A UK-listed issuer filed a share allocation notice last week, and by the time it reached me it carried a Web3 tag. I read it three times looking for the primitive — a contract address, a validator set, a settlement layer, a token supply schedule. There is none of it. What the notice contains is five numbers: 2,710,442 shares allocated; £1,891,933 in gross proceeds; approximately £0.698 per share; a net receipt rate of 98.25%; and 41,213,788 shares remaining under the same subscription agreement, unallocated, with no published timetable attached.
Four of those figures describe something that has already happened. The fifth describes something that hasn't. In a market where price reprices daily and liquidity is thin, only the fifth is load-bearing for anyone holding the equity.
I want to be precise about what this document is before I argue about what it isn't. A UK subscription agreement is not a placing and it is not an open offer. Under a placing, a broker canvasses institutional demand, builds a book, and clears at a price that reflects the marginal buyer at that moment. Under a subscription agreement, the issuer and a defined counterparty agree terms directly. There is no book. There is no price discovery. There is a negotiated number attached to a facility that gets drawn in tranches.
Why companies choose the structure is unremarkable and honest. It is fast. It needs no bookbuild, no roadshow, no coordinated marketing window. For a small-cap issuer that needs cash before a quarter closes, speed dominates price. What the structure surrenders in exchange is transparency: the full agreement — drawdown rights, pricing formula, any floor, any anti-dilution clause — sits between two parties. The public gets a one-page notice and an invitation to trust the arithmetic.
That asymmetry is the entire story here.
The 98.25% net receipt rate is the anomaly, and almost nobody reads it.
Strip fees from £1,891,933 at a 1.75% all-in cost and you get roughly £1,858,824, with about £33,109 left behind in transaction costs. For a UK small-cap equity raise, that is remarkably cheap. Placings on the London market routinely clear at 3% to 6% all-in once you stack broker commission, legal counsel, reporting-accountant work, and the ancillary filings. A 1.75% load is not the cost structure of intermediated distribution. It is the cost structure of a bilateral transaction — a direct subscription in which the only parties earning anything are the issuer's own advisers, and in which nobody stands between issuer and buyer to guarantee price integrity.
Follow that consequence through. If no intermediary underwrites the deal, no intermediary has underwritten the price. The £0.698 is not a market-clearing number. It is an agreed number. It can sit above or below the screen, and the filing does not say which. For an outside holder trying to value the register, that is a missing input, not a rounding error.
There is a second consequence. A £0.698 anchor tells you what one counterparty was willing to pay, at one moment, for a size it could handle. It tells you nothing about depth. It is a single print in a market with no continuous quote. In crypto terms, it is an OTC fill, not an order book — and OTC fills get priced for the fact that they are OTC.
The facility is 6.17% utilised. The remaining 93.83% is an unmodeled supply overhang.
Do the arithmetic. 2,710,442 divided by the sum of 2,710,442 and 41,213,788 gives 6.17%. The company has drawn roughly one-sixteenth of the authorized subscription. At £0.698 per share, full utilisation of the remainder implies a further £28.77 million of issuance — fifteen times the size of the tranche announced this week. That is the number that belongs in the headline. It is absent because headlines carry what cleared, not what is authorized.
Here is the input that makes this filing unanalyzable from the outside: the notice publishes the numerator of dilution and withholds the denominator. There is no total shares outstanding figure. There is no fully diluted share capital figure. There is no indication of what proportion of the existing register the 41,213,788 represents. A holder reading this cannot compute their own dilution ratio. They can compute the issuer's issuance pace. They cannot compute what it costs them.
What that pace implies depends entirely on an intention that is not disclosed. Drawing 6.17% of a facility in a single tranche can mean a company that needs one more shipment of capital and no more. It can equally mean the first of sixteen equal instalments, which would put the full facility on the register inside four years at quarterly cadence — and faster than that if the drawdown right sits with the company rather than on a fixed schedule.
