its credit IS Galaxy's data-center credit.

Lens 3 · Competitive Advantages (moats)

Digital Assets segment — moat: MEDIUM, eroding-from-strength.

• Institutional trust + breadth — Galaxy is one of the few crypto firms with a credible institutional brand, a registered investment adviser, KPMG as auditor since 2021, and a TradFi-fluent senior team. That is a genuine switching-cost/relationship moat with allocators.
• Balance-sheet scale — the ability to warehouse risk and quote large blocks (e.g. the single 80,000 BTC trade in Q3-2025 that drove a $17.7B revenue jump ) is a capital moat narrow competitors can't match.
• Bargaining power — weak vs. exchanges (it depends on a handful of platforms for liquidity/custody), moderate vs. lending counterparties (over-collateralized), moderate vs. asset-management clients (sticky but fee-competitive). Crypto trading spreads are structurally compressing as the market institutionalizes — the moat here is relationships + scale, not pricing power.

Data Centers segment — moat: HIGH if executed, but unproven.

• The durable moat is interconnected power + permitted land at scale — 1.63 GW of ERCOT-approved capacity and 3,400+ MW of potential on owned land is genuinely scarce in the AI-power land-grab. Once a hyperscaler-grade tenant is energized under a 15-year lease, the asset is a quasi-utility annuity with very high switching costs (you cannot re-home 500 MW of GPUs easily).
• Caveat: this moat is prospective. Galaxy has zero data-center operating revenue and is a first-time large-scale AI-DC developer. The moat converts from "story" to "real" only on flawless Phase-I→III delivery.

Net: the combination is the pitch — a cash-generative (in good quarters) trading franchise funding/de-risking a high-moat infrastructure annuity. The bear retort (Lens 13) is that you're paying for two execution-dependent things at once.

Lens 4 · Segments

Critical accounting point first: GAAP "Revenue" is gross principal turnover and is economically misleading. Galaxy books the full proceeds of every digital-asset sale as revenue and the cost as "Transaction expenses," which net to ~zero. So focus on net economics, not the headline.

Headline (gross) revenue:

The FY25 revenue jump (+42%) was almost entirely the single 80,000-BTC Q3 trade plus higher average prices — not a durable growth signal.

The economically meaningful net lines (FY2025):

• Fees: $158.6M (+54% YoY) — boosted by the Forward Industries Solana-treasury PIPE placement-agent fee + higher AUM fees
• Blockchain rewards (staking): $220.0M (+22%) — Galaxy retains only 5–10% of third-party staking rewards
• Interest income: $147.9M (+73%) — on the $1.79B average loan book
• Proprietary mining: $14.5M (−77%) — ceased at Helios Q1-2025 for the AI conversion; mining-hosting revenue went to $0 (was $31.6M)
• Analyst-cited true economics: ~$426M FY25 adjusted gross profit is the right denominator, not the $60B.

By reportable segment — pre-tax income:

This table is the whole thesis in four rows. The operating business (Digital Assets) is growing and profitable — pre-tax income quadrupled FY24→FY25. The consolidated loss is entirely Treasury & Corporate — i.e. mark-to-market on Galaxy's own crypto/investment book, plus notes interest, plus the LUNA settlement and mining-impairment charges. The house book gave back in FY25 (crypto round-trip + Oct-2025-peak-to-2026 drawdown) what it made in FY24.

By geography (pre-tax): Domestic −$359.3M vs Foreign +$88.6M (FY25). Most positive economic income is recognized offshore (legacy Cayman/foreign-subsidiary structure); the US entity carries the losses + a $208.1M NOL carryforward.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print: Q1-2026, filed 2026-05-08)

The print:

• Revenue (gross) $10.04B (−23% YoY); net loss $(216.3)M; EPS −$0.48 basic / −$0.49 diluted on 192.1M wtd Class A shares.
• Firmwide Adjusted EBITDA −$187.5M — negative across all segments incl. Digital Assets (−$19.4M).
• Driver: a $284.4M digital-asset impairment swamped a positive +$171.8M gains-from-operations / +$279.0M net gain on digital assets. This is the accounting asymmetry: upside is partly capped at "lower of cost or observable fair value" while downside hits as impairment — so a down-quarter for BTC reads worse than the economics.
• Balance sheet shrank with crypto: total assets $11.3B→$10.0B; digital assets $4.5B (BTC −22% in the quarter); cash $1,246M→$910.7M; P&E $1,423M→$1,777.9M (+$354.8M — the Helios build accelerating even as the book bled).

