Circular financing scale

Vendor-financing commitments signed in the trailing four quarters between AI-stack suppliers and their lab customers (equity stakes, guarantees, take-or-pay contracts, backstops), from filings and company announcements; derived metric circular_commitments_total. Letters of intent, talks and self-reported aggregates are on the ledger but excluded from the total.

Reason through this

The reading

Over a trillion dollars of circular commitments in a year, most of it filed. The Nvidia 5-year CDS (on the ledger) is the market's stress read on it.

Direction over the last 4 readingseach dot links to its record

in the rule's windowearliermoves within ±$20.0B read as stable, and a move counts only if more than half the steps go its way; this window moved +$218B and 2 of 3 steps rose; higher reads concentrating

reported

11 readings drawn from the derived metric circular_commitments_new_4q; 55 observations in the evidence.

Derived (circular_commitments_new_4q) from: SEC EDGAR filings (8-K, 6-K, 10-Q exhibits); NVIDIA Newsroom; Official Microsoft Blog; Anthropic newsroom; About Amazon company news; CNBC; TechCrunch; SiliconANGLE; CoreWeave newsroom; Microsoft Source (news.microsoft.com); SoftBank Group press releases. SEC EDGAR archives. NVIDIA Newsroom. Official Microsoft Blog. Anthropic newsroom. About Amazon. CNBC. TechCrunch. SiliconANGLE (relaying the WSJ report). CoreWeave (company announcement). Microsoft Source. SoftBank Group.

The numbers

as of	reading	value	record
2026-09-30	—	$998B	reported
2026-06-30	—	$1.2T	reported
2026-03-31	—	$979B	reported
2025-12-31	—	$781B	reported
2025-09-30	—	$418B	reported
2025-06-30	—	$62.1B	reported
2025-03-31	—	$57.3B	reported
2024-12-31	—	$14.2B	reported
2024-09-30	—	$8.2B	reported
2024-06-30	—	$6.2B	reported
2024-03-31	—	$16.3M	reported

Why it matters

When Nvidia guarantees OpenAI's leases, Microsoft's backlog is half OpenAI, and Amazon funds the customer that commits to AWS, revenue and capital are the same dollars going round the stack. The ledger is where the circularity is filed.

What cuts against it

Commitments are not cash; "up to" amounts and ten-year floors are counted at face; equity in a customer is only circular if the customer spends it with the investor.

How it is measured

derived metric circular_commitments_new_4q (formula in the semantic layer)
source About Amazon company news · default tier 4 · Amazon; short quotation
source Anthropic newsroom · default tier 4 · Anthropic; short quotation
source CNBC · default tier 5 · CNBC; short quotation
source CoreWeave newsroom · default tier 7 · CoreWeave press release; short quotation
source Official Microsoft Blog · default tier 4 · Microsoft; short quotation
source Microsoft Source (news.microsoft.com) · default tier 4 · Microsoft; short quotation
source NVIDIA Newsroom · default tier 4 · NVIDIA; short quotation
source SEC EDGAR filings (8-K, 6-K, 10-Q exhibits) · default tier 4 · Public domain (U.S. government)
source SiliconANGLE · default tier 5 · SiliconANGLE; short quotation
source SoftBank Group press releases · default tier 4 · SoftBank Group; short quotation
source TechCrunch · default tier 7 · TechCrunch; short quotation
proxy capital_flow · cadence quarterly

Three quarters of the trailing-year flow; moves under $20B are one small deal. Rising commitments concentrate the stack's financing on a few counterparties, and unlike the cumulative total this series can fall.

Applied to metric:circular_commitments_new_4q.

Model comparison (lower AIC fits better; both on ln(horizon) residuals)

fit	estimate	95% interval	n	R²	AIC	inputs
`circular_commitments_total`	$1.4T	—–—				sources1 2+53 more

Derived rows (11)

as of	value	inputs
2026-09-30	$998B	sources1 2 3+31 more
2026-06-30	$1.2T	sources1 2 3+32 more
2026-03-31	$979B	sources1 2 3+27 more
2025-12-31	$781B	sources1 2 3+22 more
2025-09-30	$418B	sources1 2 3+13 more
2025-06-30	$62.1B	sources1 2 3+5 more
2025-03-31	$57.3B	sources1 2 3+6 more
2024-12-31	$14.2B	sources1 2 3+4 more
2024-09-30	$8.2B	sources1 2 3+2 more
2024-06-30	$6.2B	sources1 2 3+1 more
2024-03-31	$16.3M	source1

