Coinbase Global, Inc. (NASDAQ: COIN) — A Regulated Exchange Annuity Wrapped Around a Crypto-Beta Engine

An independent fundamental research note Report date: 2026-06-10 Price (as of 2026-06-10): ~$154 · Market cap: ~$40.5–48B · Enterprise value: ~$38–45B · 52-week range: $139.36 – $444.65 Sector / sub-sector: Financials — Capital Markets / Diversified Capital Markets (crypto exchange & onchain infrastructure) Fiscal year-end: December · CIK: 0001679788 · Shares outstanding: ~222M Class A + ~41M Class B (~263M total)

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information only — not investment advice. The analytical body that follows (Sections 1–15) takes no position and carries no price target; it analyzes valuation only as embedded expectations.

Verdict: HOLD / accumulate on weakness toward the mid-$120s and below. Not a short here. Conviction: medium. Directional zone: I’d be a committed buyer of this franchise in the $110–135 band (roughly 2.2–2.6× book and a level that prices in a real crypto winter), a willing holder in the $135–210 range, and a seller of new money above ~$260 absent a fresh crypto bull market plus proof that Subscription & Services has truly de-cyclicalized the P&L. At ~$154 the stock has already round-tripped ~65% off its $445 high and trades near trough-cycle pricing on book (~3.0× P/B, 24th percentile of its own history) — the easy short was at $400, not here.

Tag: “The toll booth on a road whose traffic you cannot forecast.”

Coinbase is the best-run, most-trusted, most-regulated franchise in crypto — and that is exactly why it is dangerous to misclassify. The bull wants to pay a CME/ICE multiple for it because Subscription & Services (stablecoin reserve income, staking, custody, Coinbase One) is now 44% of revenue and genuinely annuity-shaped. The bear correctly notes that the other ~56% is pure crypto-price beta, that reported net income is dominated by mark-to-market swings on a proprietary BTC/ETH book management keeps adding to, that the retail take rate (~1% vs Binance’s ~0.1%) is a documented melting iceberg, and that five-year peak-to-peak revenue is flat ($7.35B in 2021, $7.18B in 2025). Both are right. The honest synthesis: this is a wonderful, cash-generative, fortress-balance-sheet operator strapped to an earnings stream you genuinely cannot underwrite quarter-to-quarter, run by a founder who controls a majority of the votes, allocates capital pro-cyclically (he bought back $1.9B of stock at ~$303 right before a net-loss quarter), and runs a side bet on Bitcoin with shareholder money. The franchise quality is real and the valuation is no longer euphoric — but the price still embeds a normal-to-better crypto cycle, not the winter that the -31% Q1’26 revenue print suggests may be arriving. I want to own the toll booth; I want to own it after the cycle’s gravity has done more of its work, or once the S&S annuity has proven it can grow through a down-tape rather than merely soften less. The single fact that flips me bullish: two consecutive down-tape quarters in which S&S grows while transaction revenue falls — proof the diversification is real. The single fact that flips me bearish: a sustained take-rate step-down combined with a crypto-winter volume collapse while the multiple is still north of ~30× normalized earnings.

1. Executive Summary

Coinbase Global is the largest US-regulated cryptocurrency exchange and the most institutionally credible crypto-financial-services platform in the world. It serves three customer groups — consumers (retail trading, self-custody via the Base App), institutions (prime brokerage, custody, financing), and developers (the Coinbase Developer Platform) — across four exchanges (Coinbase Exchange, Coinbase International, Coinbase Derivatives, and the August-2025-acquired Deribit, the global leader in crypto options), the Base Ethereum Layer-2 network, and a stablecoin distribution franchise that captures roughly half of all USDC economics through a contractual revenue-share with Circle.

The business earns money two ways. Transaction revenue (59% of FY2025 net revenue, $4.06B) is volume-based trading fees — overwhelmingly retail/consumer ($3.32B) at a take rate roughly 10× Binance’s, plus a smaller, lower-fee institutional book ($0.48B). Subscription & Services revenue (41%, $2.83B) is the more durable layer: stablecoin reserve income ($1.35B, +48% YoY, COIN’s share of the Treasury-bill yield on USDC), blockchain rewards / staking ($0.68B), interest & finance income ($0.25B), and custody plus the Coinbase One subscription. The single most important structural trend in the company is the deliberate mix-shift toward S&S — 37% of revenue in FY2024, 41% in FY2025, and 44% in Q1 2026 — the management response to the central problem of the business.

That central problem is brutal cyclicality. Net revenue went $7.84B (FY2021) → $3.19B (FY2022) → $3.11B (FY2023) → $6.56B (FY2024) → $7.18B (FY2025), then fell 30.5% year-over-year in Q1 2026 to $1.41B, producing a $394M GAAP net loss. Across a full cycle (2021 peak to 2025) the top line is essentially flat — five years of zero net-revenue growth at the peaks — which establishes how much of historical “growth” was crypto-price beta rather than franchise expansion. Reported earnings are doubly distorted: at the operating line by two-directional volume leverage (operating margin swung from 35% in FY2024 to 20% in FY2025 to -1.5% in Q1 2026), and below the line by mark-to-market swings on a proprietary Bitcoin/Ether treasury (a +$687M gain in FY2024 became a -$529M loss in FY2025 and a -$482M loss in Q1 2026). Net income is not a usable run-rate metric for Coinbase; only normalized operating income is, and it just touched zero.

The competitive position is genuine but narrow. Coinbase’s durable advantages are a regulatory/compliance + trust intangible (≈80 licenses, a decade-plus operating record, ~12% of all crypto held in custody — “more than the next four competitors combined,” 12 consecutive quarters of native-unit inflows even as prices fall) and a contractual stablecoin annuity (the Circle Agreement auto-renews in three-year terms and “cannot be terminated” if thresholds are met). Its weaker claims are to retail “network effects” — switching costs for retail traders are low, and the ~1% retail take rate is compressing in plain sight (a documented -$384.4M FY2025 drag as users migrate to Advanced and zero-fee Coinbase One). The whole investment debate reduces to one race: can the durable S&S/institutional/derivatives layer scale faster than the retail spot-fee engine erodes?

Capital allocation is mixed-to-poor and governance is weak. Management financed opportunistically (zero-coupon convertibles issued into equity strength) and used rich stock as M&A currency (the $4.3B Deribit deal), but it also executed $1.9B of buybacks at an average ~$303/share into a cyclical peak, runs a proprietary crypto book that amplifies the P&L’s existing beta, and pays non-founder executives in large time-based RSUs weakly tied to per-share value. Brian Armstrong controls ~49.6% of the vote alone and a majority with affiliates — Coinbase is a controlled company, and external checks on all of the above are minimal.

At ~$154 the stock trades at ~3.0× book (24th percentile of its own history), ~6–7× EV/revenue, and a trailing P/E (~57–67×) that is uninterpretable at a cycle turn. The market is pricing a trough, not euphoria — but to justify the current EV against a sane (~20–25×) mid-cycle multiple, it must be underwriting roughly $1.6–2.4B of sustainable mid-cycle net income, above the FY2025 trough and near the FY2024 near-peak. This note takes no position and sets no price target. It identifies what must be true for those embedded expectations to hold, and what would falsify them.

