The $300 Million That Wasn't: Auditing Evernorth's XRP Treasury Arithmetic
Hook: The Rounding Error That Ate a Treasury
Numbers lie in the spaces between them.
The merger proxy for Evernorth Holdings — the Delaware corporation scheduled to trade on Nasdaq under the ticker XRPN — prints its subscription figures to the dollar. Two hundred twenty-four million, five hundred fifty thousand. Ten million, five hundred thousand, deferred. A thirty-million-dollar convertible note. Trust proceeds near forty-eight million. Add, subtract the line items nobody quotes in the headline, and you land at roughly three hundred three million dollars.
The press release that went out around the shareholder vote did not use those numbers. It used one number: "approximately $300 million."
Round down. Clean headline. Nobody argues about three million dollars of rounding on a three-hundred-million-dollar story.
The interesting number is not the three million. It is the two hundred fourteen million. Because that money — the pre-funded slice of the private placement — had already been converted into XRP. By November, roughly eighty-four million, three hundred seventy thousand tokens sat on the balance sheet, bought at an average of two dollars and fifty-four cents. Two hundred fourteen million dollars of cash, already spent. Already irreversible.
A single dollar cannot buy two different things. Unless you are writing a press release. Logic remains; sentiment fades.
That is the anomaly I want to pull apart. Not the ticker. Not the shareholder vote. The arithmetic.
Context: What a Digital Asset Treasury Actually Is
Strip the marketing and a Digital Asset Treasury — DAT, if you read the S-4s for a living — is a holding company with a legally distinct problem. It has no operating business. It has no revenue. It has an equity structure designed to absorb capital from public markets and redirect that capital into a single asset, held on the balance sheet, marked to market, and presented to investors as a proxy exposure.
MicroStrategy is the canonical BTC version of this machine. Evernorth intends to be the XRP version, and it intends to do it through a SPAC.
A SPAC is a shell. A blank-check corporation raises money in an IPO, parks it in a trust account, and spends two years hunting for a target. If it finds one, the target merges into the shell, the shell becomes the operating entity, and the ticker changes. If the shell fails to find a target, the trust returns to shareholders and everyone walks away with their principal plus a sliver of interest.
Armada Acquisition Corp. II is the shell. Evernorth is the target. RippleWorks — a fundraising and investment vehicle affiliated with the Ripple ecosystem — is one of the contributors.
The structure matters because SPAC trust capital is not free. Every shareholder who votes yes also holds a redemption right. They can take their pro-rata share of the trust and leave before the merger closes. That right is the single most important variable in the entire transaction, and it does not appear in any headline.
I have spent enough of my professional life inside capital structures of this shape to recognize the pattern. In 2017, while the ICO circuit was still pretending that token sales were a novel form of fundraising, I spent three months reverse-engineering the 0x protocol's exchange contracts on GitHub. Not the whitepaper. The Solidity. Order matching, signature verification, the exact byte layout of the fill function. The whitepaper made promises the code did not keep, and the code made promises the whitepaper had never mentioned. That was the moment I stopped reading narratives as primary sources.
The same instinct applies here. A SPAC merger is a smart contract written in legal prose. It has state variables (the trust account), conditionals (redemption rights), and a settlement path (the closing on October 7). If you want to know what the entity will actually hold, you do the arithmetic. You do not read the press release.
The tape I want to reconstruct: what Evernorth says it will have, what the filings say it will have, and what can be spent.
Core: Decomposing the Capital Stack
The Three-Hundred-Million-Dollar Sentence
The headline figure distributed to the market before the vote was "approximately $300 million in gross proceeds." Decomposed against the proxy's precise language, the stack looks like this:
A private placement subscription of $224.55 million, of which $214.05 million was pre-funded and $10.5 million deferred. A $30 million 4% payment-in-kind convertible note, contingent on the completion of the merger. An estimated $48 million in trust proceeds remaining after anticipated redemptions — a figure that is itself an assumption, not a fact, because the actual redemption rate is not known until the closing.
Sum the gross: $303.05 million before fees, before transaction expenses, before the sponsor's promote is settled. Call it three hundred three.
Now run the subtraction that the headline never performs. Of that three hundred three million, $214.05 million had already been spent on XRP by November. It cannot be spent again. The cash is not sitting in a bank account waiting to be deployed. It is a token position with a cost basis of two dollars and fifty-four cents per unit.
What remains as genuinely new, deployable, gross capital?
The deferred subscription: $10.5 million, contingent on the merger closing.