Compare that to how the identical problem is handled in crypto, and the irony gets thick. A token generation event carrying 41 million units sitting unvested would, in any serious project, ship with a published unlock schedule — cliff dates, linear tranches, monthly granularity, often a live dashboard. Retail can model a vesting curve down to the week. They can see the wall before it lands. Set that against this notice: an authorized facility fifteen times the drawn amount, no timetable, and no disclosed contractual commitment for when the remaining shares hit the register.
Code does not lie, but it often omits the truth. A vesting contract that says "unlocked" will unlock. It will not tell you that the unlock lands the same week as a market-making agreement, a treasury reallocation, and a governance vote. The contract is accurate and the picture is incomplete. This filing is the equity-market expression of the same failure mode: every figure is true, and the document as a whole reveals almost nothing about supply risk.
I learned that lesson from implementation detail rather than theory, and the detail is where it always bites. Theoretical cryptography survives or dies at the point where it meets a real system under real load. The same is true of capital structures. The document states what the parties agreed. It does not state what the market will do when the agreement executes.
Watch the tail, not the median. When I benchmarked Optimistic Rollups against ZK-Rollups in 2023 — 10,000 simulated transactions across Arbitrum and StarkNet, gas per transfer and time-to-finality logged per run — the headline metrics converged. Both architectures looked comparable at the median. The differentiation lived exclusively in the tail. Under congestion, the p99 finality spread widened from single-digit percentages into structural asymmetry. Medians are marketing. Tails are engineering.
Apply the same lens. The 2,710,442 shares are the median. Small, already priced, already in the float. The tail is the 41,213,788 that remain undrawn — and the tail is where a holder either survives or doesn't.
Scalability is a trilemma, not a promise — and so is capital formation. Raise at speed, raise at a defended price, and raise without diluting existing holders. Pick two. This facility picked speed and price. It paid for both in dilution rights, and it deferred the bill.
Which raises the variable the notice leaves open, and which decides whether this is a nuisance or a structural problem: whether the remaining 41,213,788 shares are priced at a fixed £0.698 or against a floating benchmark — typically a discount to a volume-weighted average over a trailing window.
Fixed price means the company sold forward at a level it believes it can defend. Floating price means the facility is an at-the-market programme wearing a subscription agreement's clothing. Both produce identical disclosures at announcement. Eighteen months later they diverge completely. The first dilutes predictably. The second dilutes fastest exactly when the equity is weakest, because a discount-to-VWAP mechanism issues more shares as the price falls. That is the mechanical shape of a death spiral, and it is invisible in this document.
The chain is only as strong as its weakest node, and in information pipelines the weakest node is always the classifier. A crypto news aggregator that routes a plain-vanilla UK share allocation into a Web3 category is not lying. It is doing something more common and less forgivable: deriving a category from an unreliable signal — a company name containing the string "Web" — and publishing it as if it were a classification.
Every downstream reader inherits that label. Terminals inherit it. Auto-generated digests inherit it. A retail investor scrolling a feed inherits it and prices the asset against it. Nobody re-derives it, because verification is expensive and trust is cheap. This is the same failure I spent 2022 quantifying on lending protocols: I modeled that a 15% deviation in a price feed, sustained through lighthouse-node delay, could liquidate roughly $2 billion in positions — not because the feed was false, but because it was late. Stale data is more dangerous than wrong data, because it looks correct. A lagging price reference reads as authoritative inside a liquidation engine. A wrong category label reads as authoritative inside an information engine.
In a bear market this stops being academic. Drawdown is visible. It prints daily and every holder watches it. Supply expansion is not visible. It surfaces only at disclosure events, which is exactly why it does more damage: the loss arrives without a ticker. Protocols that bleed liquidity providers over seven days get written up and dissected. Issuers quietly expanding their register by a factor of fifteen get filed under "Web3" and ignored.
So watch the next disclosure from this issuer, not this one. If a second tranche lands within roughly one quarter and prices at or below £0.698, the subscription agreement is functioning as a slow ATM, and the 41,213,788 shares will convert into register weight faster than any published timetable implies. If nothing lands for two or three quarters, the facility is a standby line — expensive insurance the company hopes never to draw.
You will not learn which from this document. That is the point. The filing is complete, accurate, and insufficient — and it will never carry a tag that says so.