2. Business Overview

What Coinbase is. Founded in 2012 and now headquartered in New York, Coinbase operates a vertically integrated stack of crypto-financial infrastructure under one consumer brand and a set of regulated entities (FY2025 10-K, Item 1). It is, at once: (i) a retail crypto brokerage and wallet — the Coinbase app and the self-custody Base App, through which ~9.2M average monthly transacting users buy, sell, send, stake, and hold crypto; (ii) an institutional prime broker, custodian, and financing desk serving asset managers, hedge funds, corporates, and the ETF issuers that need a qualified custodian; (iii) a derivatives venue, hugely expanded by the August 2025 acquisition of Deribit, the world’s largest crypto-options exchange by volume and open interest; (iv) the primary distributor of USDC, the second-largest dollar stablecoin, via a long-standing revenue-share with issuer Circle; and (v) the operator of Base, an Ethereum Layer-2 network, and the Coinbase Developer Platform (“crypto-as-a-service”), which sells wallet, custody, staking, and onchain rails to other companies.

How it makes money — two pillars and a tail. Coinbase reports revenue in two principal buckets plus a small “other” line. The structure and five-quarter trajectory are the core of the thesis (FACT, FY2025 10-K Note 4 “Revenue”; Q1 2026 10-Q):

A few facts in this table carry most of the weight:

• Consumer transaction is still the single largest line ($3.32B FY2025) — and it shrank 3% YoY despite a 7% rise in consumer trading volume. The gap is a -$384.4M drag from a “lower average blended fee rate” as customers migrate from the high-fee Simple interface to the low-fee Advanced platform and the zero-fee Coinbase One subscription (FY2025 10-K, MD&A). The melting iceberg is melting inside the filing.
• Institutional carries the volume, consumer carries the revenue. FY2025 institutional trading volume was $982B vs $239B consumer, yet institutional produced $480M of revenue vs consumer’s $3.32B — institutions pay a fraction of the retail take rate. This is the structural reason any “share gains in volume” must be discounted: not all volume is created equal economically.
• Stablecoin revenue is now the growth engine of the durable layer, up 48% to $1.35B and ~19% of net revenue. It is COIN’s contractual share of the reserve income (essentially the Treasury-bill yield) on USDC in circulation, both on- and off-platform, paid by Circle. FY2025 growth came from higher USDC balances (+$417.7M on-platform, +$314.1M off-platform) partly offset by -$290.8M from an 89bp decline in rates — the rate sensitivity is explicit and material.

Key performance indicators (FACT, FY2025 10-K): average Monthly Transacting Users (MTUs) of 9.2M (+10% YoY); Assets on Platform of $376B at year-end (-7% YoY, price-driven; management cited >$500B at the Q3 2025 peak); total trading volume of $1,221B (+3%); USDC on platform of $9.26B (+52%). Management emphasizes 12 consecutive quarters of net native-unit inflows — i.e., customers add more coins even when the dollar value of assets falls on price — as evidence the franchise widens through down-tapes.

Recurring vs. cyclical. The S&S layer is genuinely more recurring than transaction fees — stablecoin reserve income is contractual and auto-renewing, Coinbase One has >1M paid subscribers, custody fees scale with assets. But “recurring” overstates the decoupling. Stablecoin revenue is a joint function of balances (sticky) and interest rates (exogenous and now falling); staking rewards move with crypto prices and protocol reward rates; interest income is rate-levered. This is a cyclical trading business with a real but macro-exposed annuity overlay — structurally similar to Robinhood’s three-line model (comparable retail-brokerage disclosures), but more crypto-pure: where crypto is ~20% of Robinhood, it is essentially 100% of Coinbase.

Verdict. Coinbase is a crypto-cycle-levered transaction franchise actively and credibly diversifying into a stablecoin/services annuity. The diversification is real and accelerating (S&S 37% → 41% → 44% across five quarters), and the unit-growth metrics show a franchise that genuinely widens in down markets. But the annuity is itself rate- and crypto-price-sensitive, so “diversified” overstates how decorrelated the P&L actually is — a point Q1 2026 proved when S&S itself fell 16% sequentially.

3. Industry Dynamics

Structure. The crypto-exchange industry is a globally fragmented oligopoly with a concentrated top and a long competitive tail. Binance is the global #1 by spot volume — a multiple of Coinbase’s global volume, charging ~0.1% fees — operating largely outside US regulation. Coinbase is the dominant US-regulated venue and a growing global #2/#3. Kraken, Gemini, Bitget, OKX, and Bybit round out the major exchanges. But the more important competitive story is adjacent entrants: Robinhood (crypto ~20% of its revenue; it acquired Bitstamp and WonderFi to deepen crypto), PayPal (PYUSD stablecoin), and the traditional-finance giants — BlackRock and Fidelity (whose spot Bitcoin and Ether ETFs directly disintermediate the “buy-and-hold” use case that was historically Coinbase’s highest-margin retail flow), plus banks and brokers entering post-regulatory-clarity. Coinbase itself flagged in Q1 2026 that “Morgan Stanley and other TradFi banks are offering better prices on their brokerages.”

Cyclicality is the defining structural feature. Net revenue has now suffered two ~60% peak-to-trough collapses in five years (2021→2022 and the rollover underway in 2026). Management’s own framing: “our revenue is inherently nonlinear. A significant portion moves in line with crypto asset prices and trading volumes” (Q1 2026 call). No amount of product diversification has yet broken this — the Q1 2026 print (-31% YoY, GAAP loss) is the latest proof.

Fee compression is the central structural threat. The migration of Coinbase’s own retail users toward Advanced and zero-fee Coinbase One is documented in the FY2025 filing as a -$384.4M revenue drag. Coinbase’s blended retail take rate sits an order of magnitude above Binance (~0.1%) and approaches infinity above TradFi brokers offering crypto at zero. Management’s defense — “our clients are not choosing us because we’re the cheapest… [but because we’re] most trusted, easiest to use” (Haas, Q1 2026) — is the classic premium-pricing argument, and management itself concedes “over the long term… fees could come down as things become commoditized.” The premium is real but eroding.

Regulatory landscape — a genuine regime shift. This is the most important change in the industry in a decade, and it cuts in Coinbase’s favor:

• The 2023 SEC enforcement action alleging Coinbase operated as an unregistered securities exchange — a multi-year existential overhang — was dismissed in 2025 under the new SEC posture (FACT).
• The GENIUS Act (stablecoin regulation) became law in mid-2025, providing a federal framework for dollar stablecoins like USDC.
• The CLARITY Act (crypto market structure) is, per management, heading toward markup and a likely signed law “by the end of summer” 2026, with a Tillis/Alsobrooks compromise protecting activity-based stablecoin rewards (Grewal, Q1 2026).
• The CFTC cleared Coinbase (and Kalshi) for US crypto futures/derivatives in May 2026.

This is a transition from regulation-by-enforcement to a defined framework — a clear positive for the regulated-incumbent thesis, because compliance cost is a barrier to entry that protects scaled, licensed players. The caveat (the same one the HOOD report flags): the tailwind is administration-dependent and reversible, and the public Armstrong-vs-Jamie-Dimon spat over the CLARITY Act (May 2026) is a reminder that powerful incumbents are fighting the framework.

