Skip to content

AI Market · Forecast desk · Live · capital

43%

A second AI infrastructure contract above $500M discloses both a future service date and an option, delayed-draw, earnout, or guarantee allocating schedule risk by December 31, 2026.

Probability history

0%25%50%75%100%09-05deadline

Called at 68% on September 5, 2026. Not yet repriced. 73 days to deadline.

How the number was built

Final method: stacked-fixed-weights (resolved n=74 < 80). Each rung is an independent estimate; the final is a stacked combination.

Stated (issue date)68%
Reference class (n=37)46%
Recalibrated57%
Feature model28%
Council41%
Final43%

Drivers

  • Reference class qtype:disclosure runs 0.46 (n=37) vs stated 0.68.
  • Hierarchical calibration moves stated 0.68 to 0.57 (lens capital, kappa_stated=80.0, kappa_lens=200.0).
  • Feature model at 0.28: conjuncts=3, named disclosure=True, horizon=117d.
  • Council of 5 at 0.41.

Cautions

  • all 5 votes came from one vendor (openai); extremization disabled, persona diversity only

Council

5 personas on one model (openai/gpt-5.6-sol); other vendors were unavailable this run, so read this as one model's persona-averaged view, aggregated by log odds mean without extremization.

41%

grid and power economist · openai/gpt-5.6-sol

Large AI commitments remain plausible given the reported capex and RPO growth, but the resolution standard is unusually restrictive. Contracts and 8-Ks often provide deployment or service timing; they rarely publish an ISO queue identifier, and a same-disclosure cross-walk is rarer still. Only 73 days remain, and this must be the second qualifying contract above $500M. Partial credit materially supports the estimate because a service date alone is plausible. Still, the disclosure and multi-leg ledger base rates are only 0.461 and 0.433, while higher Treasury yields and wider high-yield spreads modestly impede capital-intensive agreements. The stated 0.68 is therefore too high.

35%

semiconductor supply chain analyst · openai/gpt-5.6-sol

AI capital commitments remain exceptionally strong, as indicated by large issuer RPO and capex figures, supporting another contract disclosure. However, the resolution standard is unusually restrictive: it requires a second contract above $500M, an eligible filing by year-end, and a same-disclosure crosswalk containing both a future service date and queue identifier. Service dates are plausible, but queue IDs are rarely included in customer contract filings, making partial credit materially more likely than a full hit. Only 73 days remain, while disclosure and conjunctive reference classes are below 0.5. Higher rates and wider lower-grade spreads also modestly reduce near-term financing and signing odds.

41%

capital markets credit analyst · openai/gpt-5.6-sol

The disclosure hurdle is unusually strict: a second contract must exceed $500M and connect a future service date to a specific queue identifier in an eligible, cross-walked disclosure. Rapid RPO and capex growth supports additional large AI infrastructure agreements, but queue identifiers rarely appear in transaction announcements or 8-Ks. Service timing alone is substantially more likely, making partial credit material. Only 73 days remain, spanning limited year-end filing windows, while higher Treasury yields and widening lower-quality credit spreads may slow definitive financings. The forecaster’s 0.68 also exceeds its realized 0.548 outcome rate in that confidence band and the 0.433 multi-leg base rate.

42%

superforecaster statistician · openai/gpt-5.6-sol

I use the 0.433 multi-leg base rate, corroborated by the 0.461 disclosure base rate. Large public-company RPO and capex figures support continued AI contracting, but they do not verify this unusually specific disclosure. Resolution requires a second >$500M contract plus both a future service date and queue identifier, cross-walked in an allowed filing, within 73 days; disclosure timing and the conjunctive structure materially reduce success. Partial credit limits the downside if only one required detail appears. The forecaster’s 0.60–0.69 band realized 0.548 versus 0.660 stated, indicating overconfidence, so 0.68 is too high.

46%

geopolitics regulation analyst · openai/gpt-5.6-sol

The capital backdrop supports continued large AI contract flow, given exceptional public-company capex and RPO growth. But resolution is document-specific: a second contract above $500M must emerge within 73 days and pair a future service date with an identifiable queue entry in the same disclosure. Queue identifiers usually reside in interconnection records, while issuer announcements more often provide only timing and value. Partial credit raises the estimate because a service date alone is plausible. Relevant ledger rates—disclosure 0.461, multi-leg 0.433, and >90-day horizon 0.404—also argue against the forecaster’s 0.68.

Evidence the engine used

direct · undated

SEC.gov | EDGAR Full Text Search

Cited by grid-and-power-economist: Large AI commitments remain plausible given the reported capex and RPO growth, but the resolution standard is unusually restrictive. Contracts and 8-Ks often provide deployment or service timing; they rarely publish an ISO queue identifier, and a same-disclosure cr