vs consensus / market reaction: Revenue beat (~$10.04B vs ~$8.75B est, +$1.29B) but a wider GAAP loss; non-GAAP EPS −$0.49 was framed as a "beat by $0.44" on lowered numbers. Stock fell ~5% pre-market. Net read: the market shrugged at the crypto-driven loss and kept its eyes on Helios — which is the entire decoupling thesis.

Critical context — the prior-quarter contrast. Q3-2025 was a record: +$505M net income, diluted EPS $1.01. So in three quarters Galaxy went +$505M → (FY-Q4 loss) → −$216M. The earnings are a leveraged bet on the crypto tape, full stop. Management's preliminary Q2-2026 guide of ~$90M adjusted EBITDA signals a sharp sequential rebound as crypto stabilized off the lows — consistent with the volatility, not a break from it.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts=0). From web-sourced call coverage:

• Tone shift over the last ~4 quarters: Q3-2025 call = triumphant (record print, 80K-BTC trade, up-list momentum). Q4-2025 / Q1-2026 calls = deliberate pivot in the narrative from "crypto earnings power" to "data-center milestone." Management now leads with Helios/CoreWeave delivery and contextualizes the crypto loss as mark-to-market.
• Recurring phrases: "uncorrelated, long-term, predictable revenue," "$15 billion-plus infrastructure buildout," "power-ready," "first data hall delivered to CoreWeave."
• What they stopped emphasizing: proprietary bitcoin mining (now de-emphasized/wound down at Helios) and trading-volume bragging — replaced by AUM growth + infrastructure.
• Sentiment read: management is consciously repositioning GLXY as an AI-infrastructure equity with a crypto cash engine, because that is the higher multiple. Credible, but it is also exactly what a CEO does when the core business just printed a loss.

Lens 7 · Comps

GLXY is GAAP-loss-making, so P/E is n/m and the right frame is SOTP / book / Adjusted-EBITDA — that is itself the analyst consensus ("Galaxy's headline revenue is misleading; value the parts" ). Peer multiples below are ``, mid-2026, and are directional (sources vary):

Market-cap conflict (surfaced, not hidden): sources report GLXY mkt cap anywhere from $6.55B to $13.33B on adjacent dates. The spread is the dual-class / LP-unit structure: ~192.7M Class A shares × ~$34 ≈ $6.5B Class-A float cap, but fully-diluted (Class A + 198.4M Class B/LP units + exchangeables) the diluted weighted share count is ~366M → ~$12–13B fully-diluted. Use the fully-diluted ~366M count for any per-share valuation — the Class-B/LP units are economically equivalent Class-A claims.

Mechanism-appropriate comps (not pure multiples): for the trading/lending leg, COIN/HOOD/MARA-style crypto-equities; for the Helios leg, neocloud/AI-DC landlords (CoreWeave itself as counterparty, plus IREN, Core Scientific-style miner-to-AI converters, and merchant-power DC developers). GLXY is one of the cleanest miner-to-AI-DC conversion plays alongside the former-bitcoin-miner cohort — that is the comp that matters for the re-rating.

Lens 8 · Stock-Price Catalysts (what actually moves GLXY)

Pattern over the last ~18 months:

• April 2025 — CoreWeave 15-yr lease signed → the single most important value-creation event; reframed the equity.
• May 2025 — Reorganization + Nasdaq up-list (GLXY) → unlocked US index/institutional eligibility; structural re-rating catalyst.
• June 2025 — $477.8M equity raise @ $19.00; Oct 2025 — $460M strategic PIPE @ $36.00 (incl. insiders selling 3.75M shares) → financing the buildout; the $19→$36 step-up in four months tracks the re-rating.
• Aug 2025 — $1.4B Deutsche Bank Helios facility → de-risked Phase-I funding.
• Aug 2025 + Jan 2026 — Phase II / Phase III CoreWeave leases (→526 MW) and 2026 — +830 MW ERCOT approval (1.63 GW total) → expanded the annuity; each lifts the SOTP.
• Q3-2025 record print ($505M) and Q1-2026 loss (−$216M) → earnings move the stock, but less than the crypto tape and Helios news now do.
• June 2026 — bitcoin −~48-51% from Oct-2025 ATH → and yet GLXY is +45.8% TTM, ~$34, mid-upper of its $16.43–$45.92 range.

