Full dataset available on GitHub: fortune-500-ai-analysis
This analysis examines AI mentions in SEC 10-K filings across the Fortune 500 from 2022 to 2025. We tracked how often companies mention AI, whether they frame it as a risk or opportunity, and whether AI disclosure correlates with company performance.
Key Findings
AI disclosure has become nearly universal. 85% of Fortune 500 companies now mention AI in their annual filings, up from 29% in 2022. Total AI mentions across all filings grew 601%.
Companies have shifted from optimism to caution. In 2022, 30% of AI-discussing companies framed AI only as an opportunity. By 2025, that fell to 4%. Today, 78% of AI-discussing companies acknowledge both risks and benefits.
High-growth companies talk more about AI, but early adopters didn't grow faster. Companies in the top growth quartile mention AI twice as often as those in the bottom quartile. However, companies that began discussing AI in 2022 did not outperform those that started later or never mentioned AI at all.
Risk awareness is high. 75 companies (18% of AI-discussers) see AI purely as a risk. The remaining 82% who mention AI acknowledge both opportunities and challenges.
Sector attitudes vary widely. 50% of Transportation & Logistics companies that discuss AI see it only as a risk. Just 2% of Software & Technology companies do.
Almost no one quantifies AI returns. Companies selling AI infrastructure (NVIDIA, Broadcom) report concrete revenue gains. Companies buying AI almost never do. Over 20 companies now explicitly warn that AI investments "may not result in material benefits."
Some of the largest companies say nothing. Berkshire Hathaway and Exxon Mobil (combined revenue over $720 billion) have never mentioned AI in four years of filings. 77 companies remain silent in 2025.
Adoption Trends
AI disclosure in Fortune 500 filings has grown substantially since 2022, with a notable acceleration following the release of ChatGPT in late 2022.
Figure 1: AI adoption rate and GenAI mentions over time
Adoption by Year
Between 2023 and 2024, AI adoption more than doubled (33% to 69%), coinciding with the mainstream emergence of ChatGPT and generative AI. By 2025, 36% of all companies (181 out of 500) specifically reference generative AI or LLMs, representing 43% of the companies that discuss AI.
Figure 2: Total AI mentions across all Fortune 500 filings
Top 10 Companies by AI Mentions
Figure 3: All top 10 are Software & Technology companies
How Companies Frame AI
Companies mention AI in different contexts within their 10-K filings. We classified each company based on where AI appears:
- Risk Only: AI appears only in risk factor disclosures (e.g., competitive threats, regulatory uncertainty, cybersecurity)
- Benefit Only: AI discussed only as a business opportunity or capability
- Both: Company acknowledges AI as both a risk and an opportunity
Figure 4: Companies viewing AI as 'pure opportunity' declined from 30% to 4% (of AI-discussing companies)
In 2022, nearly one in three companies that mentioned AI discussed it only as an opportunity, with no acknowledgment of risks. By 2025, that share fell to just 4%. Meanwhile, the proportion of companies framing AI as both risk and opportunity grew from 58% to 78%.
Figure 5: Distribution of AI framing over time
AI Talk and Company Performance
A natural question is whether companies that talk more about AI perform better. We examined this relationship from several angles.
High-Growth Companies Mention AI More Often
Companies in the top 25% of revenue growth mention AI more than twice as often as companies in the bottom 25% (27.6 vs 12.8 average mentions). This pattern is consistent across all growth quartiles.
Figure 6: Average AI mentions and median revenue growth by quartile
But Early Adopters Did Not Grow Faster
If AI disclosure reflected genuine strategic advantage, we would expect companies that began discussing AI early (before the ChatGPT hype) to have outperformed those that joined later or never mentioned AI.
They did not. At least not dramatically. Early adopters (companies discussing AI since 2022) had a median revenue growth of +5.0%. Companies that never mentioned AI had a median growth of +1.9%. The difference (about 3 percentage points) suggests early adopters may have performed slightly better, but causation remains unclear.
