Which Companies Actually Use AI? Behind the Index Bloomberg Covered

337 companies in the S&P 500 mentioned artificial intelligence on their Q1 earnings calls, according to FactSet.

That number tells you something important: talk is easy to measure. What companies actually do with AI has been much harder to pin down, because until now the data came almost entirely from self-reported surveys. And self-reported surveys overreport. Ask an executive whether their company is adopting AI and the answer is nearly always yes. Ask the evidence instead, and the answers get more interesting.

That gap, between what companies say about AI and what they demonstrably do, is the problem the AIDE Index was built to solve.

What Bloomberg found in the data

This week, Bloomberg’s Robb Mandelbaum profiled the AIDE Index, our evidence-based measurement of AI execution across the S&P 500. Read the piece here.

The most telling examples in the piece are the mismatches. At most companies, what leaders say about AI tracks with what they do. But as Bloomberg reports from the financial services sector, Capital One’s executives say relatively little about AI while the company sits on the leading edge of deployment, and at MetLife, leaders talk up AI a lot but have built little.

Those mismatches are not anomalies in the data. They are the reason the data exists.

How the index works

The AIDE Index scores every S&P 500 company on four pillars: Literacy (does leadership KNOW AI), Advocacy (do they SAY it publicly), Orientation (do they PRIORITIZE it strategically), and Implementation (do they BUILD it).

Each pillar draws on six evidence channels, spanning executive backgrounds, public statements, strategic disclosures, job postings, patents, and deployed technology. No surveys. No self-reporting. Every signal comes from what companies and their leaders have actually done or published.

Three design choices matter here. First, scores are normalized against sector peers, because comparing a utility to a software company on raw AI signals tells you about industries, not companies. Second, findings are verified by independent models before they enter the index. Third, human reviewers adjudicate the edge cases that automated pipelines get wrong.

The result places every company in one of four archetypes on the AIDE Matrix: AI Trailblazers (high intent, high integration), AI Visionaries (high intent, low integration), Stealth Adopters (low intent signals, high actual integration), and Emerging Adopters (early on both dimensions). The full scoring approach is documented in our published methodology.

The matrix is the honest way to compare across sectors. A raw score comparison between an Energy company and a software company is a comparison of industries wearing company names. The quadrants compare each company to the peers it actually competes with.

What it means

For boards, the finding underneath the rankings is uncomfortable: AI literacy among S&P 500 board members and executives is not where it needs to be. The fix is not another strategy memo. It is executive education, systematic upskilling, and recruiting directors and operators who have built with AI, not just read about it.

For investors, the asymmetry is the story. Talk is priced in. When 337 companies mention AI on earnings calls, the market has fully absorbed the narrative. Evidence is not always priced in, and the distance between the two is where the signal lives.

For journalists, sector-level cuts of the index are available on request.

What comes next

Bloomberg’s piece is a snapshot. The index underneath it is a full map, and over the coming weeks we will publish sector-by-sector findings from it.

Our suggestion: watch the Stealth Adopters. The companies quietly building more than they say may turn out to be the most interesting story in the data.

Frequently asked questions

Source: AIDE Institute, AIDE Index 2026. Company scores are normalized within GICS sectors; the AIDE Matrix quadrant is the only cross-sector comparison the methodology supports. Sector-level findings cited above are as reported by Bloomberg.