Hook
The most consequential number in this story is not six million. It is zero.
Zero is how many of the source fields in the original dispatch were populated. No execution price. No settlement date. No counterparty. No custodian. No wallet address. No exact token count. Every field that would let a reader verify a single sentence came back marked 'none.' And sitting on top of that empty scaffolding is a claim large enough to bend the narrative of a $400 billion asset class: that one Corporate entity, Bitmine, now holds something close to six million ETH.
That is roughly five percent of Ethereum's total supply. For scale, the Ethereum Foundation's treasury is a rounding error beside it. The largest liquid staking protocol's controlled stake is the same order of magnitude. No publicly traded company on any exchange has ever held a comparable share of a major proof-of-stake network, and no oil major, no sovereign fund, no central bank reserve portfolio has ever concentrated this much of a single digital asset on one balance sheet.
Then the fragment that actually circulated: $68 million of additional ETH purchases.
No unit price. No quantity. No date of execution.
Twelve years of doing this work has taught me that the shape of a disclosure tells you more than its contents. A serious treasury update ships with a wallet address, an average cost basis, a settlement window, and a named custodian. This one shipped with none of those. That absence is not sloppiness. It is architecture. And if you are the kind of reader who only looks at the headline number, you are the intended audience.
Context
Pull back. The Digital Asset Treasury model — DAT, for those who need the acronym — is not complicated. It is a capital-markets flywheel with three moving parts, and it has been running in public since August 2020.
The model works like this. A company issues equity — common stock, converts, preferred, sometimes a mix — at a price above the net asset value per share of the crypto it already holds. Proceeds buy more crypto. Per-share crypto holdings rise without existing holders paying a cent for it. That is accretion, and it is the entire thesis. Then you repeat. MicroStrategy industrialized this with bitcoin starting in 2020 and turned a mid-cap analytics vendor into a leveraged bitcoin proxy with a market capitalization several times its software revenue. By 2024 the playbook was public domain. By 2025 a cohort of imitators had pointed the same machine at ETH: Bitmine, SharpLink, ETHZilla, Bit Digital, and a rotating cast of smaller shells and SPAC remnants, each one promising shareholders a slice of the same trade.
The ETH-specific twist is staking. A bitcoin treasury generates nothing. It sits in a cold wallet and depreciates against the dollar until it does not. An ETH treasury can pledge its stack to consensus and earn protocol-level yield — currently in the low single digits annualized, typically quoted somewhere between 2.5 and 4 percent depending on network activity and the split between consensus-layer and execution-layer rewards. That is real revenue. It is not a token emission subsidy. It is a payment from the protocol for performing a job, and it is the single most important structural difference between the BTC version of this trade and the ETH version.
Which is exactly why the ETH version is both more interesting and more fragile than the one it was copied from.
Set the backdrop, because the numbers matter and most coverage skips them. Ethereum's total supply is somewhere around 120 million ETH. Total staked is in the 34 to 36 million ETH band, roughly 28 to 30 percent of supply. Since EIP-1559 activated in August 2021, ETH has run a low-inflation or mildly deflationary supply curve, with issuance partially or fully offset by base-fee burn depending on block-space demand. Six million ETH is five percent of supply. If the dispatch's claim that 'most of the tokens are earning staking yield' holds, that is somewhere between 15 and 18 percent of the entire staked set controlled by one legal entity.
Sit with that for a moment. Fifteen percent of Ethereum's validator weight is not a portfolio position. It is an infrastructure claim.
And the market context here is not a bull market. We are in a sideways tape. Consolidation, chop, range-bound price action, declining realized volatility, funding rates that hover near neutral, and a retail cohort that has largely stopped paying attention. This is precisely the environment where marginal announcements get inflated into directional signals, because there is nothing else to trade on. In a trending market, a $68 million purchase is noise. In a sideways market, it is a headline.
That is the setup. Now the dissection.
The Arithmetic of a Buy That Changes Nothing
Start with the only piece of hard, checkable arithmetic in the entire dispatch.