The convertible note: $30 million, contingent on the merger closing.
The trust proceeds: approximately $48 million, contingent on the redemption rate staying bounded.
Total identifiable new gross sources at settlement: approximately $88.5 million. Before fees. Before the mechanics of closing eat their share.
That is the number that should have led the press release. It is roughly thirty percent of the number that did.
Why $214 Million Stops Being Purchasing Power
The essential mechanic — and this is where retail readers get lost — is that pre-funded subscription capital in a SPAC transaction is not held in escrow pending deployment. It is released to the target or its agents as consideration, typically well before the closing, subject to the subscription agreements. In Evernorth's case, the filings indicate the pre-funded portion was converted into XRP in November. The money left. The tokens arrived.
This is not fraud. It is standard. It is also frequently obscured in public communication because it makes for a worse headline. A treasury company that already bought its bag does not have fresh purchasing power. It has a historical position. Those are different claims, and conflating them inflates the perceived demand-side impact on the underlying asset.
If you want XRP exposure, and you believe the treasury premium thesis, you care about how much new buying pressure the vehicle can aggregate. If that number is unknown, you assume it is large. If the number is $88.5 million gross, the aggregated pressure is small relative to XRP's daily notional volume, which routinely clears hundreds of millions to billions across centralized venues.
Frictionless execution, immutable errors. Once the cash is spent, the position it bought is permanent. The narrative around it is not.
The In-Kind Illusion
RippleWorks contributes roughly 211.3 million XRP as an in-kind subscription. A related-party subscription adds another 50 million units. The combined in-kind contribution sits near 261 million tokens.
This is not the same as cash purchase, and the difference is structural, not cosmetic.
Cash purchase consumes market liquidity. It moves the price. It takes size against the order book. In-kind contribution transfers an existing position from one balance sheet to another without touching the tape. From the market's perspective, an in-kind contribution is a reclassification, not a bid.
Worse, the RippleWorks contribution is reportedly revocable under conditions tied to the completion of the merger. That clause is the whole story. It means RippleWorks retains an option — the freedom to walk if the merger terms cease to suit it — while Evernorth carries the downside of a position it may never actually control.
That is an asymmetric contract. Read it the way you would read a token vesting schedule. The counterparty has an out. The counterparty has the information. The counterparty has the asset. You have the disclosure obligation.
I have audited enough DeFi rails to recognize the shape. In 2020, during the DeFi Summer, I audited twelve Uniswap V2 forks for small DAOs in Chengdu. Forty-five distinct logic flaws across slippage tolerance and reentrancy surfaces. The one that killed projects was never the exotic bug. It was always the clause nobody modeled: the privilege held by the counterparty that the contract implicitly assumed away. In-kind contributions with revocation rights are the legal-prose equivalent of an unbounded approval.
The 473 Million Token Problem
The headline positions often cite a post-closing XRP holding of roughly 473 million units. Break that number down and it dissolves.
Eighty-four million, three hundred seventy thousand were already bought with cash, at two-fifty-four. Two hundred eleven million, three hundred thousand are RippleWorks in-kind, revocable. Fifty million are related-party subscription, contractual. The remainder is a mixture of committed but unclosed allocations and rounding effects.
Summing these categories and presenting the total as a unified holding is a category error. Worse, the filings suggest that in November — before the merger even closed on paper — the language describing "more than 473 million XRP held or committed" was already in circulation. A number introduced before settlement cannot be treated as incremental at settlement.
Metadata is fragile; code is permanent. In 2021, I wrote a Python routine to check metadata integrity across ten thousand NFT tokens on Ethereum. Fifteen percent depended on centralized IPFS gateways that could and did go dark. The tokens were immutable. The metadata was not. The same rot applies to financial disclosure. A holding figure that mixes realized positions, committed positions, and revocable positions is metadata. Only the settled position is code.
Until you can separate "held" from "committed" from "revocable," the 473 million is a marketing artifact, not a balance sheet.
The PIK Note and the Compounding Clock
A four percent payment-in-kind convertible note does not pay interest in cash. It pays interest in principal. The debt accretes. It compounds. It becomes a larger claim over time, and it converts to equity on terms the holder selects, typically when conversion is most advantageous to them.
Evernorth's note matures in 2031. That is a comfortable runway on a short time horizon and a slow bleed over a long one. Between now and 2031, the note's notional grows by roughly four percent annually, and its eventual conversion will dilute equity at a moment the company has the least capacity to absorb it — likely a moment when XRP has appreciated, the equity has rallied, and the noteholder's conversion calculus is maximally favorable.