What the pattern reveals: the market has partially decoupled GLXY from spot bitcoin and is pricing the Helios/CoreWeave annuity + AI-infrastructure optionality as the swing factor. Crypto still matters (it drives the printed earnings and the book NAV), but the re-rating lever is data-center execution and capacity announcements.

---

Phase C — Judge people & books

Lens 9 · Management

Mike Novogratz — Founder, CEO, controlling shareholder.

• Track record: ex-Goldman Sachs partner (1998), ex-Fortress Investment Group principal (macro). Founded Galaxy 2015, CEO since Dec 2017; took it public (TSX 2018, Nasdaq 2025). Genuine macro/markets pedigree and a TradFi rolodex — a real asset in institutionalizing crypto.
• Skin in the game: controls Class B common stock via Galaxy Group Investments LLC — Class B carries governance but no economic return (founder control structure); Forbes pegs his net worth ~$2.7B, down from a ~$4.8B peak. Economic-ownership figures conflict wildly across sources (0.65%–10%) because of the dual-class/LP structure — treat as "controlling founder, large but cyclical stake."
• Capital-allocation history: aggressive serial acquirer — Helios/Argo mining (Dec 2022), GK8 from Celsius bankruptcy (Feb 2023), CMF node operator (Jul 2024), Fierce→GalaxyOne (Dec 2024). The Helios→AI-DC pivot is a genuinely bold, value-additive reallocation (mining infra → AI annuity). But the prop book (Treasury & Corporate) is where capital has been destroyed in down-cycles — ROE is wildly cyclical (positive FY23/24, deeply negative FY25). He is a founder-operator/macro-trader archetype, not a steady capital-compounder — implies upside torque + drawdown risk.
• Red flags (related-party — material, disclosed): The CEO owns a private aircraft and a private boat used by the company ("on terms advantageous to the Company," arms-length per policy — $1.1M aircraft cost in FY25). Sub-advisory arrangements with a GDI beneficial owner ($2.7M advisory fees FY25). A former chairman/board-member consulting agreement (1.5M RSUs + 500k options). Tax distributions to LP unitholders ($49.3M FY25), majority to related parties. None individually damning, but the pattern = a founder-controlled vehicle with meaningful related-party surface; govern accordingly.

Bottom line: the right jockey for the story (macro credibility + dealmaking to build the AI-DC franchise), but the structure concentrates control and the prop book makes results founder-temperament-dependent. The Oct-2025 PIPE letting executives sell 3.75M shares @ $36 (near the high) is a modest governance yellow flag on insider alignment timing.

Lens 10 · Forensic Red Flags

Acting as a forensic analyst — labeled figures only.