Marathon capital-cycle read. Crypto exchange and onchain infrastructure is a textbook late-cycle capital-flooding setup: high returns at the 2021 and 2024–25 tops drew in TradFi entrants (BlackRock, Fidelity, PayPal, Morgan Stanley), new venues, and a wave of capital into stablecoins and tokenization. Marathon’s supply-side lens predicts mean-reversion of returns as capacity catches up — and the fee-rate decline already visible in Coinbase’s filings is precisely that signal arriving. Regulation distorts the cycle in Coinbase’s favor (compliance as a moat), but the parts of the business attracting the most new capital — stablecoins, tokenized equities — are exactly where competitive return-erosion is most likely.

Verdict: a structurally challenged industry with a structurally attractive regulated niche. The core spot-trading market is cyclical, fee-compressing, globally low-margin (Binance economics), and under assault from zero-fee TradFi and ETF disintermediation — a bad industry on Greenwald/Marathon terms. The redeeming feature is the regulated US/institutional/custody sub-segment and the stablecoin reserve annuity, where compliance barriers and a USDC near-duopoly (Coinbase + Circle) create defensible economics. Net: bad at the trading core, good in the regulated/stablecoin layer — and the investment case is a bet on which one dominates the future mix.

4. Competitive Position

Coinbase has a genuine but narrow moat, and naming the mechanism precisely in Greenwald’s taxonomy matters because two of the three claimed advantages are real and one is largely marketing.

1. Regulatory/compliance + trust intangible — the real moat. Coinbase holds ~80 licenses, has a decade-plus compliance track record, works with five G-SIB banks and ~150 government agencies, and custodies ~12% of all crypto in existence — “more than the next four competitors combined” (Q4 2025 call). In Greenwald’s framework this is a demand-side captivity + intangible (brand/regulatory) advantage: customers trust Coinbase with custody and consolidate to it in stress. The financial fingerprint is visible and falsifiable — Coinbase sustains a ~1%+ retail take rate that Binance structurally cannot charge, and it gains share in down markets (all-time-high trading market share in the down Q1 2026 tape; 12 straight quarters of native-unit inflows). “When conditions are difficult, people go to where they trust” (Armstrong, Q1 2026). This is the most durable piece because compliance is a rising barrier to entry that the GENIUS/CLARITY regime is actively codifying. It ties to a financial outcome that would deteriorate without it (pricing power and counter-cyclical inflows) — so it qualifies as a moat.

2. The stablecoin/USDC annuity — a structural, contractual moat. The Circle Agreement gives Coinbase a share of reserve income on all USDC in circulation, on- and off-platform. Critically (FACT, FY2025 10-K; Q1 2026 call): the agreement auto-renews into perpetuity in three-year terms and “cannot be terminated” so long as product/company/reseller thresholds are met, and the revenue share is tied to USDC supply, unaffected by the rewards-legislation fight. Coinbase captures ~50% of all USDC economics, and >25% of USDC is held in its products. This is the closest thing to a true annuity Coinbase owns — but it is leveraged to two things outside the company’s control: interest rates (-$290.8M of revenue in FY2025 from an 89bp rate decline) and USDC’s share versus Tether and a growing field of bank- and PayPal-issued stablecoins.

3. Liquidity “network effects” — the weakest claim, largely marketing. Management asserts a “powerful network effect” from “pooled global liquidity.” Pressure-tested, this is thin for retail: switching costs are low (a user can open a Kraken or Robinhood account in minutes; the only friction is moving custodied assets), and spot liquidity is globally fragmented with Binance — not Coinbase — holding the deepest global books. It is moderately real for institutional/derivatives, where Deribit’s options open-interest concentration is a genuine liquidity moat. The retail network-effect claim does not tie cleanly to a financial outcome and should be treated as marketing, not moat.

Is the ~1% retail take rate a melting iceberg? Yes — but melting slowly, and the entire investment debate is the speed of the melt versus the speed of the replacement. Eroding: the filing-documented -$384.4M blended-fee drag, the zero-fee Coinbase One tier, and TradFi offering “better prices.” Resilient: Coinbase still posts a take rate ~10× Binance’s and gains share in down markets, evidence of trust-based pricing power. The honest read is that Coinbase is racing to replace the eroding retail premium with (a) higher-margin new asset classes (derivatives, prediction markets, perpetuals) and (b) the stablecoin annuity before it fully melts. That race is the thesis.

Direct comparison:

• vs Binance: loses on global scale, volume, and fees; wins on US regulation, trust, institutional access, and being a listed, investable, audited company.
• vs Kraken/Gemini: wins clearly on scale, brand, product breadth, and balance sheet.
• vs Robinhood: Robinhood is broader (options is its largest line, plus equities, with ~88% PFOF economics) and cheaper, but crypto is only ~20% of HOOD; Coinbase is the crypto-depth, institutional, custody, and derivatives leader. The two are converging — both chasing “trade everything” (Coinbase’s “Everything Exchange” vs Robinhood’s multi-asset app) and tokenized equities.

Verdict: a durable advantage in the regulated/custody/stablecoin layer; an eroding premium in retail spot trading. The moat is real and financially visible where it can be tied to outcomes (counter-cyclical inflows, ~1% take rate, the locked Circle annuity, ~12% of world crypto custodied). It is weak-to-illusory where management leans hardest on retail “network effects.” The thesis hinges on whether the durable layer scales faster than the retail premium erodes — and Q1 2026 (S&S -16% QoQ) is a warning that even the durable layer is not as decoupled as advertised.

5. Growth History and Forward Opportunities

The whipsaw, and the single most damning fact. Net revenue ran $7.35B (2021) → $3.15B (2022) → $3.10B (2023) → $6.56B (2024) → $7.18B (2025), and is rolling over again (-31% YoY in Q1 2026). Over a full cycle — 2021 peak to 2025 — total revenue is essentially flat. Five years of zero net-revenue growth at the peak-to-peak level is the most damning single fact about the “growth” story, and it confirms how much of historical “growth” was crypto-price beta rather than durable franchise expansion. Any forward growth narrative must clear this bar.

Underneath the beta, there is real secular progress. This is what separates Coinbase from a pure crypto-beta vehicle:

• S&S revenue compounded durably: $1.41B (2023) → $2.31B (2024) → $2.83B (2025), +23% in FY2025 and, per management, “5.5× the prior-cycle peak in 2021.” This is the genuinely growing, less-cyclical layer.
• Stablecoin grew 48% in FY2025 to $1.35B — the standout line.
• MTUs +10% and 12 consecutive quarters of native-unit inflows — the unit base grows even when dollar Assets-on-Platform fall on price. This is the strongest single piece of evidence the business is more than price beta.
• Revenue breadth: 12 products at >$100M annualized run-rate, half at >$250M. Derivatives reached >$200M annualized run-rate; prediction markets (via a Kalshi partnership) hit $100M annualized within two months of launch.

Organic vs acquired. The FY2025 institutional growth was substantially acquired — the +$152M derivatives uplift was “due mainly to the acquisition of Deribit” (Aug 2025), and Echo (onchain capital formation) was a Q4 2025 tuck-in. Stablecoin, Base, MTU growth, and Coinbase One are organic.