Against a position bought at two-fifty-four and currently quoted near one-fifty-three, the note is a negative-carry instrument attached to a negative-carry asset. Held long enough, the compounding works against the equity holder, not for them.
The debt does not have to be repaid tomorrow. That is the trap. It has to be repaid eventually, and the terms are written by the other side.
The Redemption Variable
Everything above rests on a variable that has not yet been printed: the SPAC redemption rate.
Trust proceeds are not a fixed number. They are the residue after every shareholder who wants their money back has taken it. If ten percent redeem, the trust contributes roughly forty-eight million. If fifty percent redeem, the trust contributes roughly twenty-seven million, and the deployable gross collapses toward sixty-seven million. If seventy-five percent redeem, Evernorth closes with less new capital than a single mid-sized DeFi protocol raises in a seed round.
Redemption is the shareholders' put option. It is exercised when the market distrusts the merger. It is therefore negatively correlated with the merger's success — the exit door swings widest exactly when you least want people walking through it.
A DAT vehicle that closes with thin trust capital is not a treasury. It is a shell holding historical positions and a revocable contribution. The buying power that gave the press release its headline simply does not exist.
I have seen this shape before, in different clothing. In 2022 I audited three cross-chain bridges and found integer overflow bugs in two of them — the kind that produce silent wrong answers rather than loud failures. The pattern holds across domains. The dangerous number is never the one that breaks. It is the one that quietly computes wrong and keeps reporting a plausible result. A SPAC trust account after redemption is exactly that number. It reports a residue. Whether the residue is meaningful depends entirely on a rate nobody publishes until it is too late to change it.
The Disclosure Asymmetry
Here is the part that reads like a controlled experiment in financial narrative design.
The proxy, filed with the SEC, breaks out subscription figures to the dollar. It enumerates conditions, deferrals, contingent tranches, and the terms under which contributions can be revoked. The proxy is verbose. It is tedious. It is correct, in the sense that any careful reader who does the arithmetic reaches the same conclusions I have written here.
The press release, distributed to the market, does none of that. It offers a single rounded figure — "approximately $300 million" — and lets the reader import their own assumptions about deployability.
The distance between the two documents is not a mistake. It is a design. Disclosure regimes tolerate it because the information is technically present, if buried. Market participants tolerate it because rounding is unremarkable. Between those two tolerances sits a spread between perception and reality, and that spread is the return the sponsor collects.
Vulnerabilities hide in plain sight. Nothing about Evernorth's arithmetic is concealed. It is merely unaggregated. The work required to aggregate it is small — a spreadsheet, an afternoon. The number of participants who will do that work is smaller.
Core, Continued: The Valuation Frame
Step back and mark the position correctly.
Evernorth has no operating cash flow. It has no protocol revenue. It has no product. Its net asset value is a linear function of the XRP price, discounted by the market's assessment of its financing structure and the probability that its in-kind contributions survive closing.
If XRP trades at one-fifty-three and the company's average cumulative cost is nearer two-fifty-four across the cash-purchased position, the company is carrying unrealized losses on its realized book. Add the debt overhang — the PIK note, growing toward 2031 — and add the redemption risk on trust capital, and the pro-forma balance sheet is less attractive than the pitch deck.
None of this is catastrophic. A DAT can operate with negative mark-to-market in its first year and recover. MicroStrategy survived far worse drawdowns in BTC. The question is whether the structure can finance those drawdowns. MicroStrategy financed its way through BTC volatility by issuing equity at premiums to net asset value and by issuing convertibles against a liquid, deep, institutionally accepted asset. Its financing capacity was a function of BTC market depth.
XRP's institutional market depth is a fraction of BTC's. The depth of potential convertible buyers is narrower. The pool of equity investors willing to underwrite a treasury premium thesis on XRP specifically is smaller. The vehicle's refinancing capacity is therefore lower than the MicroStrategy template implies. This is the load-bearing assumption that no press release states, and it is the assumption most likely to fail.
I want to be precise about what I am and am not claiming. I am not claiming Evernorth is a Ponzi. A Ponzi requires paying early investors from later investors' capital with no legitimate economic activity. Evernorth does have an economic activity — it holds an asset. I am claiming that Evernorth is a leveraged, single-asset, financing-dependent vehicle whose purchasing power is correlated with the price of the asset it intends to purchase. That correlation is the structural vulnerability, and it is not a bug. It is the design.