1. Gross-up revenue accounting (the headline trap). $60.4B "revenue" is principal-basis turnover that nets to ~zero against $60.2B transaction expenses. Not improper (ASC 606 principal treatment), but it structurally obscures the real ~$426M of gross profit and invites both bull (huge "revenue") and bear (it's fake) misreads. KPMG flagged digital-asset existence/rights and Level-3 investment fair value as the two Critical Audit Matters — exactly the soft spots.
2. Impairment asymmetry / earnings quality. Digital intangibles carried at "lower of cost or observable fair value" mean down-moves hit as impairment ($753.7M FY25; $284.4M Q1-26) while up-moves are constrained — GAAP net income materially understates economics in down-quarters and is noisy throughout. Adjusted EBITDA is unaudited and self-defined (adds back settlement, mining impairment, reorg, derivative marks) — directionally useful but management-flattering; treat with caution.
3. Level-3 / mark-the-book risk. $0.7B current + $1.0B non-current investments, a chunk Level-3 (market approach, unobservable inputs — comparable-company multiples, volatility). Valuation is management judgment; in a crypto/venture downturn these marks lag and then catch down (as Q1-26's $233.6M investment markdown shows).
4. Concentration (counterparty + platform + DeFi). Two platforms = 38% of revenue; three counterparties = 45% of loans; Aave/Kamino/Spark = 70% of DeFi balance. Over-collateralized, but a smart-contract exploit or platform failure is a tail loss.
5. Leverage + covenant exposure. $3.0B notes payable: 2026 exchangeable ($445M due 2026, eff 7.0%), 2029 ($402.5M, eff 9.2%), 2031 ($1.3B, eff 0.9%), + Deutsche Bank Helios facility $878M/$1.4B, SOFR+4.75%, floor 2.5%, matures Aug-2028, DSCR ≥1.40 covenant, max loan-to-cost 80%. The Helios facility's covenants bind to data-center stabilization — a CoreWeave delay or tenant stress could pressure them.
6. Off-balance-sheet customer safeguarding. Obligated to safeguard customer digital assets not on the balance sheet; no loss accrued, but a custody failure is an unquantified contingent liability.
7. Opportunity-zone tax trap (timing). ~$634.8M contributed to QOZ subsidiaries in FY25; in Dec-2026, Galaxy must recognize capital gains on 90% of contributed amounts, potentially forcing additional tax distributions to NCI holders. A known, dated cash/tax event to model.
8. Dilution. Shares outstanding +39% YoY via LP-unit exchanges + the $478M and $460M raises; diluted share base ~366M. Persistent equity issuance funds the buildout but dilutes Class-A holders.

Regulatory findings (required sub-section).

• SEC LR / AAER: None. Verified via SEC EDGAR EFTS (LR + AAER), search window 2021-06-20→2026-06-20.
• 10-K Item 3 / Note 18 (Legal Proceedings — the company's own disclosure): LUNA Matters. On 2025-03-27 Galaxy settled civil claims with the New York State Attorney General over its investments, trading, and public statements on the LUNA digital asset (late-2020→2022). Accrued legal provision $151.0M (discounted at ~4.3%); undiscounted $160.0M, payable $40M (2026) / $60M (2027) / $60M (2028), plus compliance enhancements on crypto public statements and trading. Separately, a December-2022 proposed Ontario class action names GDH Ltd., the CEO, and the former CFO over alleged LUNA-related misrepresentations (secondary-market acquirers May-2021→May-2022); leave/certification scheduled to be heard June 2026 — outcome/impact not estimable.
• Non-SEC enforcement (web): No new material FTC/DOJ/CFTC/CFPB action surfaced beyond the NYAG LUNA settlement. The NYAG settlement is the one material, resolved matter; the Ontario class action is the live overhang.
• Net: the accounting is auditor-clean (KPMG unqualified) but the business model is mark-to-market-noisy and concentration-heavy; the one real legal scar (LUNA) is settled-and-provisioned, with a Canadian tail.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (EPS, next three fiscal years — FY2026E / FY2027E / FY2028E)

No forecast.ts create logged (watchlist/breadth mode — skip the Brier log per skill). All outputs ``, arithmetic shown; GLXY EPS is dominated by the crypto tape, so point EPS is low-confidence — the honest output is scenario ranges, not a number.

Building blocks: FD share base ~366M and rising ~5–10%/yr from LP exchanges/issuance; Digital Assets segment run-rate pre-tax ~$190–250M (FY25 actual $193.9M, FY24 $47M — call it $150–250M normalized, crypto-volume-dependent); Treasury & Corporate = ±$300–600M swing entirely on crypto marks (un-forecastable directionally); Data Centers ramps from ~$0 (FY25) toward the CoreWeave annuity — >$1B/yr average gross over 15yr once ~800 MW is energized, but FY26 contribution is small (Phase-I 133 MW partial-year), FY27 partial (Phase-II H1-27), FY28 the 526 MW substantially online; notes interest ~$60M+ and rising with Helios draws.