Forward opportunities (management framing — treat as hypothesis):

1. The “Everything Exchange” — tokenized stocks (~10,000 tickers rolling out), equity perpetuals (20× leverage, ex-US), prediction markets, commodities futures (gold/silver/oil, +4× QoQ volume), DEX trading of millions of tokens, and token sales (the Monad sale was oversubscribed at $269M). This is the explicit strategy to replace eroding spot-crypto fees with new asset-class fees.
2. Stablecoins / payments — USDC scaling, Base as a settlement layer, and the CDP / x402 agentic-commerce protocol (99% of x402 transactions settle in USDC, 90%+ on Base). The agentic-payments narrative is early and unproven but strategically coherent.
3. Derivatives — Deribit integration (expected complete in 2026) and US crypto options (“actively working on it,” no firm timeline).
4. Institutional — prime custody, ETF custody (including staking ETFs), and financing (average daily loan balances at an all-time-high $1.4B in Q1 2026).
5. Tokenization / RWA — management cites real-world-asset tokenization growing toward ~$16T by 2030, with 45 major financial institutions moving tokenization “from concept to production” in Q1 2026.

Is forward growth durable or just crypto beta? Mixed, improving. The S&S/stablecoin/MTU/native-inflow trends are evidence of some durable, less-cyclical growth — this is not pure beta. But the burden of proof is high (flat five-year peak-to-peak revenue); the new Everything-Exchange products are still small and many are regulation- or integration-gated; the largest FY2025 institutional growth was bought (Deribit), not organic; and the best durable line (the stablecoin annuity) is rate-exposed, with falling rates a direct headwind into 2026. Management’s Q2 2026 restructuring (a reduction-in-force, a ~$500M cost target, an “AI-native” pivot, and a $50–60M restructuring charge) signals that management itself expects a tougher near-term revenue regime.

Verdict: medium-quality growth, trending higher-quality. The transaction core is low-quality growth (crypto beta, fee-compressing, flat over a full cycle). The S&S/services layer is genuinely higher-quality and compounding, and the unit metrics show the franchise widening even in down markets. The decisive question is whether the durable layer can out-grow the structural erosion of the retail spot-fee engine plus the rate headwind — unproven, but the trend (S&S 37% → 44% in five quarters, 12 products >$100M) is moving the right way.

6. Financial Quality

Multi-year P&L (FACT, EDGAR XBRL reconciled to the 10-K Consolidated Statements of Operations):

The single most important read: revenue is wildly cyclical and net income is dominated by below-the-line crypto marks, not operations. Q1 2026 revenue fell 30.5% YoY and the company swung to a $394M net loss despite an essentially breakeven operating result (-$21M) — the loss was almost entirely a $482.4M mark-to-market loss on crypto held for investment (Q1 2026 10-Q). FY2025 net income of $1.26B was less than half of FY2024’s $2.58B even though revenue grew 9.4%; the swing came from a $528.9M loss on crypto held for investment in FY2025 versus a $687.1M gain in FY2024 — a ~$1.2B negative swing in a single non-operating line. Net income is not a usable run-rate metric for Coinbase; it is crypto-price beta wearing an income-statement costume. Normalize it out before any valuation conclusion.

Operating expense structure — hardening, not scaling (FACT, 10-K):

Three flags. (1) Sales & marketing tripled FY2023→FY2025 ($332M → $1,059M), driven by USDC rewards — the pass-through paid to users to hold USDC — which rose from $35M to $441M. This is the cost of “buying” the stablecoin float: a customer-acquisition subsidy that scales linearly with the float, not a fixed cost that delevers. It is a rebate war, not a moat. (2) Technology & development jumped to $526M in Q1 2026 alone (vs $355M in Q1 2025, +48%), largely Deribit integration and headcount — the wrong direction on a -30% revenue quarter, and the proximate trigger for the Q2 2026 restructuring. (3) G&A rose to $1.62B, including legal settlements.

The consequence is the central financial verdict: operating margin is two-directional volume leverage, not durable scale economics. Transaction revenue is ~95% incremental margin, so when volumes are high (FY2021, FY2024, H1 2025) operating leverage looks spectacular; when volumes fall (FY2022, FY2023, Q1 2026) the fixed-ish cost base plus the variable USDC subsidy crush margins. The leverage is real but symmetric — it is crypto-price beta, not a compounding-margin business.

Cash flow and the custodial distortion. Operating cash flow was -$1.59B (FY2022), +$0.92B (FY2023), +$3.10B (FY2024), +$2.43B (FY2025), and +$0.18B (Q1 2026). But these headline figures are heavily distorted by customer custodial funds ($5.35B at FY2025, matched by an equal custodial liability) that gross up both sides of the balance sheet and inject swings unrelated to corporate economics. Any OCF read must be done net of custodial flows; the headline numbers are not corporate free cash flow. Net income and OCF diverge sharply because crypto marks are non-cash and SBC is large — so in the Q1 2026 down-mark quarter OCF was positive (+$183M) while net income was negative (-$394M). This is the reverse of a typical quality-of-earnings red flag, but it underscores that neither metric is clean.

Balance sheet (FACT, FY2025 10-K / Q1 2026 10-Q):

The balance sheet is a fortress on liquidity and increasingly levered on debt. Debt is ~$7.2–7.5B aggregate principal, almost entirely opportunistic financing: ~$1.74B of senior notes (3.375% 2028, 3.625% 2031), and a stack of zero- or near-zero-coupon convertibles — $1.27B 2026 converts (0.5%, due June 2026), $1.27B 2030 converts (0%), and $1.50B each of 2029 and 2032 converts (0%, issued Aug 2025). Conversion prices are high (e.g., the 2029 converts at ~$454/share), so dilution risk is remote at current prices, but the 2026 converts ($1.27B) are a near-term maturity, likely refinanced or settled given the cash position. Book value per share is ~$51 (equity $14.79B against ~263M shares; it fell to $13.48B at Q1 2026 on the net loss). ROE is ~6.7% TTM — a poor structural return for a “platform,” and almost entirely a function of where Bitcoin traded in the quarter. ROIC is not meaningfully computable (the denominator is inflated by ~$5.6B of Deribit goodwill+intangibles and the numerator is crypto-mark-distorted) — state it as not meaningful rather than fabricate precision.

Dilution and SBC. SBC was $821M (2021), $1,566M (2022), $781M (2023), $913M (2024), $839M (2025) — ~12% of revenue, high but trending down as a percentage from the 2022 peak. Gross shares grew ~5.6% in FY2025 (to ~267.8M issued), driven by the Deribit stock consideration and SBC, partly offset by buybacks.

Quality-of-earnings flags: (1) crypto marks dominate net income — normalize them out; (2) goodwill/intangibles ballooned from ~$1.19B to ~$5.57B on Deribit, creating future impairment risk if crypto-derivatives volumes disappoint; (3) FY2025 “Other operating expense, net” jumped to $356M (from $8M), composition partly Deribit/legal — an open question; (4) a large FY2024 deferred-tax benefit flattered that year, and deferred-tax assets fell from $941M to $571M; (5) the 19%-of-revenue single counterparty (Circle/USDC) is a concentration the income statement does not flag.

Verdict. Economics do not durably improve with scale. The apparent operating leverage is bidirectional volume leverage on a crypto-price-correlated revenue base, amplified by a hardening fixed-cost stack and a USDC-rewards subsidy that scales with the float. Reported profitability is dominated by non-operating crypto marks. This is high-quality infrastructure attached to a low-quality, beta-driven earnings stream; normalized operating income is the only honest base, and it just went to zero in Q1 2026.

