23%

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

The deadline has passed, and the supplied digest identifies no qualifying primary disclosure. AI infrastructure spending clearly reached enormous scale, but the rule requires an unusually specific combination: a Fortune 500 name, an explicitly stated workload above 50MW, on-prem characterization, and a dated primary document. Corporate disclosures usually report capex, GPU counts, facility capacity, or cloud commitments rather than workload-equivalent MW. Hyperscaler-scale projects do not automatically establish an enterprise on-prem workload, and ambiguity resolves against the forecaster. The broad universe leaves some chance that an unobserved filing or release qualifies, but 0.6 was too high for this documentation-sensitive condition.

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

The deadline has passed, and the supplied digest identifies no qualifying primary disclosure. More importantly, >50MW is exceptionally large for an enterprise on-prem AI deployment; installations at that scale are usually hyperscaler campuses, colocation projects, or planned capacity rather than a named customer’s operational workload. Even if such infrastructure existed, resolution requires the Fortune 500 name and explicit power equivalent in a timely press release, 8-K, or comparable primary document. Ambiguous claims about campus capacity, future buildout, or cloud-hosted workloads fail. These documentation constraints make the forecaster’s 0.6 materially too high.

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

The deadline has passed, but the supplied digest contains no qualifying disclosure. A >50MW on-prem AI workload is exceptionally large for a non-cloud enterprise, and companies usually describe AI capacity through accelerator counts, investment dollars, or data-center totals rather than workload power. The rule additionally requires a Fortune 500 name, an explicit >50MW figure, and a dated primary disclosure; inferred facility capacity or vendor claims without customer confirmation fail. Hyperscaler capex growth supports large deployments generally but does not establish an on-prem enterprise workload under this wording. The forecaster’s 0.6 was high relative to both the strict documentation test and the disclosure reference-class rate.

superforecaster statistician · openai/gpt-5.6-sol

Base rate: 0.461 for disclosure questions. I adjust sharply downward because resolution requires a rare conjunction: a Fortune 500 name, an explicitly on-prem AI workload, a stated >50MW equivalent, and a qualifying primary disclosure by a hard date. The deadline has passed, and the digest shows extraordinary AI capex but no qualifying workload-level disclosure; capex, RPO, or model releases do not satisfy the rule. Ambiguity resolves against the forecaster. The forecaster’s 0.60–0.69 band realized only 0.548, also indicating overconfidence, though the question-specific disclosure burden is the larger penalty.

geopolitics regulation analyst · openai/gpt-5.6-sol

The deadline has passed, and the supplied evidence contains no qualifying primary disclosure. Large AI capex and model-release activity do not establish a named Fortune 500 enterprise’s on-prem workload above 50MW. Enterprises usually disclose GPU counts, total facility capacity, or spending rather than power attributable specifically to an AI workload. Even a clearly larger installation fails if the disclosure lacks the explicit MW/kW figure, qualifying company name, primary-document status, or on-prem characterization. The strict ambiguity rule therefore makes the original 0.6 forecast too high despite the broad pool of potential companies.

https://www.sec.gov/edgar/search/

direct · undated

SEC.gov | EDGAR Full Text Search

Cited by geopolitics-regulation-analyst: The deadline has passed, and the supplied evidence contains no qualifying primary disclosure. Large AI capex and model-release activity do not establish a named Fortune 500 enterprise’s on-prem workload above 50MW. Enterprises usually disclose GPU counts, tot