The honest synthesis: FY2026 GAAP EPS is most likely modestly negative-to-breakeven (the H1-26 crypto drawdown already booked Q1 losses; Q2 ~$90M AEBITDA guide helps but the year is hostage to BTC). The real EPS inflection is FY2027–2028 as Helios revenue converts — that is when the equity stops being "leveraged BTC" and starts being "AI-DC annuity + crypto kicker." Conviction on the base case is LOW precisely because half the P&L is an un-forecastable mark-to-market book. No tracked forecast logged.

Lens 12 · Bull vs Bear

Bull case. Galaxy is the cleanest crypto-to-AI-infrastructure conversion equity. The Digital Assets franchise is growing and profitable (segment pre-tax quadrupled FY24→FY25 to $193.9M) and throws off cash + a 1,600-counterparty institutional moat + KPMG-audited credibility — it is the funding engine. Bolted on top is a scarce, permitted, interconnected 1.63 GW (→3,400 MW potential) West-Texas power+land bank anchored by a 15-year CoreWeave lease worth >$1B/yr average — a quasi-utility annuity uncorrelated to crypto that the market is only beginning to capitalize. Management (Novogratz, ex-GS/Fortress) has the macro credibility and dealmaking to fund the $15B+ buildout (already $1.4B DB facility + two equity raises). As Helios revenue converts in FY27–28, GLXY re-rates from a volatile crypto-beta toward an AI-DC multiple — and the SOTP already looks compelling on ~$426M adjusted gross profit + the data-center NAV, which is why consensus is Buy (57% Strong Buy), avg PT ~$41.30 vs ~$34 spot. Earnings surprise to the upside: any crypto rally (à la Q3-2025's $505M print) is a free option on top of the infra story.

Bear case (permanent-impairment risks).

1. CoreWeave single-tenant concentration. The entire data-center thesis rests on one tenant's 15-year credit. CoreWeave is itself a highly-levered, AI-cycle-dependent, cash-burning neocloud. If CoreWeave restructures, renegotiates, or the AI-DC demand curve rolls over, Galaxy is left with a half-built, debt-financed campus and a $1.4B DB facility whose covenants bind to stabilization. This is the one risk that could permanently impair the buildout.
2. The earnings are un-ownable noise. Half the P&L is mark-to-market on a $4.5B crypto book + Level-3 venture marks. In a sustained crypto bear (BTC already −48-51% from ATH as of June-2026), the book bleeds, GAAP losses recur, dilution accelerates to fund Helios, and the equity de-rates on "another loss quarter."
3. Multiple already pricing the dream. At ~$34 (+45.8% TTM, near the upper-half of range) the stock has already re-rated on Helios optionality that hasn't produced a dollar of revenue. A Hold-rated SOTP puts probability-weighted fair value at $22–26 (15–28% downside) if you haircut CoreWeave execution. Piper Sandler cut its PT on valuation.

Pre-mortem (18 months out, thesis broke): Most likely failure path — CoreWeave Phase-II/III slips or is renegotiated amid an AI-capex digestion, Helios delivery dates push right, the DB facility covenant gets tight, and a crypto bear keeps the book in the red — so the "uncorrelated annuity" that justified the re-rating fails to show up on time while the crypto engine simultaneously stalls. The stock round-trips to the low-$20s. Secondary path: a custody/counterparty/smart-contract loss event detonates the concentration risk.