Impermanent loss is a feature, not a bug. Read that again in the DAT context. The vehicle loses capacity precisely when the market needs it most. It gains capacity precisely when the asset is most expensive. This is not a flaw in execution. It is a mathematical property of the structure.
Contrarian: The Redemption Rate Is the Whole Trade
Here is what most coverage of the Evernorth transaction misses.
The press release debate — three hundred million versus three hundred three million, or three hundred million versus eighty-eight and a half — is a sideshow. The real variable is not in the announcement at all. It is in the trust account, and it will not be known until the closing, at which point it will be too late for anyone who did not price it in advance.
Conventional analysis treats SPAC redemption as a technicality. The sophisticated frame treats it as the primary pricing input. Preferred redemption rates on large SPAC deals have hovered between eighty and ninety-five percent in many cases over the past few years. If Evernorth's redemption rate lands anywhere in that range, the trust contribution goes to near zero, and the actual deployable capital reduces to the deferred subscription plus the convertible — roughly forty million dollars, gross, contingent.
A forty-million-dollar treasury is not a treasury. It is a rounding error on a Nasdaq listing.
There is a second-order effect that no one talks about. RippleWorks's in-kind contribution is revocable. If the redemption rate is high enough to suggest the merger will underdeliver against its stated vision, RippleWorks faces a simple calculus: contribute 211 million XRP into a thin-cap vehicle, or walk and keep the position. A rational RippleWorks walks. And the moment one major contribution becomes revocable, the others become questionable, because the conditions are rarely independent.
This is the contagion risk embedded in conditional capital structures. Each condition looks benign in isolation. Bundled, they produce a knife-edge: the deal either closes with all contributions intact and capital sufficient, or it closes with contributions withdrawn and capital insufficient, and there is not much middle ground.
Trust no one; verify everything. In 2026 I audited one of the first AI-driven trading bots integrated with a decentralized oracle network. Twelve instances where the model's heuristic outputs bypassed the smart contract's input validation layer. The failure mode was never malicious. It was the model optimizing for a goal the contract had not defined. Conditional SPAC structures fail the same way. The sponsor optimizes for closing. The contributor optimizes for optionality. The shareholder optimizes for redemption. Nobody is optimizing for the treasury's actual purchasing power, because no single party's payoff is tied to it.
That is the blind spot. It hides behind precise arithmetic in the proxy and vague language in the press release, and it will resolve in a single disclosure on a single date.
A Detour Into Regulatory Contours
I should say something about the regulatory frame, because it is load-bearing and usually ignored.
The XRPN shares themselves are unambiguously securities. Nasdaq-listed equity, Delaware corporation, SEC reporting obligations. The Howey analysis on the stock is trivial.
The interesting question is the underlying asset. XRP spent three years in litigation against the SEC. The final judgment left the token in a partially ambiguous regulatory posture, which is precisely the posture that suppresses institutional participation. A treasury company that holds XRP on its balance sheet inherits that ambiguity as a disclosure risk.
Now layer the European regime on top. MiCA nominally gives Europe clarity on stablecoins and crypto-asset service providers. What it actually gives is a compliance burden calibrated to institutions that can absorb it. Reserve requirements for stablecoin issuers and CASP licensing costs will compress small projects out of the market, and the survivors will be the balance sheets large enough to subsidize legal departments. A US-listed XRP treasury that touches European institutional capital will face both regimes simultaneously. The compliance cost is real, and it is a drag on treasury NAV.
The disclosure asymmetry I described earlier also has a regulatory edge. If the SEC concludes that the gap between "approximately $300 million" and the actual deployable figure was marketed as purchasing power — as distinct from gross sources — there is a theoretical misstatement exposure. I rate it low-probability and low-severity, because the precise figures are in the proxy and the doctrine tolerates rounding. But the doctrine also treats the impression created by disclosure as material, and the impression created here was of a larger machine than the arithmetic supports.
This is not the cryptocurrency industry's first encounter with the gap between narrative and disclosure. It will not be the last. The pattern repeats because the disclosure regime rewards precision and the marketing regime rewards rounding, and the two run on the same wire.
Second-Order Contours and the Relative Value of Concentration
The more interesting question, at the margins, is whether Evernorth's structure teaches us something about the DAT model in aggregate. It does.