If ETH traded between $3,000 and $4,000 across the purchase window — a reasonable range for the current consolidation band — then $68 million converts to roughly 17,000 to 22,700 ETH. Against a stated base of approximately 5.9 million ETH, that is between 0.29 and 0.38 percent of existing holdings.
| Metric | Value | Interpretation | |---|---|---| | Notional purchase | $68,000,000 | Headline number | | Implied ETH acquired | 17,000 – 22,700 | At $3,000–$4,000/ETH | | Existing holdings | ~5,900,000 ETH | 'Close to 6 million' | | Incremental share | 0.29% – 0.38% | Rounding error | | Stated target | 5% of supply (~6,000,000 ETH) | Already ~98% achieved | | Distance to target | ~100,000 ETH | Under 2% of current stack |
Read the bottom two rows again. If the five-percent target is six million ETH, and the entity already holds 'close to six million,' then this purchase is not a strategic expansion. It is a completion trade. It is a purchase made to close a gap that is already almost closed.
So what is the information content? It is not financial. A $68 million bid inside an asset with tens of billions in daily spot volume, executed across multiple venues, is absorbed without leaving a footprint. The direct price impact is essentially unmeasurable. If you modeled it, you would find it buried inside the confidence interval of a single hour's trading.
The information content is behavioral. It tells you three things: the entity still has access to capital or cash, the entity still wants to be seen buying, and the entity still controls its own disclosure cadence.
That third point is the one nobody prices. A high-frequency cadence of small announcements is not a treasury strategy. It is an attention strategy. When a company discloses purchases weekly or biweekly in small increments and never in a single consolidated filing, it is manufacturing a news cycle. Each release gives the sell-side a reason to publish, each publication gives the retail channel a reason to post, and each post gives the equity a reason to hold a bid. The cumulative PR value of twelve $68 million announcements is vastly greater than the PR value of one $816 million announcement, even though the balance sheet outcome is identical.
I have seen this pattern before, in a different costume. In 2020, during DeFi Summer, I was part of a University of Pennsylvania student group manually tracking simulated yield across three Yearn Finance vaults. The 'gurus' in the Discord kept posting daily screenshots of compounding returns. The numbers were real. The presentation was the product. When I cross-checked slippage assumptions and posted the discrepancies, I was called a noob for three weeks. Then one protocol got reaped, and the numbers stopped being screenshots. That was the moment my writing shifted from observational to forensic permanently. Announcement cadence is a metric. Treat it like one.
The Staking Black Box
Now to the sentence that should have generated ten times more scrutiny than it did: 'most of the tokens are earning staking yield.'
That sentence hides an entire operational architecture, and the architecture determines the risk. There are at least four distinct ways to stake 5.9 million ETH, and they do not have remotely similar failure modes.
| Implementation | Mechanism | Primary Risk | Counterparty Exposure | |---|---|---|---| | Self-operated validators | Proprietary nodes, own keys, own clients | Slashing, downtime, key compromise, operational failure | None external | | Staking-as-a-service | Delegated to Kiln, Coinbase Cloud, Figment, etc. | Provider insolvency, provider slashing, fee drag | High | | Liquid staking derivatives | Hold Lido stETH, Rocket Pool rETH, etc. | Smart contract risk, peg risk, validator set risk | High, stacked | | Restaking / LRT | EigenLayer-style layered security | Cascading slashing, correlated AVS failure | Very high, compounded |
The dispatch tells us nothing about which one is in use. That gap is not academic. If the entity is running self-operated validators, it has taken on slashing risk and key-management risk, but it has retained control and captured the full yield. That is the highest-margin, highest-operational-burden path. It requires a real infrastructure team, client diversity discipline, monitoring, and a key ceremony that would make a bank auditor sweat.
If the entity has delegated to a third party, it has outsourced the operations and imported a counterparty. That counterparty's solvency, insurance, and slashing policy become line items on the parent's risk register, whether or not they appear in a filing.
If the entity is holding liquid staking tokens, the exposure stacks. You now have ETH risk, plus staking protocol risk, plus peg risk on the derivative itself, plus whatever governance risk sits in the protocol's token. That is not one position. It is four positions wearing one ticker.
There is a fourth scenario that almost nobody discusses in public and that I would want to rule out before writing a single positive word about a structure like this: restaking. If a meaningful portion of the stack is securing additional networks via restaking, the slashing surface expands dramatically and correlations between failure modes go up, not down. A single correlated event across multiple actively validated services could, in theory, penalize the same capital twice. The dispatch does not mention restaking. It also does not rule it out. In a disclosure environment where every source field reads 'none,' the absence of a denial is not a denial.