Contrarian view (what the market refuses to see): The bull crowd treats CoreWeave as a riskless annuity and the bear crowd treats GLXY as just levered BTC. Both miss that the two legs are correlated through the same macro variable — AI-capex enthusiasm and crypto risk-appetite are both long-duration-liquidity trades; a genuine liquidity/rate shock (the exact June-2026 setup: hawkish Fed + ETF outflows) hits both the crypto book and CoreWeave's demand at once. The "diversification" is less than advertised. Conversely, in a liquidity-easing regime both legs rip together — so GLXY is a higher-beta liquidity play, not a hedged barbell.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• Where revenue is concentrated: economically, on (a) a handful of trading platforms (38% of revenue via two) and (b) prospectively, one tenant (CoreWeave) for the entire growth narrative. Shift either and the story breaks.
• Why the moat is weaker than bulls think: crypto trading spreads are compressing as TradFi enters; the "institutional brand" moat is shared with Coinbase, and the data-center moat is unproven — Galaxy has never operated a large-scale AI DC and is a first-time developer competing against CoreWeave's own DC partners, IREN, Crusoe, and the hyperscalers' self-build.
• Most dangerous competitor bulls underestimate: CoreWeave itself — the tenant could in-source/self-develop future capacity, and every former bitcoin miner (IREN, Core Scientific, Riot, etc.) is racing into the same miner-to-AI conversion, commoditizing the "stranded-power-to-GPU" trade and compressing lease economics on future phases.
• Worst capital-allocation / governance items: prop-book value destruction in down-cycles; serial acquisitions of uncertain ROI; founder-controlled dual-class with no Class-B economics; related-party plane/boat; executives selling 3.75M shares into the Oct-2025 PIPE near the high; persistent dilution (+39% shares YoY).
• Assumptions that must hold for today's price (~$34, ~$12–13B FD): (1) CoreWeave performs for 15 years; (2) Phases I–III deliver on time and on budget within the DB covenant; (3) the AI-DC lease multiple the market is awarding survives AI-capex digestion; (4) the crypto book doesn't force a dilutive capital raise at a bad price.
• If growth disappoints 20–30% (CoreWeave slips a phase + crypto stays soft): SOTP compresses to the low-$20s — the Hold-case $22–26 fair value; in a hard AI-capex + crypto double-down, high-teens is plausible.
• Single scenario that permanently impairs: CoreWeave default/restructuring mid-build leaving Galaxy with a leveraged, under-leased campus and a covenant breach — low-probability near-term (CoreWeave is well-capitalized and AI demand is hot in mid-2026) but non-trivial and existential to the DC thesis over a 15-year horizon.

Lens 14 · Management Questions (ordered by information value)

1. CoreWeave concentration: what are the specific tenant-credit protections, security/letters-of-credit, and termination/step-in rights in the 15-year lease, and what happens to the Helios economics and the DB covenant if CoreWeave restructures or defers Phase II/III?
2. Helios delivery: are Phase I (full 133 MW), Phase II (~260 MW, H1-2027), and Phase III on schedule and on budget, and what is the current loan-to-cost vs the 80% covenant and the DSCR runway to the 1.40 floor at stabilization?
3. Capital plan: the $15B+ buildout — how much is equity vs project debt vs tenant prepayment, and at what BTC price / stock price would you not issue equity to fund it?
4. The other ~830 MW + 3,400 MW potential: what is the contracting pipeline beyond CoreWeave, and will you diversify the tenant base before energizing more capacity?
5. Prop book: what is the firm's risk-limit framework on the Treasury & Corporate digital-asset book, and would you ever hedge it to stop GAAP losses swamping a growing operating business?
6. Normalized economics: what is the right "through-cycle" Digital Assets segment pre-tax margin and AUM trajectory we should underwrite, stripped of crypto marks?
7. GalaxyOne: what are the unit economics, customer-acquisition cost, and 3-year revenue ambition, and is retail a real leg or an option?
8. Opportunity-zone gain (Dec-2026): quantify the expected capital-gain recognition on the ~$635M QOZ contributions and the resulting tax-distribution cash outflow.
9. Dilution: what is the expected fully-diluted share count trajectory through Helios stabilization, and how do you weigh Class-A dilution against deleveraging?
10. LUNA / Ontario class action: beyond the $160M NYAG settlement, what is the realistic range of outcomes on the Canadian class action heard June-2026, and are there other unresolved regulatory inquiries?
11. Counterparty/platform concentration: how are you reducing the 38%-of-revenue two-platform dependence and the top-3-borrower 45%-of-loans concentration?
12. Staking economics: the 5–10% retained take-rate on third-party staking — is that defensible as competition enters, and what is the AUM-staked growth runway?
13. Mining wind-down: is proprietary bitcoin mining fully exited, and what residual capex/impairment remains from the conversion?
14. Re-rating thesis: what milestones would you point to that should move GLXY from a crypto-beta multiple to a data-center/infrastructure multiple, and on what timeline?
15. Founder control: how should minority Class-A holders think about the dual-class structure, related-party arrangements, and succession given the founder-controlled vehicle?

---