OpenAI published September research on how workers use AI across tasks beyond their usual occupations. That is evidence about observed usage in the study, not a direct forecast for the revenue or valuation of every AI-linked company.

What the evidence shows

The September OpenAI research discusses patterns of worker use. It is useful evidence about task adoption under its methodology, but it is not an audited revenue statement for every vendor. The BIS review of the digital economy offers a broader financing perspective. These sources can inform a question, not settle a stock valuation.

At company level, compare contracted revenue, realized gross margin, inference costs, infrastructure depreciation and capital spending. A large model-user count can coexist with costly service delivery. A chip supplier, cloud operator and application vendor capture different shares of spending and face different competition.

Why it matters

For investors, adoption, monetisation and infrastructure return are distinct questions. A company can report strong AI demand yet face high compute and power costs. Broad claims that all AI spending will earn the same return need company-level disclosures and comparable methodology.

Deeper context and limits

AI spending has a chain of suppliers and customers. Data centers buy power, networking, memory and accelerators; application firms pay for inference; enterprises pay when tools improve workflows enough to justify cost. Revenue recognized by one layer may be another layer’s capital expenditure. That is why adoption and profitability cannot be collapsed into one growth number.

Research surveys have limits: self-reported use, sample selection and task definitions affect results. An issuer’s earnings can add financial evidence, but forward orders are still subject to cancellations and changing pricing. Analysts should compare revenue growth with gross margin and free cash flow over time, and be explicit when a vendor’s own study is part of the argument.

Bull, bear and neutral cases

Confirmation and invalidation

Look for consistent disclosures across quarters and independent customer evidence. Invalidate a monetization thesis if revenue and margins fail to follow usage.

Check the study’s sample and definitions, then compare actual customer revenue, capital expenditure and margins in issuer filings. Treat a vendor’s research as one source and label forecasts as forecasts.

Reader checklist

• Read the study’s methods and sample.
• Separate vendor claims from company filings.
• Compare capital expenditure with realized cash generation.

Published 26 September 2026. This is a dated report, not a live price feed or personal investment advice. Source documents may be revised after publication.