Figure 7: Median revenue growth by AI adoption timing
Similarly, companies that specifically mention generative AI did not outgrow those that don't. GenAI adopters had a median growth of +4.2% compared to +2.7% for non-adopters, a difference that is not statistically distinguishable from random variation.
Company Size Explains Much of the Pattern
Larger companies mention AI more frequently and are more likely to mention generative AI specifically. Half of the largest 125 companies (by revenue) mention generative AI, compared to 28% of the smallest 125. This relationship is statistically significant.
Figure 8: AI and GenAI adoption rates by company size
Figure 9: Company size may explain both AI mentions and growth
This suggests the correlation between AI talk and growth may reflect a common underlying factor: larger, better-resourced companies have more capacity both to invest in AI initiatives and to grow revenue. The relationship between AI disclosure and performance does not appear to be causal.
Summary
High-growth companies mention AI more often, but early AI adopters did not outperform late adopters or non-adopters. The correlation between AI disclosure and growth appears to reflect company size rather than a causal relationship. Talking about AI may reflect success more than it predicts it.
The ROI Disclosure Gap
If companies are investing heavily in AI, are they seeing returns? We examined whether Fortune 500 companies quantify AI's impact on their business in their 10-K filings.
The answer: almost never.
The Picks-and-Shovels Pattern
The only companies consistently reporting concrete AI revenue are those selling AI infrastructure, not those buying it. This mirrors the Gold Rush, where the real money went to those selling picks and shovels.
NVIDIA — Data center revenue grew 93% to $115 billion, driven by AI chip demand.
Broadcom — AI-related revenue reached $12.2 billion, up 220% year-over-year.
Micron — Data center revenue more than doubled, with AI demand cited as the primary driver.
For these companies, AI is the product. Revenue attribution is straightforward.
Buyers vs. Sellers
For companies using AI rather than selling it, the picture is different. Claims of AI benefits are abundant. Quantified returns are rare.
Microsoft describes "substantive productivity gains" from Copilot but provides no revenue or efficiency figures.
Alphabet claims AI has "significantly lowered costs" across operations, with no numbers attached.
Accenture reports $3 billion in GenAI bookings, but bookings measure contracts signed, not value delivered.
The pattern holds across sectors: companies describe AI as transformative, but almost none quantify what that transformation has produced.
The Exception: CarMax
CarMax stands out as one of the few non-infrastructure companies to report concrete AI metrics:
"Skye, our AI-powered virtual assistant, is now able to independently answer over half of the questions our customers ask it, reflecting a more than 20% year-over-year improvement. Additionally, the rate of fully self-progressed online sales grew by 25% in fiscal 2025."
This level of specificity is rare. Most companies that discuss AI benefits use language that is enthusiastic but unmeasurable.
The "No Assurance" Trend
Perhaps more telling is a growing pattern of explicit warnings. Over 20 companies now include language cautioning shareholders that AI investments may not pay off.
Fiserv captures the paradox directly:
"These investments may not result in material benefits... Our competitors may implement similar or more effective technologies, reducing any potential competitive advantage. However, if we fail to invest adequately in AI and other emerging technologies, our competitive position and growth prospects may be adversely affected."
In other words: we must invest to stay competitive, but we can't promise it will work.
Verizon is similarly blunt:
"If our AI-related efforts do not evolve at a pace consistent with the developments in artificial intelligence... our business, reputation, financial condition, and results of operations could be adversely affected."
The Competitive Necessity Paradox
This creates a strategic double bind. Companies face pressure to invest in AI or risk falling behind competitors. But they also face pressure to demonstrate returns to shareholders. When returns are uncertain or unmeasurable, disclosure becomes hedged.
The result is a gap between AI as a strategic priority (which it clearly is for most Fortune 500 companies) and AI as a demonstrated value driver (which almost none can prove in their filings).