Here is where I want to plant a flag that will annoy people. Yield is a sedative; volatility is the needle. The 2.5 to 4 percent staking return is the part of this story that gets quoted because it sounds like revenue. It is not the part that determines whether the entity survives. The entity survives or dies on the price of ETH, and everything else is a rounding adjustment. A 3 percent yield on a six-million-ETH stack is real money — roughly $540 million annually at a $3,000 ETH price. But a 30 percent drawdown in ETH destroys eighteen months of that yield in a single afternoon. The sedative numbs you. The needle is what actually enters the vein.
And there is a concentration question inside the yield that nobody has asked out loud. Fifteen to eighteen percent of the staked set under one operator is not a yield story. It is a consensus story, and consensus is where Ethereum's credibility actually lives.
Concentration Is a Consensus Problem, Not a Marketing Problem
Ethereum's security model does not just depend on how much is staked. It depends on who is staking and how independently they operate.
The community has litigated this before. Lido's share of staked ETH became a recurring governance flashpoint precisely because the network's credible neutrality is a function of validator diversity, not validator quantity. Client diversity has been an explicit engineering priority since the 2022 merge, for the specific reason that a supermajority client bug could trigger a finality crisis. Geographic distribution of validators matters because jurisdiction-level censorship is a real, demonstrated hazard. MEV-Boost relay dependency matters because a small number of relays have at various points processed the majority of blocks.
Into that delicate equilibrium, insert a single corporate entity with 15 to 18 percent of the staked set.
| Dimension | Why It Matters | Effect of Single-Entity Concentration | |---|---|---| | Client diversity | A dominant-client bug can halt finality | Unknown client mix; undisclosed | | Geographic distribution | Jurisdictional censorship risk | Unknown; likely concentrated | | Relay dependency | Block censorship and MEV capture | Unknown; likely default relays | | Governance signaling | Social-layer legitimacy | A single corporation gains outsized voice | | Exit liquidity | Coordinated exits can destabilize | Enormous, undiversified |
I want to be precise here, because this is the part where careless analysts overstate the case. Holding a large stake does not give an entity the power to rewrite history or steal funds. Ethereum's slashing and fork-choice rules make that prohibitively expensive and socially suicidal. What it does give an entity is influence over the social layer, the governance layer, and the exit layer — and those three are where the actual fragility lives.
The exit layer is the one I would watch most closely. If a single operator controls 15 to 18 percent of staked ETH and decides, for whatever reason, to unbond, the withdrawal queue becomes a systemic event. Ethereum's exit queue is deliberately rate-limited to protect consensus. It is not designed for a single counterparty walking out the door with a double-digit share of the validator set. That is not a hypothetical attack. It is a scenario you can model in a spreadsheet, and I have not seen a single published model of it from anyone covering this cohort.
And here is the uncomfortable corollary: this concentration is entirely reversible. A corporate treasury has no protocol-level lock-in. It can unwind in a quarter. That means the ecosystem's exposure to this entity is asymmetric in the wrong direction — the network gets a small permanent security benefit while the entity retains total optionality.
It is a one-way dependency. Ethereum does not need Bitmine. Bitmine is a leveraged bet on Ethereum and nothing else.
The mNAV Flywheel, and the Reflexivity Nobody Prices
The real risk in this structure is not on the blockchain. It is on the stock exchange.
Every DAT runs on a single number: mNAV, the market-to-net-asset-value multiple. It is the ratio of the company's market capitalization to the fair value of the crypto it holds.
When mNAV is above 1.0, an equity issuance is accretive. You sell a share for more than the crypto backing that share, buy crypto with the proceeds, and every existing holder's per-share crypto goes up. That is the engine. That is the entire flywheel.
When mNAV drops below 1.0, the same action inverts. Issuing equity now dilutes per-share crypto holdings. The company is selling dollars for eighty cents. The rational move flips from issuing to buying back — and buying back requires free cash the model does not generate.