7. Capital Allocation

M&A. The headline event is Deribit, acquired August 14, 2025 for total consideration of $4.295B ($721.5M cash + $3.573B in Class A stock, with $150M of cash in a 15-month indemnity escrow). The purchase-price allocation added ~$2.82B of goodwill and ~$1.35B of customer-relationship intangibles. This is a defensible strategic move — Deribit is the global crypto-options leader and a genuine liquidity moat — executed with a sensible currency (rich stock). But two cautions: the headline ~$2.9B announced price rose to $4.3B as Coinbase stock appreciated between signing and close, meaning shareholders paid more in real economic terms than advertised; and it loaded ~$4.2B of goodwill+intangibles onto the balance sheet for a derivatives business whose volumes are themselves crypto-cyclical (impairment risk). Smaller historical deals (One River, FairX, Bison Trails, Agara, Echo) were minor by comparison.

Buybacks — the standout capital-allocation flag. The board authorized a $1.0B repurchase (Oct 2024), raised to $2.0B (Oct 2025). By March 31, 2026, $1.9B had been utilized to repurchase only 6,278,390 shares — an average price of ~$303/share. Coinbase deployed essentially the entire authorization buying its own high-beta stock near cyclical/valuation highs, funded by a balance sheet it was simultaneously levering with convertibles, right before a -30% revenue quarter and a net loss. This is the textbook pro-cyclical capital-allocation error — buying maximum stock at peak optics rather than husbanding capital for the trough when the same shares trade at half the price. There is no common dividend.

Proprietary crypto treasury. Coinbase holds $2.0B of crypto for investment (up from $1.55B), “mainly Bitcoin and Ethereum… we have actively increased our investment in Bitcoin since [Q1 2025].” This book produced the +$687M gain (FY2024), -$529M loss (FY2025), and -$482M loss (Q1 2026) that dominate reported net income. Coinbase is running a proprietary, undiversified directional crypto bet on the corporate balance sheet — a mini-MicroStrategy layered on top of the operating business. For an analyst this is a capital-allocation negative: it amplifies the very beta that already dominates the P&L, reduces transparency, and is not a use of shareholder capital that builds the customer franchise.

Compensation and incentive alignment (2026 DEF 14A). Base salaries are modest (Armstrong $1.0M, Choi $0.875M, Haas/Grewal/Brock $0.730M). Armstrong takes no new equity — his 2025 total comp of $9.71M was $1M salary plus $8.71M of “all other” (overwhelmingly personal security) — because he already controls the company and holds large legacy option positions. Other NEOs are paid overwhelmingly in large, time-based RSUs, not performance-gated PSUs: Choi’s 2025 total was ~$22.2M (incl. a ~$20.0M RSU grant), Grewal ~$20M, Haas ~$14.4M. Alignment is therefore via raw stock-price exposure rather than ROIC or per-share-value metrics — and for a business this beta-driven, RSU-heavy comp pays out richly in up-cycles regardless of capital efficiency. Weak incentive design.

Founder control. Per the 2026 proxy, Armstrong controls ~49.6% of total voting power alone (via super-voting Class B), and a majority with affiliated trusts — Coinbase is a controlled company. Public shareholders have negligible governance influence over the board, comp, the crypto-treasury policy, the buyback timing, or M&A. That is a structural governance discount that must be weighed against any quality argument.

Verdict. Capital allocation is mixed-to-poor. Pluses: opportunistic zero-coupon convertible financing during strength, and using rich stock as M&A currency (Deribit). Strong minuses: (1) pro-cyclical buybacks (~$1.9B at ~$303/share into a downturn); (2) a proprietary BTC/ETH treasury that amplifies the P&L’s existing beta and caused the Q1 2026 loss; (3) time-based RSU comp weakly tied to per-share value; (4) a founder-controlled governance structure that removes the external check on all of the above. Management has not demonstrated counter-cyclical capital discipline — the single most important capital-allocation virtue for a cyclical business.

8. Changes and Headwinds — Last Two Years

Strategic and corporate changes:

• S&P 500 inclusion (effective ~May 19, 2025) — a major index/demand event that broadened the shareholder base and likely contributed to the run toward the $445 high.
• The $4.3B Deribit acquisition (closed Aug 2025) — the largest deal in company history, transforming Coinbase into the global crypto-options leader and adding ~$4.2B of goodwill+intangibles.
• $3.0B of new zero-coupon convertibles issued (Aug 2025) plus the 2030 converts (Mar 2024) — an aggressive, opportunistic terming-out of cheap debt during equity strength.
• The “Everything Exchange” pivot — tokenized equities, perpetuals, prediction markets (Kalshi), commodities futures, and DEX trading, all launched/expanding through 2025–26 to diversify away from spot-crypto fees.
• A Q2 2026 reduction-in-force and “AI-native” restructuring (~$500M cost target, $50–60M charge) — management bracing for a softer near-term revenue regime.

Regulatory and legal developments (net-positive but two-edged):

• The 2023 SEC enforcement action was dismissed in 2025 — removing a multi-year existential overhang.
• The GENIUS Act (stablecoins) became law (2025); the CLARITY Act (market structure) is expected to be signed by late summer 2026; the CFTC cleared US crypto futures (May 2026).
• A mid-2025 security/extortion incident (an insider-bribery data-access event) — a reminder that custody/security is the existential operational risk for a trust-based franchise.

Market headwinds:

• The crypto tape rolled over in early 2026, taking Q1 revenue down 31% YoY and the stock down ~65% from its high — the defining near-term headwind.
• Take-rate compression continued (the -$384.4M FY2025 blended-fee drag).
• Falling interest rates are a direct headwind to the stablecoin annuity (-$290.8M from -89bps in FY2025).

Verdict. The last two years simultaneously strengthened the structural franchise (regulatory clarity, S&P 500 status, Deribit/derivatives leadership, the S&S mix-shift) and exposed its cyclical fragility (the Q1 2026 revenue collapse and net loss, pro-cyclical buybacks, the treasury-book drag). On balance the changes strengthen the long-term franchise while doing nothing to resolve — and arguably amplifying, via the treasury book — the near-term earnings cyclicality.

9. Risk Analysis

Catastrophic / total-loss risk. A true total loss is low-probability given the cash fortress and the dominant franchise. The two non-linear tail risks are a catastrophic custody breach and a structurally adverse regulatory reversal — either could permanently impair the trust moat. The more probable severe scenario is not insolvency but a prolonged crypto winter compressing transaction and S&S revenue simultaneously, exposing that the “diversification” is correlated and re-rating the stock toward a cyclical-trough multiple.

10. Valuation Discussion — Embedded Expectations

The core problem. Coinbase’s reported earnings whipsaw violently with the crypto cycle (FY2024 $2.58B → FY2025 $1.26B → Q1 2026 a $394M loss). Consequently the trailing P/E (~57–67×) and forward P/E (~31× on some data providers vs ~79× on others — the divergence is itself the tell) are functionally uninterpretable. A single-year P/E for Coinbase prices a moment in the crypto cycle, not the business. The valuation question must be reframed onto cycle-robust anchors and a normalized earnings base.