What Companies Say About AI
Top Risk Concerns
When companies discuss AI as a risk, cybersecurity is the dominant concern, mentioned by nearly half of all companies with risk disclosures. Regulatory uncertainty and competitive threats follow. Notably, workforce displacement ranks relatively low.
Figure 10: AI risk themes by number of companies
Top Benefit Themes
On the benefit side, efficiency and productivity gains lead, followed by analytics and innovation. The presence of "Content/Creative" (78 companies) reflects the impact of generative AI on creative and marketing workflows.
Figure 11: AI benefit themes by number of companies
Other Notable Quotes
Opportunity-Focused
Alphabet
"We have invested more than $150 billion in research and development in the last five years... We believe AI is a profound platform shift, one that can bring meaningful and positive change to people and societies across the world."
Classification: Both | 127 AI mentions
Tesla
"Further improvements and deployment of our FSD (Supervised) capabilities, including through increased AI training compute by over 400% in 2024 and the introduction of our purpose-built Robotaxi product, Cybercab."
Classification: Benefit-only | 6 AI mentions
Mastercard
"We launched Decision Intelligence Pro, the next generation of our Decision Intelligence real-time fraud solution. This enhancement, which leverages generative AI techniques to produce additional data points to help assess the validity of a transaction, boosts fraud detection rates."
Classification: Both | 60 AI mentions
Salesforce
"We introduced Agentforce, a new layer of our trusted platform that enables companies to build and deploy AI agents that can respond to inputs, make decisions and take action autonomously across business functions."
Classification: Both | 133 AI mentions
Risk-Focused
Apple
"This risk may be exacerbated by the use of new and emerging technologies, including machine learning and artificial intelligence, which can involve, among other things, the acquisition and use of copyrighted materials for training as well as the potential reproduction of copyrighted materials in their outputs."
Classification: Risk-only | 10 AI mentions | Despite launching Apple Intelligence
Costco
"Some competitors have greater financial resources and technology capabilities, including the faster adoption of artificial intelligence, better access to merchandise, and greater market penetration than we do. Our inability to respond effectively to competitive pressures could result in lost market share."
Classification: Risk-only | 2 AI mentions
News Corp.
"Generative AI-powered chatbots, search overviews and other tools using models trained or grounded on the Company's content or that produce responses that contain, are similar to or are based on the Company's content without permission, attribution or compensation, have, and may continue to, reduce traffic to, and subscriber demand for, the Company's digital products."
Classification: Both | 94 AI mentions
Sector Patterns
Attitudes toward AI vary substantially by industry. Some sectors overwhelmingly view AI as an opportunity; others are more cautious.
Figure 12: Percentage of companies in each sector classified as 'risk-only'
In Transportation & Logistics, 50% of companies that mention AI discuss it only as a risk, the highest of any sector. In contrast, just 2% of Software & Technology companies are classified as risk-only; 96% acknowledge AI as both risk and opportunity.
Figure 13: Sector anxiety levels have shifted significantly since 2022
Transportation & Logistics showed the largest shift: from 0% risk-only in 2022 to 50% in 2025. Financial Services moved in the opposite direction, declining from 17% to 9% risk-only as the sector has embraced AI for fraud detection, customer service, and process automation.
GenAI Adoption by Sector
Figure 14: Media & Entertainment leads in GenAI adoption; Construction and Chemicals report zero
Non-Technology GenAI Leaders
Several companies outside the technology sector have notably high GenAI adoption.
The AI Infrastructure Premium
Companies that sell AI infrastructure (chips, servers, networking equipment) mention AI significantly more often than other companies. This reflects the fact that AI is a direct driver of their current revenue, not a future capability or operational tool.
Infrastructure sellers mention AI 6.5 times more often than the rest of the Fortune 500.
The top AI infrastructure companies by disclosure volume include NVIDIA (127 mentions), Intel (114 mentions), and Broadcom (89 mentions). For these companies, AI demand directly affects quarterly revenue and is therefore material information for investors.