Run the arithmetic on the current position, using round numbers and clearly labeled assumptions.
| Step | Assumption | Value | |---|---|---| | ETH held | Disclosed | 5,900,000 ETH | | ETH price | Current range midpoint | $3,500 | | Gross asset value | 5.9M x $3,500 | ~$20.65B | | Shares outstanding | Illustrative | 100,000,000 | | NAV per share | $20.65B / 100M | $206.50 | | Market price | Illustrative premium | $310.00 | | mNAV | 310 / 206.50 | 1.50x | | New issuance | 5,000,000 shares | $1.55B raised | | ETH purchased | $1.55B / $3,500 | 442,857 ETH | | New ETH per share | 6,342,857 / 105M | 60.41 ETH per 1,000 shares | | Old ETH per share | 5.9M / 100M | 59.00 ETH per 1,000 shares | | Accretion | 60.41 vs 59.00 | +2.4% |
Now flip the sign. Set mNAV at 0.85x. The same issuance destroys 2.4 percent of per-share holdings. There is no operational lever to pull. The company cannot stake its way out of a premium collapse.
This is what George Soros called reflexivity, and it is not a metaphor here. The price of the equity influences the fundamental accretion, the accretion influences the narrative, and the narrative influences the price of the equity. It is a closed loop with a positive feedback term — until it is not.
I have watched this exact geometry before, and I do not mean that rhetorically. In 2022, two days before the Terra collapse went terminal, the thing that struck me was not the technical failure of the peg mechanism. It was the same self-referential structure: an asset whose value derived from a promise to buy more of the asset. I did not write about it cleanly at the time. I was too deep in the market's collapse, so instead I hosted a weekly Crypto Triage mixer in Manhattan and let developers and traders argue in a room with bad lighting. What came out of those conversations is the analysis I trust most today, because it was stress-tested by people who had lost money. The pattern is identical. The only thing that changed is the collateral quality and the fact that the new version wears a suit and files with the SEC.
A DAT is a reflexive structure. Reflexive structures do not fail slowly. They fail when the feedback term flips sign, and the sign flips when the marginal buyer stops showing up.
In a sideways market, the marginal buyer always stops showing up eventually.
Custody: The Layer Nobody Audits Until It Fails
Six million ETH is roughly $20 billion at current prices. Ask yourself a simple question: where is it?
There are only a handful of credible answers. A qualified custodian such as Coinbase Prime, BitGo, Anchorage, or Fidelity Digital. A multisig arrangement with disclosed signers and a disclosed threshold. A cold storage architecture with an institutional key ceremony. Or some combination of the three, with a portion hot for staking operations and a portion cold for reserves.
The dispatch mentions none of this. That matters because custody is the single largest operational attack surface in the entire structure, and it is the one that most equity analysts have no framework for evaluating.
I learned this the blunt and personal way. In 2021, working as a junior analyst, I attended NFT NYC and ended up sitting with a group of Axie Infinity players who had lost their savings to a phishing site that mimicked the official launcher. I pulled the contract interaction logs for eleven of them that weekend. It was not a protocol bug. It was signature spoofing — users signed a transaction that granted an unlimited allowance to an attacker-controlled address. The protocol worked exactly as designed. The users were the vulnerability.
That week I adopted a rule I have never broken: no promotional language for any project whose key-management architecture I cannot verify. The rule has cost me access and it has cost me invitations and I would do it again tomorrow.
For a treasury of this size, the questions are specific and answerable:
- What is the multisig threshold, and who holds the keys?
- Is there a separation of duties between transaction initiators and approvers?
- Are signing keys generated in hardware security modules with attestable provenance?
- Is there slashing insurance, and what is the deductible?
- What is the recovery plan if a custodian becomes insolvent or is compelled by a regulator to freeze assets?
- Are staking withdrawal credentials pointed at a contract or at an entity-controlled address?
The last question is the one that separates amateurs from professionals. Withdrawal credential type determines who can actually move funds out of a validator. If withdrawal credentials are pointed at a smart contract, that contract becomes a permanent, load-bearing dependency for every validator the entity runs. If they are pointed at a plain address, the entire security model reduces to the key management of that address.
None of this is disclosed. We audit the code, but we mourn the users. In this case there is no code to audit — just an announcement, and a balance sheet, and an undisclosed number of people who have no idea what they actually own.
Disclosure Quality Is Itself a Risk Asset
The absence of sourcing is not just an annoyance for analysts. It is a priced-in risk that nobody has bothered to price.