Peer comp table:

*The ~31× vs ~79× forward-P/E divergence across data providers reflects wildly different sell-side estimates for a cycle-turn year.

The comp set frames the central tension precisely. The bull wants Coinbase valued like CME/ICE/CBOE — a regulated-exchange annuity at ~19–22× earnings, ~17–20× EV/EBITDA, with 60%+ margins and recurring data/clearing revenue. But those venues earn the multiple on structurally recurring, low-cyclicality revenue. Coinbase’s S&S is heading that way, but the bulk of profit is still transaction revenue — pure volume/price beta. Coinbase therefore sits awkwardly: it trades at an exchange-like-or-richer multiple (~36–38× EV/EBITDA) on brokerage-or-worse earnings cyclicality. On the crypto-beta axis (MSTR/IREN), Coinbase is far higher quality — a real, cash-generative operating business, not a levered NAV vehicle — but it shares the same fundamental driver: the BTC/crypto price level.

Cycle-robust anchors. On an own-history valuation index (2026-06-09), Coinbase’s P/E sits at the 86.6th percentile (expensive) but its P/B at the 23.7th percentile (cheap) and P/S at the 36.3rd, composite 48.9th. The split is the whole story: on earnings it looks expensive (because earnings are trough-ish), but on book and sales it looks cheap-to-mid versus its own history. At ~$154 the stock is -65% from its $445 high and below both its 50-day (~$190) and 200-day (~$250) moving averages. P/B ~3.0× is the most cycle-robust anchor, and it says the market is pricing a trough, not a bubble.

Sum-of-the-parts. The right frame separates the cyclical trading engine from the more-durable S&S annuity. S&S is now ~44% of net revenue (~$2.83B FY2025), and the bull case assigns it an exchange-data-like multiple — justified by the contractual, auto-renewing Circle annuity, >1M Coinbase One subscribers, and custody fees — such that a large share of the ~$40–48B market cap can be “explained” by S&S alone, with the trading franchise valued as a cheap option on the next crypto up-cycle. The bear’s rebuttal: S&S is not truly decoupled — stablecoin revenue is a function of crypto risk appetite and short-term rates and flows through a third party (Circle) — so it is annuity-shaped but still macro-levered, as the Q1 2026 -16% QoQ S&S print demonstrated.

Embedded expectations. Take EV ~$38–45B against a mid-cycle earnings base. To justify the current EV at a ~20–25× mid-cycle P/E (an exchange-like multiple, generous given the cyclicality), the market must be underwriting roughly $1.6–2.4B of sustainable mid-cycle net income — meaningfully above the FY2025 trough ($1.26B) and near-to-above the FY2024 near-peak ($2.58B). Translated to revenue at Coinbase’s ~25–40% incremental margins, that implies a mid-cycle revenue base in the high-$7B to ~$9B+ range (vs $7.18B FY2025, ~$6.3–6.6B TTM) — meaning the market is paying for growth and de-cyclicalization, not merely a cycle rebound.

What the market appears to be pricing correctly: that Coinbase is the dominant, most-trusted US regulated venue gaining global share; that S&S/stablecoins genuinely lower revenue beta over time (the 44% mix-shift is real); and that the franchise is trough-discounted, not euphorically priced (P/B 24th percentile).

What the market may be pricing incorrectly (either direction): bull risk — that take rates hold and S&S is a true rate-independent annuity (if GENIUS/CLARITY constrains stablecoin reward economics or rates fall, the annuity shrinks); bear risk — that the market under-credits the optionality of the Everything Exchange (derivatives at ~$200M ARR, prediction markets ~$100M ARR within two months, Deribit options, tokenization) — genuinely new, higher-margin lines that did not exist 18 months ago.

Scenario framing (explicit crypto-cycle assumptions; no price target):

The current ~$40–48B sits between Base and Bull — it requires a normal-to-better cycle and continued S&S/Everything-Exchange execution. It does not embed a true crypto-winter Bear (in which fair value would be materially lower) and it does not embed a euphoric Bull (the book multiple is only ~3×). No price target — this is the embedded-expectations read only.

11. Variant Perception

Consensus. The Street is split and de-rating into the cycle turn: an analyst rating of 3.69 (13 buy / 14 hold / 2 sell), a mean target of ~$231 (recorded only as a third-party signal), and a fresh Baird cut to $142, Neutral (2026-06-05). The prevailing view is “great franchise, but earnings are falling and we cannot multiple them.” The stock is ~65% off its high precisely because consensus lost conviction on near-term earnings, not on the franchise.

Strongest bull case — “the AWS of crypto / the primary financial account.” (1) Regulatory clarity unlocks TAM — CLARITY expected signed by late summer 2026, GENIUS already law; management likens the coming wave to GENIUS’s aftermath, with “a couple hundred large companies” integrating crypto and buying Coinbase’s custody/CDP rails. (2) S&S + derivatives + Base de-cyclicalize the mix — S&S already 44%, stablecoin revenue scales with USDC supply, Coinbase One >1M subs is a true subscription annuity, and the Everything Exchange added $200M+ (derivatives) and $100M+ (prediction markets) of ARR in under a year. (3) Trust moat — all-time-high market share in a down quarter, ~12% of all crypto custodied, 12 products >$100M ARR. (4) Optionality — tokenized equities, agentic commerce (x402), DeFi.

Strongest bear case — “crypto-price beta dressed as a platform.” (1) Terminal take-rate compression — retail take rates an order of magnitude above zero-fee/low-fee competitors; management’s own “fees could come down as things become commoditized” is an admission, and zero-fee Coinbase One is itself the compression vector. (2) Revenue is still crypto-price beta — Q1 2026 revenue -31% YoY, the GAAP loss driven by crypto marks, and much of S&S (stablecoin balances, staking, interest) co-moves with crypto risk appetite, so the de-correlation is overstated in a true winter. (3) Stablecoin economics depend on Circle + rates + regulation — ~50% of USDC economics is a revenue-share with a third party, and a large slice is net-interest income on reserves; GENIUS’s limits on passive yield and any rate-cut cycle both compress it. (4) Cyclical-peak risk — if FY2024 was near peak earnings, normalized earnings are well below the optics and the franchise must grow into an exchange multiple from a shrinking base.

The assumptions that matter most, and what falsifies each side:

Short interest as a signal. 25.6M shares short, ~11.7% of float, short ratio 2.4. Elevated for a large-cap but not an extreme crowded short, and the low 2.4-day cover ratio means the position is liquid — not a coiled squeeze. The skew says “contested, leaning skeptical,” consistent with the 14 holds + 2 sells. An open question is how much of the short is fundamental versus convertible-arbitrage hedging against the converts.

