Put it on tomorrow, Put it on the slate. You can sell the sunrise. There’s a meter on my gate.
“Mr Merdle is the man of this time,” announces Mr Dorrit, well into Little Dorrit. “The name of Merdle is the name of the age.” People speak of Merdle constantly and repeat his name with respect, while the novel withholds any credible account of what his business is. His undertakings bring in such vast sums that people regard them as national benefits, and the confidence spreads through London the way Dickens says an epidemic does, until even Bleeding Heart Yard catches it. The Yard is a place “where there was not one unappropriated halfpenny,” and its residents follow Merdle’s fortunes as keenly as the Stock Exchange does. Mr Plornish, trowel in hand on somebody else’s roof, tells his neighbors that Merdle is the one to put them all to rights.
Merdle turns out to be a forger, which is where the comparison stops. The books examined in what follows are, as far as anyone outside can tell, the books. The sharper point concerns the Yard. Dickens understood that the people with no shares in a great promise are rarely outside it. They live downstream of it, and they are usually the last to see the terms.
A ticket and a timetable
Imagine two instructions reaching the same company on the same morning. A customer asks it to stop a renewal. An executive tells its suppliers to prepare for demand that has yet to arrive. The customer receives a ticket number and a survey about her satisfaction. The executive’s forecast becomes a construction schedule and, somewhere in Ohio or Indiana, a fenced field with a sign on it.
When the Answer Counts followed the first kind of instruction into the machinery that should give it effect, and found how easily a person’s words can be received without anything changing. It asked four things of an instruction: a named owner, a defined effect, a deadline, and evidence that it took effect. The executive’s instruction meets the same test and fails it in the opposite direction. The forecast carries a commercial timetable the suppliers can invoke, and no date on which the promised local benefit becomes demandable. An authorized signature makes other people rearrange their balance sheets (and, eventually, their electricity bills) before anyone knows whether the confidence behind it was justified. A promise becomes power when someone else must budget around it.
AI companies approaching the stock market offer a spectacularly expensive version of that relationship. They sell services people already use, while asking investors to finance capacity for a far larger business than the one that exists. The old argument about whether the machines will transform society now arrives stapled to agreements that oblige someone to pay before the transformation has earned its keep.
The pitch and the proceeds
Anthropic announced a confidential draft registration statement on June 1, and OpenAI announced its own on June 8. A confidential submission is a draft shown to the regulator, which is a long way from shares a pension fund can buy. Reuters reported on October 14 that Anthropic is expected to reach the market first, after the US midterm elections. OpenAI, by most accounts, now leans toward 2027; the New York Times reported in June that Sam Altman regarded listing below a trillion-dollar valuation as a non-starter, which is one way of telling the market what it thinks. Expected dates have a way of becoming previous expected dates, so treat the calendar as weather.
An initial public offering does two different things that tend to share a single headline. Where the company sells new shares, fresh money goes into the business, to build or to repay borrowing. Where existing holders sell, the money goes to them, and the company receives the pleasure of their continued goodwill. Most large offerings mix the two, and the SEC’s own bulletin for ordinary investors sends readers to the use-of-proceeds section for that reason. A first-day pop celebrates both kinds of sale indiscriminately (markets are generous like that).
A valuation, meanwhile, prices a claim on the future. It is a statement about what someone will pay for ownership. Suppliers, unhelpfully, invoice in money. A large valuation helps a company raise more of it, which is useful right up to the day it is needed and the market is in a different mood.
The first signature
The financing debate usually opens with the largest number in the room. Reuters reports that Anthropic’s confidential prospectus describes at least $518 billion of infrastructure spending over roughly a decade. The figure is a sum of agreements with different exits: some payable whether or not the capacity is used, one cancelable on ninety days’ notice. Read as cash already spent, it is merely alarming. Read as a set of dated obligations with different doors, it becomes informative, and The Bill Has a Date, later this week, takes the doors one at a time.
Long commitments can be perfectly sensible. No one pours concrete for a data center on the strength of a vibe, and a laboratory that waited for guaranteed demand before reserving capacity would find the capacity reserved by someone else. The contract is where the parties decide who carries the uncertainty while the building goes up. It names someone who can demand payment and someone who must answer, which is more than most promises manage.
People living near the proposed facility enter through a different door altogether. A utility agreement may protect their tariff; a planning hearing may let them object. Their rights come from those arrangements, and none of them appears in the share offer.
Bleeding Heart Yard, revised edition
Public investment and public exposure reach different people. Someone can buy shares, or decline to. Someone else meets the same expansion through the electricity system their household depends on, or the transmission line across the back field. Whether that second person pays anything is a matter of evidence. A map pin beside a data center proves proximity; the approved tariff and the cost allocation establish who pays.
A useful technology and the price of its maker are separate accounts too. Customers may value the service while investors expect more growth than its economics can carry. A business may flourish financially while a local agreement leaves its neighbors with weak remedies. The accounts can disagree for years, which is bad news for anyone hoping for a tidy verdict.
Hence a ledger, kept line by line, with columns chosen for their dullness. What was promised. Who promised it. Who is exposed if it fails. Who can call the promise in.
Ledger, entry one. Commitment: build capacity for demand that has yet to arrive. Made by: the lab, in its supply agreements. Exposed if it fails: suppliers and the lenders standing behind them. Who can call it in: the counterparties named in the contracts, and so far nobody else.
(A disclosure the ledger asks of its own keepers: the practice behind this newsletter drafts in company with AI models, including models made by two of the firms under examination here. Readers can weigh that as they would any other interest, and are invited to audit the arithmetic.)
Put It on Tomorrow belongs to the woman down the road. It is nocturnal new wave with a dub bassline that supplies all the swagger, and the singer has read the small print. She enjoys the sales pitch enough to dance to it: someone draws a window in the wall and sells the view beyond, and a room full of gentlemen likes the paper it was on. The chorus keeps the slate. In the bridge she admits she likes a little light and would like to leave it on. The instruments stop before the last one, and she asks it plainly, of someone who can answer: who put my name on this? The video animates its storyboard in clay and keeps a ledger of its own between the verses.
Some promises get a ticker. Others get a meter on the gate.
Companions
- When the Answer Counts, on instructions whose receivers own the unfinished work.
- The Account Must Have an Address, on an answerable principal.
- Not Our Building, on consequences that cross institutional boundaries.
- What We Made, on contested accounts of capability and risk.
- The track: Put It on Tomorrow.
Bibliography
Anthropic, ‘Anthropic confidentially submits draft S-1 to the SEC’, 1 June 2026.
Dickens, Charles, Little Dorrit (London: Bradbury and Evans, 1857), Book the Second, chapters 5 and 13.
OpenAI, ‘Confidential submission of draft S-1 to the SEC’, 8 June 2026.
The Next Web, ‘OpenAI may delay its IPO to 2027 to hold out for $1tn’, June 2026, reporting the New York Times.
US Securities and Exchange Commission, ‘Updated Investor Bulletin: Investing in an IPO’, 14 October 2022.
Vinn, Milana, and Krystal Hu, ‘At Anthropic pre-IPO meet: “Thinking” caps, surging revenue, no human extinction’, Reuters, 14 October 2026.
Wang, Echo, and Sabrina Valle, ‘Anthropic’s $518 billion AI buildout hinges largely on deals that cannot be canceled, filing shows’, Reuters, 29 September 2026.