If Bitmine is a US-listed entity — a reasonable inference given that the DAT model has concentrated in US public markets and that the drafting style of the dispatch echoes standard corporate release formatting — then it operates under the SEC's disclosure regime. That means periodic reports, material-event disclosures, audited financials, and fair-value accounting for digital assets under the updated crypto measurement framework that took effect for fiscal years beginning after December 2023.
| Disclosure Element | Why It Matters | Present in Dispatch | |---|---|---| | Execution price or VWAP | Verifiable cost basis | No | | Settlement date | Materiality window | No | | Precise holdings | Balance-sheet accuracy | No — 'close to 6 million' | | Custodian identity | Counterparty risk | No | | Staking method | Slashing and counterparty risk | No | | Validator count | Concentration analysis | No | | Wallet addresses | Direct on-chain verification | No | | Financing structure | Dilution and leverage | No |
Eight for eight missing. That is not a rounding error in disclosure practice. That is a systematic pattern.
I want to be fair here, because the Cold Dissector's job is not to assume fraud. There are legitimate reasons a company keeps some of these details close. Publication of wallet addresses invites targeted attacks on key-management infrastructure. Precise staking architecture can be operationally sensitive. Counterparty terms may be under confidentiality agreements. These are real considerations and they are widely accepted practice in the institutional crypto space.
But there is a difference between withholding operational detail and withholding arithmetic. A company that will not tell you its average cost basis on a $68 million purchase is not protecting its infrastructure. It is protecting its narrative.
The accounting dimension deserves more attention than it gets. Fair-value measurement means every quarterly earnings release will now swing violently with ETH's price. Shareholders who think they bought a treasury-management business have actually bought a mark-to-market derivative with a corporate wrapper and a legal personality. And unlike a futures position, this one has no defined expiry, no margin call, and no hedge — which is either the best or worst feature of the structure depending on which direction ETH goes. Structurally, it is the worst, because the entity has no mechanism to protect itself on the downside. There is no counterparty on the other side of a six-million-ETH short.
The Arms Race Nobody Wins
The DAT cohort is competing for a finite pool of narrative oxygen, and the competition is entering its crowded phase.
| Entity Type | Asset | Differentiation | Strategic Position | |---|---|---|---| | Bitcoin pioneer | BTC | First mover, brand, index inclusion | Deep moat | | ETH treasury A | ETH | Scale, staking integration | Front runner, thin moat | | ETH treasury B | ETH | Financing structure | Follower, thinner moat | | ETH treasury C | ETH | Yield engineering | Follower, thinnest moat | | Smaller shells | ETH | Leverage and speed | Existential dependence on premium |
Look at the rightmost column. Only one of these positions has an actual moat, and it belongs to the bitcoin first mover. Everyone else is running the same playbook with the same asset, the same flywheel, and the same dependency on a premium that is itself a function of how many of them exist.
This is the part of the thesis that bulls systematically underweight. When there was one ETH treasury vehicle, its stock traded at a premium because it was scarce. When there were two, the premium compressed. When there are eight, the premium reflects nothing but marginal flow. Every ETH treasury trades in MicroStrategy's shadow, and none of them can escape it, because the entire category was defined by a single company's precedent.
The strategic response options are limited and none of them are great. You can differentiate on staking yield, but the yield spread between the best and worst operator is measured in tens of basis points — not enough to sustain a multiple. You can differentiate on financing cost, but that is a race to the bottom that ends with the weakest balance sheet failing first. You can differentiate on brand, but brand in crypto has a half-life measured in market cycles. Or you can differentiate on disclosure and transparency — which, given the eight empty fields above, nobody in this cohort is currently doing.
In a sideways market, the arms race gets worse, not better. Flat prices mean flat NAV growth. Flat NAV growth means the only way to grow per-share holdings is to issue. Issuing requires a premium. Premiums compress when there is nothing new to say. And the way to have something new to say is to keep announcing purchases — which brings us all the way back to the $68 million headline that started this dissection.
The announcement is not a distraction from the strategy. The announcement is the strategy.
What Five Percent Actually Does to the Float
The bull case on concentrated ETH treasuries rests on a supply argument: buy and stake, remove tokens from circulation, tighten float, support price. It sounds intuitive. It is mostly true and almost entirely irrelevant at this size.