12. Fact vs. Interpretation Table

13. Open Questions

1. How much of stablecoin revenue survives a Fed cutting cycle? Each 100bps of rate cuts is a direct hit to a now-19%-of-revenue line; the FY2025 -$290.8M from -89bps quantifies the sensitivity.
2. Will GENIUS/CLARITY rulemaking constrain the stablecoin reward economics that underpin the Circle revenue-share, and could Circle renegotiate the split as USDC scales?
3. What is the true mid-cycle normalized operating income? FY2024 ($2.31B op income) was near-peak; Q1 2026 was zero. The honest base is somewhere between, and it determines whether the stock is cheap or expensive.
4. Does the Everything Exchange (derivatives, perps, prediction markets, tokenized equities) reach material scale, or do these remain sub-$500M curiosities gated by regulation and integration?
5. Composition of the FY2025 $356M “Other operating expense” spike (Deribit integration? legal settlements?) and the status/cost of any residual legal matters.
6. How much of the 11.7% short interest is fundamental conviction vs convertible-arbitrage hedging?
7. Will management ever run the balance sheet counter-cyclically — i.e., buy back stock in the trough rather than the peak, and stop adding to the proprietary BTC book?

14. What Must Be True (Bull and Bear, each with a falsification test)

For the BULL to be right (the stock compounds from here as a de-cyclicalizing, regulated crypto utility):

1. The retail take rate must compress only gradually, with new higher-margin asset classes (derivatives, perps, prediction markets, Deribit options) replacing the lost spot-fee economics.
2. S&S must grow through down-tapes, proving genuine de-cyclicalization — not merely soften less than transaction revenue.
3. USDC supply and the Circle annuity must compound (toward $150–200B+) with the reward model surviving GENIUS/CLARITY, and rates must not collapse the yield.
4. Regulatory clarity must be net-additive — the TAM unlock (enterprise onboarding via CDP/custody) must exceed the new competition it invites from well-capitalized TradFi entrants.

Falsification test for the bull: two consecutive down-tape quarters in which S&S falls alongside transaction revenue, and/or a blended take rate that steps down sequentially for 2–3 quarters. Either would prove the diversification is illusory and the franchise is still pure beta.

For the BEAR to be right (the stock is crypto-price beta dressed as a platform, mis-valued at an exchange multiple):

1. Take rates must enter a structural race toward zero as Binance, zero-fee tiers, and TradFi commoditize trading.
2. The crypto cycle must turn down and stay down, compressing transaction and S&S revenue simultaneously.
3. Stablecoin economics must shrink on falling rates and/or adverse stablecoin rulemaking.
4. Normalized mid-cycle earnings must prove to be well below the FY2024 optics, exposing the exchange-like multiple as unjustified.

Falsification test for the bear: a normal-volatility quarter that prints net income at or above the FY2024 run-rate with a stable-to-rising blended take rate — proof that the franchise’s mid-cycle earnings power is structurally higher than the trough optics suggest.

15. Source Appendix

See Appendix B — Source Appendix below for the full, dated source list. Primary sources: Coinbase FY2021–FY2025 10-Ks (latest filed 2026-02-12), the Q1 2026 10-Q (filed 2026-05-07), the 2026 DEF 14A (filed 2026-04-24), the Form 4/144 corpus, and the 8-K material-event timeline; the SEC EDGAR XBRL company-facts API; earnings-call and conference transcripts (Q1 2026 / Q4 2025 / Q3 2025 / the Dec-17-2025 Special Call, plus 2025–26 investor-conference presentations); public market data; and public disclosures of comparable companies (Robinhood, CME, ICE, Cboe, MicroStrategy, IREN).

APPENDIX A — Standard Diligence Questionnaire

Coinbase Global, Inc. (NASDAQ: COIN) — as of 2026-06-10

Fact / Interpretation / Assumption labels are applied where it matters.

General

What thoughtful questions have other investors asked about this company? The dominant investor debate is a single classification question: is Coinbase a regulated-exchange annuity (deserving a CME/ICE multiple) or crypto-price beta dressed as a platform (deserving a cyclical-trough multiple)? Sub-questions that recur: (i) Is the ~1% retail take rate a melting iceberg, and how fast does it melt? (ii) How “recurring” is Subscription & Services really, given it is rate- and crypto-price-sensitive? (iii) What is normalized mid-cycle earnings power, given net income is dominated by crypto marks? (iv) Does regulatory clarity (CLARITY/GENIUS) help Coinbase more than it helps the TradFi entrants it invites? (v) Is the proprietary BTC/ETH treasury a sensible use of capital or a beta amplifier? (vi) How should one think about a founder-controlled company on capital allocation?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? (Interpretation) Closer to a trough/turning point than a high. FY2024 ($2.58B NI) was near-peak; FY2025 ($1.26B) was already down 51%; Q1 2026 was a $394M loss. The crypto tape rolled over in early 2026 and revenue fell 31% YoY. Earnings are depressed and falling — which is precisely why a trailing P/E is misleading.

Driven by the external environment or internal actions? Overwhelmingly external — crypto prices, trading volumes, and interest rates. Internal actions (S&S mix-shift, cost discipline, the Q2 2026 RIF) modulate the cycle but do not drive it. The proprietary crypto treasury is a self-imposed amplifier of the external driver.

How stable are revenues? Among the least stable in large-cap financials. Net revenue: $7.84B (2021) → $3.19B (2022) → $3.11B (2023) → $6.56B (2024) → $7.18B (2025) → annualizing well below that in 2026. Beta 3.38. Two ~60% peak-to-trough collapses in five years.

Outlook for products/services? The S&S/stablecoin/derivatives/institutional layer is genuinely growing (S&S 37%→44% of revenue across five quarters; 12 products >$100M ARR). The retail spot-trading core is structurally fee-compressing and cyclical. Net: improving mix, cyclical level.

How big will this market be — growing, shrinking, domestic or international? (Interpretation) The addressable market (crypto trading, stablecoins, tokenization, onchain payments) is large and secularly growing, but cyclical and globally contested. Coinbase is the US-regulated leader with a growing international footprint (Coinbase International, Deribit), but Binance dominates global volume.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More — zero-fee TradFi entrants, BlackRock/Fidelity ETFs disintermediating buy-and-hold, Robinhood/PayPal expanding, and regulatory clarity lowering the barrier for banks. Offset partially by compliance cost as a barrier protecting scaled incumbents.

How profitable is the business (ROIC, ROE)? ROE ~6.7% TTM — poor for a “platform,” and almost entirely a function of where Bitcoin traded. ROIC is not meaningfully computable (goodwill/intangible-inflated denominator, crypto-mark-distorted numerator). In peak years (FY2021, FY2024) returns look excellent; the through-cycle average is mediocre.

How profitable is the industry — competitors, barriers to entry? The regulated/custody/stablecoin niche is profitable and defensible (compliance barriers, trust). The spot-trading core earns Binance-style commoditizing economics globally. Barriers: licensing, trust/brand, and (for derivatives) Deribit’s open-interest liquidity.

Can the business be easily understood? Moderately. The two-pillar revenue model is clear, but the crypto-mark distortion of net income, the custodial-fund balance-sheet gross-up, and the Circle revenue-share mechanics require care.

Can it be undermined by foreign low-cost labor? Not labor — but by foreign low-fee exchanges (Binance) on the trading core. The regulated US niche is the defense.

Do brands matter? Yes — decisively. Trust/brand is the core moat: customers consolidate to Coinbase in stress (counter-cyclical inflows), supporting a ~10× take-rate premium over Binance.

Nature of competition / customers’ switching costs? Retail switching costs are low (open a competitor account in minutes; friction is only moving custodied assets). Institutional/custody switching costs are higher. The premium rests on trust and convenience, not lock-in.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Customer crypto held in custody (~$376B Assets on Platform) is largely off-balance-sheet under a safeguarding framework; the Circle stablecoin annuity (a contractual, auto-renewing income stream) is an economic asset not capitalized.

