Carnegie Mellon Research: Specificity of AI Disclosures Shows Significant Positive Correlation with Corporate Revenue Growth
Carnegie Mellon University and technology company Larridin released a report on August 12, finding for the first time a significant correlation between AI adoption and revenue growth. An analysis of 10-K filings from approximately 500 publicly listed companies shows that firms with the most specific AI disclosures have an average annual revenue growth rate 8 percentage points higher than those with the least specific disclosures. The findings hold even after controlling for industry, size, and prior growth, and after excluding AI chip companies such as Nvidia. The report also notes that AI adoption has no necessary link to operating profit margins or future stock price performance.
Key Findings
- A recent report jointly released by Carnegie Mellon University and technology company Larridin shows that companies with specific content in their AI disclosures havemore rapid revenue growth。
- The study notes that publicly traded companies with the most specific descriptions of their AI initiatives in their most recent 10-K filings saw year-over-year revenue growth rates averaging 8 percentage points higher than those with the least specific descriptions.
- "Many companies are now very specific in their descriptions of AI," said Ameya Kanitkar, founder and CTO of Larridin, in an interview. He believes this reflects a maturation of AI adoption, with companies shifting from experimentation to focusing on "high-value" use cases.
Deeper Analysis
The report, released on August 12, is the first study to show that strong AI adoption is "significantly associated with faster revenue growth," but Larridin emphasized in a press release that thisdoes not necessarily imply better operating profit marginsor future stock performance. Larridin is a technology company that provides an AI return-on-investment measurement platform.
The study controlled for factors such as industry, company size, and prior revenue growth. To ensure results were not unduly influenced by a few high-performing AI chip companies, the primary analysis excluded Nvidia, Broadcom, AMD, Micron Technology, and Intel. Larridin said researchers also tested a full sample including these companies and found that the overall conclusions of the study remained unchanged.
Researchers scored companies on a five-point scale based on the specificity of their AI disclosures. Of the approximately 500 companies analyzed, more than 150 received ratings in the top two tiers for disclosure specificity, with 5 receiving the highest score for reporting deployed AI use cases accompanied by quantified business results.
Visa was one of the companies that received a specificity score of 5, and after disclosing that nearly 26,000 employees had used AI-powered chat tools and describing pilot projects involving AI agents and transactions, its year-over-year revenue grew 17%. In contrast, Conagra Brands saw revenue decline 2% after discussing AI more generally, including upgrading operations with AI and connected data improvements. Larridin gave Conagra a specificity score of 2.5.
A study published in June by AI startup Blue Bridge Group AI reached a different conclusion regarding the correlation between AI disclosures and stock returns. That study found that companies with specific AI strategies hadstronger stock performance。
"Comparing these two studies is really an apples-to-oranges comparison," Kanitkar said. "Our findings only mean that, based on the specific correlation method we used, we did not find a link between AI adoption and stock performance... but that does not rule out the possibility that such a link exists."