Start with the supply stack.
| Category | Approximate Size | Notes | |---|---|---| | Total ETH supply | ~120 million | Post EIP-1559, low inflation | | Staked | 34–36 million | ~28–30% of supply | | Bitmine holdings | ~5.9 million | ~5% of supply | | Bitmine staked (claimed) | Majority | ~15–18% of staked set | | Exchange balances | Several million | Declining trend | | ETF holdings | Growing | Passive, non-staking in most cases | | DeFi locked | Low millions | Protocol-dependent | | Liquid float | Residual | Actual tradable supply |
Here is the honest math. Five percent of supply is not a small number, and if the entire position is staked and held indefinitely, it does reduce the tradable float by that amount. Over a long horizon, the marginal supply effect is real.
But it is not the effect that moves price. Price in a liquid market is set by the flow at the margin, not by the stock. ETH trades tens of billions of dollars per day. A locked five percent changes the slope of the supply curve. It does not change the intercept.
Where it does matter is psychology, and this is where I think the five-percent figure was chosen deliberately rather than derived from any portfolio model. Five percent is a milestone. It is round. It is quotable. It is a natural headline: Company X now holds five percent of all Ethereum. That sentence does work that no financial model can replicate. The five-percent target is not a target. It is a narrative anchor, engineered for repetition.
And there is a specific risk in anchors like this. When a target is close to being hit and is publicly stated, hitting it becomes a catalyst. In a sideways market, catalysts are scarce and therefore over-traded. Which means the day the five-percent milestone is formally announced is quite possibly the day the marginal buyer is exhausted — the classic sell-the-news configuration, dressed up as a triumph.
If you want a genuinely useful supply metric, do not track the single-entity position. Track the aggregate. If the DAT cohort collectively controls ten percent or more of ETH supply, the float argument becomes quantitatively serious rather than narratively interesting. That is the number to watch. It is not in any of these announcements.
Contrarian: What the Bulls Actually Got Right
I have spent a great deal of this piece taking the structure apart. Fairness requires the other side, and the other side has real arguments that the bears keep dismissing.
First, and most importantly, the staking yield is genuinely real revenue. This is not 2020. In 2020, 'yield' meant token emissions paid to mercenary capital that left the moment the number dropped. Ethereum staking is a protocol payment for consensus work, funded by issuance and priority fees, and it exists because the network needs validators. A treasury earning 3 percent on a $20 billion stack is collecting roughly $600 million annually from an economic activity that is not going away. That is a structural improvement over the bitcoin treasury model, which generates literally nothing.
Second, the buy-and-stake cohort does genuinely contribute to network security. A large, professionally operated, well-capitalized validator set increases the cost of attacking Ethereum, and it does so with capital that is economically aligned with the network's success. The concentration concerns are real, but they sit on top of a base case that is net positive for the protocol.
Third, and this is the argument the bears most consistently ignore: the DAT wrapper provides access that many institutions literally cannot get any other way. A pension fund that cannot hold spot crypto, or a wealth manager operating under a mandate that permits listed equities but not digital assets, can now get ETH exposure through a taxable brokerage account. That is not a trivial feature. It is the entire reason the premium exists, and it explains why mNAV can persist above 1 for extended periods even when the underlying logic seems thin.
Fourth, and I say this grudgingly: MicroStrategy demonstrated the model survives a seventy-percent drawdown. That is not a small data point. The reflexivity loop did not unwind in 2022 even when bitcoin fell catastrophically, because the premium held and the equity issuance stayed accretive. There is now a multi-year track record suggesting the flywheel is more durable than a purist would predict.
But here is the thing the bulls cannot get around, and it is the reason I stay on the other side of the table. Every one of those four arguments is a reason the structure is legitimate. None of them is a reason the structure is safe. Legitimacy and safety are different variables, and the market persistently prices the first while ignoring the second. The staking yield is real. The concentration is real. The access is real. The reflexivity is also real. All four can be true at once, and when the premium finally converges, three of them do not matter at all.
Takeaway
If you take one thing from this teardown, take the methodology rather than the conclusion. Watch the empty fields. Count them. A disclosure that tells you the size of a position but not its cost, its date, its custodian, or its staking method is not a data point. It is a mood. And moods, in a sideways market, are the most expensive thing you can trade against.
The real signal in this cohort was never the $68 million. It was always mNAV. Track the premium. Watch for the first quarter where a major ETH treasury announces a purchase it financed at a discount, or quietly stops issuing, or reclassifies staking revenue, or discloses a custodian change in a footnote. Those are the tells. The whale-size headline is the distraction — cold hands dissect the heat of a hype cycle, but only the ones willing to read the filing first.
