When companies get specific about AI, revenue growth looks different

Companies that provide specific evidence of how they use AI tend to record stronger revenue growth. Researchers at Carnegie Mellon University and Larridin examined a study universe of 564 companies across 12 industry sectors. Individual analyses used smaller samples depending on data availability.

The data included 478 corporate 10-K filings, more than 30,000 classified job postings, financial information, market data, and the AI Transformation Tracker built by Larridin.

The tracker assigns companies scores from 1 to 5 across three areas: AI adoption, workforce proficiency and realized impact. It also provides an overall maturity index. The January 2026 Tracker score vintage contained scores for 562 companies, corresponding to 538 after the researchers’ deduplication process.

“Generalized AI investment alone tells us little about a company’s ability to create value,” said Ameya Kanitkar, CTO of Larridin. “What matters is identifying where AI is being deployed, measuring adoption and workforce proficiency, understanding how customers and employees are benefiting, and connecting those efforts to quantifiable business results.”

Detailed AI disclosures linked to revenue growth

One of the most distinct indicators was what the researchers call “narrative concreteness.” The measure looks at how specifically a business describes its AI deployments and results in regulatory filings.

Companies that named AI systems, explained how they were being used and provided measurable outcomes tended to perform better on revenue growth. In the researchers’ adjusted model, companies at the top of the narrative-concreteness distribution were associated with 8.0 percentage points higher year-over-year revenue growth than companies at the bottom.

The researchers evaluated adoption, employee proficiency, realized impact, overall AI maturity, investment intensity, AI-focused hiring and the level of detail in corporate disclosures. Six score- and filing-based measures were significantly associated with revenue growth in unadjusted analyses. The hiring measure was not.

Several of the broader adoption and composite measures weakened once differences in industry, company size and previous growth were taken into account. Detailed descriptions of AI deployments continued to carry information about revenue growth after those adjustments.

Job postings offered another way to examine adoption. A total of 30,861 postings across 536 companies were classified to determine the share of hiring aimed at roles focused on building or operating AI and machine learning systems. The researchers caution that the job-posting data were collected after the revenue period being studied, so the hiring measure should be treated as descriptive evidence rather than a prospective predictor of revenue growth.

AI adoption shows little connection to margins

Greater signs of AI adoption were not associated with improved operating margins. No significant margin effects were found among the public signals examined.

The results provide little evidence of broad operating-margin improvements associated with the AI signals examined. The study did not directly measure individual cost categories or workforce reductions.

Stock market performance followed a similar pattern. None of the public AI signals predicted risk-adjusted stock returns over the following four months after controls and adjustments for multiple testing were applied.

AI infrastructure providers outperform

Companies supplying infrastructure for the AI market produced a different result. AI infrastructure suppliers outperformed sector- and size-matched peers by about 32 percentage points over four months.

Five large semiconductor companies, Nvidia, Broadcom, AMD, Micron and Intel, were excluded from the main analysis to prevent their performance during the AI investment boom from having an outsized effect on the results. Running the calculations with those companies included did not change the main conclusions.

The relationship between detailed AI disclosures and revenue was particularly useful in asset-heavy industries, where implementation may require changes to physical infrastructure, operations and established processes. Specific descriptions can help distinguish companies that have put AI into use from those still discussing plans or early experiments.

The results show association, not causation

The results do not establish that AI caused stronger revenue growth.

Companies that are already performing well may have more resources to deploy AI, measure its impact and provide detailed information about those deployments. Existing growth trends could also influence subsequent results, although prior revenue growth was among the factors included in the main statistical controls.

The work instead identifies a statistical relationship between observable evidence of AI use and revenue growth. The strongest result centers on concrete disclosures rather than broad claims of AI adoption, suggesting that specific information about deployed systems and measurable outcomes may provide a useful signal of how far implementation has progressed.

“The study suggests companies are using AI primarily to expand capabilities, improve customer experiences, and create new growth opportunities,” said Shixiang (Woody) Zhu, Assistant Professor at Carnegie Mellon University’s Heinz College of Information Systems and Public Policy.

“At this stage, AI’s measurable impact is appearing more clearly in revenue growth than in operating margins or stock performance, indicating that its value goes beyond cost reduction.”

from Help Net Security https://ift.tt/q7SNj1b

Related Items