As noted in the ROI section above, these are also the only companies that consistently quantify AI-driven revenue. For most other companies, AI remains an operational tool or future initiative, important but not yet a primary revenue driver.
Company Comparisons
The Magnificent 7
Among the seven largest technology companies by market capitalization, approaches to AI disclosure differ notably.
Figure 15: AI mentions and classification across the Magnificent 7
Apple is the only company in this group classified as risk-only. Despite launching "Apple Intelligence" as a major product initiative, the company's 10-K mentions AI only in risk factors, primarily concerns about copyright infringement in AI training data.
Tesla is the only company classified as benefit-only. With just 6 AI mentions, all discussing Full Self-Driving capabilities, the company does not acknowledge AI-related risks.
Amazon mentions AI 22 times but does not use the term "generative AI" once, despite operating AWS Bedrock and having invested billions in Anthropic.
Microsoft reduced AI mentions from 143 in 2024 to 117 in 2025, the only company in this group to decrease year-over-year.
Major Banks
Figure 16: AI mentions among major U.S. banks
Major banks show varying approaches to AI disclosure. JPMorgan Chase (18 mentions) and Morgan Stanley (18 mentions) frame AI primarily as a risk factor, while Goldman Sachs (41 mentions) and Bank of America (22 mentions) acknowledge both opportunities and risks.
Retail: Walmart vs. Costco
Figure 17: Two retail competitors with opposite AI postures
Walmart (18 mentions, Both)
"We continue to invest in AI and generative AI technologies to enhance our customers' shopping experience and our associate work experience and to improve efficiencies of our supply chain, operations, management functions and talent recruitment and development."
Costco (2 mentions, Risk-only)
"Some competitors have greater financial resources and technology capabilities, including the faster adoption of artificial intelligence... Our inability to respond effectively to competitive pressures could result in lost market share."
Both companies operate in the same industry, but Walmart frames AI as an investment priority while Costco frames it as a competitive threat.
The Silent Giants
Several of the largest U.S. companies by revenue have never mentioned AI in their 10-K filings over the four-year period analyzed.
- Berkshire Hathaway: Warren Buffett's conglomerate, with $371 billion in 2024 revenue
- Exxon Mobil: the largest U.S. oil company, with $350 billion in 2024 revenue
Combined, these two companies represent over $720 billion in annual revenue. Other notable companies with no AI mentions in 2025 include Delta Air Lines, Abbott Laboratories, U.S. Bancorp, and Paccar.
In total, 77 companies (with $1.6 trillion in combined revenue) remain silent about AI.
This absence may reflect different factors: conservative disclosure practices, limited AI relevance to core operations, or simply different approaches to discussing technology investments in regulatory filings.
Methodology
- Data Source: SEC 10-K filings for Fortune 500 companies, fiscal years 2022-2025
- AI Terms: "artificial intelligence," "machine learning," "deep learning," "neural network"
- GenAI Terms: "generative AI," "generative artificial intelligence," "LLM," "large language model," "GPT"
- Classification: Companies were classified based on whether AI appears in risk factor sections only, business/opportunity sections only, or both. Classification was determined by analyzing the context of each AI mention.
- Revenue and Growth Data: Revenue figures from Fortune 500 rankings. Revenue growth calculated as year-over-year percentage change.
- Statistical Tests: Spearman correlation for continuous variables. Independent samples t-tests for group comparisons. Chi-square tests for categorical comparisons. Standard significance threshold of p<0.05 (95% confidence).
Limitations
This analysis examines disclosure patterns in public filings, which may not fully reflect actual AI investment or capability. Companies may discuss AI differently in other communications (earnings calls, press releases, investor presentations).
Analysis by Pablo Rios | January 2026
Data: GitHub Repository
Press: CFO Dive Coverage
Top comments (0)