Off-balance-sheet liabilities? Customer custodial obligations are recognized (custodial fund liability $5.35B matches the asset). The safeguarding liability for customer crypto is the key item to watch under evolving accounting. Litigation/regulatory contingencies are disclosed but inherently uncertain.

How conservative is the accounting? (Interpretation) Mark-to-market on the crypto investment book is transparent but volatile — it forces real-time crypto-price swings through net income (a fair-value choice, not aggressive). Goodwill/intangibles ballooned to ~$5.6B on Deribit (impairment risk). The main quality issue is interpretive (net income is beta), not aggressive accrual.

How CapEx-hungry is the business? Light on physical capex; “capital intensity” is instead SBC (~12% of revenue), the USDC-rewards subsidy (a S&M expense scaling with float), and the optional proprietary crypto book.

Capital Allocation & Management

How much FCF, and how is it used? Operating cash flow is large but distorted by custodial flows (read net of them). Uses: a $4.3B Deribit acquisition (mostly stock), $1.9B of buybacks at ~$303/share (pro-cyclical), a growing proprietary BTC/ETH book, and convertible-debt management. No dividend.

Significant acquisitions recently? Yes — Deribit ($4.295B, Aug 2025), the global crypto-options leader; plus minor tuck-ins (Echo, etc.).

Buying back shares? Yes, but poorly timed — $1.9B of a $2.0B authorization spent near the cyclical/valuation peak at ~$303 average, just before a net-loss quarter.

Issuing large amounts of new shares to insiders? SBC ~12% of revenue and large time-based RSU grants to non-founder NEOs; gross shares grew ~5.6% in FY2025 (Deribit stock + SBC, partly offset by buybacks).

Compensation policy / motivations of management? Armstrong takes $1M salary + security, no new equity (already controls the company). Other NEOs paid in large time-based RSUs weakly gated on performance metrics — alignment via raw stock exposure, not per-share value or ROIC. Armstrong is mission-driven (“increase economic freedom”) and a long-term founder, but governance gives minority holders no check.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No — a US C-corporation common stock (NASDAQ: COIN), dual-class (Class A public, Class B super-voting). Standard 1099 treatment.

Dividend policy? None. No common dividend; subsidiary dividend restrictions noted in the 10-K.

How profitable is the business? Cyclically — spectacular in peak years, loss-making in troughs. Through-cycle ROE is mediocre (~6.7% TTM). Normalized operating margin is the honest profitability lens; net margin is crypto-mark noise.

Is net income diverging from cash from operations? Yes, and unusually — in down-mark quarters (Q1 2026) OCF is positive (+$183M) while net income is negative (-$394M), because crypto marks are non-cash. The divergence is driven by marks and SBC, plus custodial-fund noise, not by aggressive accruals.

Risks & Downside

What factors would cause the stock to decline? A sustained crypto winter (volume + price collapse), accelerating take-rate compression, falling rates shrinking the stablecoin annuity, adverse stablecoin/market-structure rulemaking, a custody/security breach, or a Deribit goodwill impairment.

Risk of a catastrophic loss? A catastrophic custody/security breach is the principal non-linear risk to the trust moat (a mid-2025 insider-bribery incident is a warning). A structurally adverse regulatory reversal is the second.

Chance of a total loss? Low. >$11B corporate cash, a dominant franchise, and manageable (mostly zero-coupon, high-strike) debt make insolvency unlikely. The realistic severe case is a multi-year earnings depression and multiple compression, not a zero.

Recent News & Events

Has the business environment changed recently? Yes — materially. Positive: the 2023 SEC suit was dismissed (2025); GENIUS (stablecoins) is law; CLARITY (market structure) is expected to be signed by late summer 2026; the CFTC cleared US crypto futures (May 2026); S&P 500 inclusion (May 2025). Negative: the crypto tape rolled over in early 2026 (revenue -31% YoY, Q1 net loss), the stock is -65% from its high, and a Q2 2026 reduction-in-force/AI-native restructuring (~$500M cost target) signals a softer near-term regime.

Significant acquisitions? Deribit ($4.3B, Aug 2025).

Change in accounting policies? Crypto fair-value/mark-to-market treatment is the dominant earnings-shaping policy; no major recent change flagged beyond ongoing crypto-accounting evolution.

Recent changes — new markets, facilities, management? The “Everything Exchange” expansion (tokenized equities, perpetuals, prediction markets via Kalshi, commodities futures), Deribit-driven derivatives leadership, Base/onchain growth, and the Q2 2026 RIF. Management team stable (Armstrong CEO, Choi President/COO, Haas CFO, Grewal CLO).

APPENDIX B — Source Appendix

Coinbase Global, Inc. (NASDAQ: COIN) — Research as of 2026-06-10

Sources are prioritized primary-first. Fact / Interpretation / Assumption distinctions are made in the memo body and diligence appendix; this appendix lists the underlying sources with dates accessed (2026-06-10 unless noted).

1. SEC Filings — Primary (EDGAR; CIK 0001679788)

2. Earnings Calls & Event Transcripts

Transcript content is treated as a hypothesis (management commentary), validated against filings and external data.

3. Market & Quantitative Data

• Public market data (accessed 2026-06-10): price ~$153.97, market cap ~$40.5–48B, EV ~$38–45B, shares ~222.4M Class A, 52-week range $139.36–$444.65, total debt ~$7.2–8.0B, total cash ~$10.4B, trailing P/E ~57–67×, P/S ~6.5–7.6×, EV/EBITDA ~36–38×, ROE ~6.7%, revenue growth -30.8%; employees 4,951; short interest 25.6M shares (~11.7% of float, short ratio 2.4); beta 3.38; 50-day MA ~$190, 200-day MA ~$250. Aggregated third-party data; reconciled to filings.
• Own-history valuation percentiles (as of 2026-06-09): price $155.50, P/E 59.1× (86.6th percentile vs own history), P/B 3.05× (23.7th percentile), P/S 7.42× (36.3rd percentile), composite 48.9th percentile.
• Recent news (as of 2026-06-10): Baird PT cut to $142, Neutral (2026-06-05); CFTC clearance for US crypto futures (2026-05-29); Armstrong–Dimon CLARITY Act dispute (2026-05-29/30). Analyst ratings/targets recorded as third-party signal only.

4. Comparable Companies (public disclosures)

• Robinhood (HOOD) — direct retail-brokerage + crypto competitor; crypto-cycle reference.
• CME, ICE, Cboe — regulated-exchange multiple/quality comparables.
• MicroStrategy (MSTR), IREN — crypto-beta reference points.

5. Analytical Frameworks

• Greenwald & Kahn, Competition Demystified — moat-type taxonomy (intangibles/regulatory + demand captivity; pressure-testing network-effect claims; ROIC/market-share tests).
• Chancellor (Marathon), Capital Returns — supply-side capital-cycle lens applied to crypto-exchange/onchain capacity and TradFi entry.

6. Note on Sourcing

No position in Coinbase is held, implied, or assumed by the author. All quantitative conclusions are reconciled to the SEC filings listed above; analyst price targets are recorded only as third-party signals, never as the author’s view.