Latest readings

2026-09-30 · $998Bas of 2026-09-30(34 obs)
2026-06-30 · $1.2Tas of 2026-06-30(35 obs)
2026-03-31 · $979Bas of 2026-03-31(30 obs)
2025-12-31 · $781Bas of 2025-12-31(25 obs)
2025-09-30 · $418Bas of 2025-09-30(16 obs)
2025-06-30 · $62.1Bas of 2025-06-30(8 obs)

Why this status, and its history

The ledger was swept under a written rule on 28 Sep 2026 (sellers of chips, cloud, data-centre capacity and power, and their AI customers, from 1 Jan 2024), adding 41 deals and correcting Amazon's Nov 2024 row to its new $4B. Over the last three quarter ends the trailing-year flow rises from $781B (Dec 2025) to $979B (Mar 2026) to $1,179B (Jun 2026), every step well beyond the $20B dead band, so the rule reads concentrating. The earlier unclear reading compared an unsearched year with a searched one.

2026-09-28unclear to concentratingconf 75 → 75 · Claude (model)
The ledger was swept under a written rule on 28 Sep 2026 (sellers of chips, cloud, data-centre capacity and power, and their AI customers, from 1 Jan 2024), adding 41 deals and correcting Amazon's Nov 2024 row to its new $4B. Over the last three quarter ends the trailing-year flow rises from $781B (Dec 2025) to $979B (Mar 2026) to $1,179B (Jun 2026), every step well beyond the $20B dead band, so the rule reads concentrating. The earlier unclear reading compared an unsearched year with a searched one.
Weighed against it: Face values count 'up to' amounts and ten-year floors; a handful of very large deals (Oracle-OpenAI $300B, Microsoft-OpenAI $250B, Amazon-OpenAI $150B across three rows) set most of the level, so one cancelled deal would move it sharply. Commitments are not cash, and the sweep may still miss deals whose dollar values were never confirmed by a party.
sources1 2 3 4 5 6+28 more
2026-09-12concentrating to unclearconf 75 → 75 · Claude (model)
The direction now reads the flow, not the stock: commitments signed in the trailing four quarters ($557B to 2026-08-17), where it previously read the cumulative total. A cumulative total can only rise, so the old rule could never return dispersing. Over the last three quarters the flow rose, rose again and then fell, so the steps disagree and the reading is unclear.
Weighed against it: The cumulative total is unchanged, still rising and still the headline figure on the ledger and in the saved analysis. One large deal moves a trailing-year flow by more than the dead band, so a single quarter's fall is not evidence that the financing cycle is cooling.
sources1 2 3 4 5 6+3 more
2026-09-10unmeasured to concentratingconf — → 75 · initial seed (model-drafted)
Cumulative signed circular commitments (equity, guarantees, take-or-pay, backstops; LOIs, talks and self-reported aggregates excluded): Dec 2025 $641B, Mar 2026 $673B, Jun 2026 $798B, Sep 2026 $903B. Rising by far more than the $20B dead band each quarter; the largest single items are OpenAI's $300B Oracle contract (reported), $250B Azure commitment (Microsoft blog), Nvidia's $105B lease guarantee (10-Q) and Anthropic's $100B+ AWS commitment (Amazon). Initial seed.
sources1 2 3 4 5 6+8 more

Why this confidence

Rests on 55 observations from About Amazon company news, Anthropic newsroom, CNBC, CoreWeave newsroom, Microsoft Source (news.microsoft.com), NVIDIA Newsroom, Official Microsoft Blog, SEC EDGAR filings (8-K, 6-K, 10-Q exhibits), SiliconANGLE, SoftBank Group press releases, TechCrunch; best evidence tier 4, grade B; current.

Confidence is independent of status: 90–95 multiple strong independent sources; 70–89 good evidence, some ambiguity; 50–69 mixed or hard to operationalise; 0–49 limited or vague evidence.

Predictions that lean on this

Happening
Anthropic's revenue keeps growing about tenfold a year. · Dario Amodei
Happening
Leopold Aschenbrenner projected that total annual AI investment could exceed a trillion dollars by 2027. · Leopold Aschenbrenner
Too early to tell
AI earns trillions of dollars a year in revenue before 2030. · Dario Amodei
Too early to tell
AI needs about two trillion dollars of revenue a year by 2030 and falls short. · Bain & Company (Global Technology Report 2025)
Too early to tell
The gap between AI spending and the revenue to repay it keeps widening. · David Cahn

More predictions that lean on this

Too early to tell
AI's revenue falls far short of what the spending on it needs to repay. · David Cahn (Sequoia Capital)
Too early to tell
The largest cloud companies spend many times more on AI buildings and chips than AI earns. · Tomasz Tunguz
Too early to tell
NVIDIA keeps financing its own customers in return for a share of their income, and uses such deals to gain a say over who gets computing capacity. · Byrne Hobart