Every DAT — BTC, ETH, XRP, or otherwise — carries a reflexive relationship between its equity price and its financing capacity. When the underlying asset appreciates, the NAV premium widens, the equity becomes a cheaper source of financing, and the vehicle can issue shares to buy more of the asset. When the underlying asset depreciates, the premium compresses to a discount, the equity becomes an expensive financing source, and the vehicle is forced to stop buying. In extreme cases, the vehicle is forced to sell to service debt.
This is a pro-cyclical financing structure. It amplifies both directions. The MicroStrategy template worked because BTC consistently validated the up-cycle for long enough to build a position large enough to survive the down-cycle. That is a survivorship outcome, not a guarantee. Vehicles that launched BTC treasuries at the top of the 2021 cycle did not all survive the 2022 drawdown.
Evernorth launches an XRP treasury into a post-halving period of suppressed mining economics and compressed altcoin liquidity. The Bitcoin halving of April 2024 cut miner revenue, and the long-run consequence is further consolidation of hash power into a shrinking set of pools. The concentration of hash power into three or four dominant pools is now close to a structural fact, and it hollows out the decentralization narrative that retail still trades on. The same consolidation logic applies to DAT financing capacity. The vehicles that can refinance will be the ones with the deepest underlying asset liquidity and the widest institutional buyer base. XRP is neither of those.
That is the ecosystem position that the Evernorth pitch does not address. The company is not just betting on XRP's price. It is betting on XRP's institutional liquidity deepening over the next five years — enough to sustain convertible issuance, enough to sustain equity issuance at premiums, enough to allow the PIK note to convert on terms that do not destroy the equity holders. None of those bets is guaranteed. All of them are correlated.
I want to be fair. There are scenarios where this works. If XRP appreciates substantially, if the redemption rate is contained, if RippleWorks's contribution holds, if the PIK note converts at a point of peak enthusiasm, the vehicle generates returns for early equity participants. That is a real scenario. It is also a scenario that requires several independent conditions to resolve favorably, and the number of independent conditions is a proxy for fragility.
Takeaway: The Date That Matters Is Not in the Deck
If you are following this transaction, ignore the headline. Ignore the ticker. Ignore the vote, which already happened.
Watch one disclosure: the post-closing redemption rate. It will determine the trust contribution. The trust contribution, in turn, determines whether Evernorth's deployable gross capital lands near eighty-eight million or collapses toward forty million. That number, printed once, in a single filing, will settle the question the press release never answered.
Watch one more: the RippleWorks condition status. If the revocable contribution loses its conditions and settles, the position is real. If the conditions fail, the position is a footnote.
And watch the tape on XRP itself, because the cost basis matters. A treasury that bought at two-fifty-four and trades the underlying at one-fifty-three is not demonstrating strategy. It is demonstrating the timing risk embedded in every financing-dependent DAT. The structure does not care about your conviction. It cares about whether you can refinance at the bottom.
Silence is the loudest exploit. The Evernorth press release did not lie. It simply declined to do the subtraction that would have changed the story, and left the arithmetic to the people who read the proxy.
That category of reader is small. It does not stay small forever. Which is why the number written on the disclosed page will, at some point, become the number the tape prices. Not the number on the press release.
The date to watch is not the vote. It is not the merger closing. It is the first 8-K after redemption, when a rounded headline collides with a precise filing and the market is forced to choose which one it believed.
Metadata is fragile; code is permanent. The filing is the code. The press release is the metadata. Verify accordingly.
Appendix: The Arithmetic, Laid Out Plainly
Because a claim without an audit trail is a narrative, and I do not traffic in narratives.
Gross sources, as stated:
Pre-funded subscription: $214,050,000, already deployed into XRP in November.
Deferred subscription: $10,500,000, contingent on closing.
PIK convertible note: $30,000,000, contingent on closing.
Trust proceeds, pre-redemption estimate: approximately $48,000,000, contingent on redemption rate.
Aggregate gross: $303,050,000, before fees and transaction expenses.
Deployable new gross sources after removing already-spent capital:
$10,500,000 plus $30,000,000 plus approximately $48,000,000 equals approximately $88,500,000, contingent on closing and a bounded redemption rate.
Effective deployable capital at a seventy-five percent redemption rate: approximately $67,000,000.
XRP purchase equivalence at $1.53 per unit, on $88.5 million gross, before fees: approximately 57.8 million tokens.
These are the numbers. They are not secret. They are simply not aggregated, because aggregation serves a purpose the press release did not have.
Read the proxy. Do the subtraction. Trust the arithmetic, not the adjective.