44% of S&P 500 Firms Now Report Measurable AI Benefits

Forty-four percent of S&P 500 companies reported at least one quantifiable benefit from artificial intelligence this year, up from 32% a year earlier, according to a Morgan Stanley analysis of roughly 18,000 corporate transcripts. The jump marks a shift from experimentation to measurable financial results, and it comes with a catch: reported AI costs are rising even faster.

The more telling number sits below the headline. Only 26% of companies reported direct financial benefits — cost savings or revenue generation — up from 16%. Another 32% cited productivity improvements. So the majority of the index still cannot point to money in the bank from AI. That gap between productivity claims and dollar benefits is the one to watch.

Which sectors are actually seeing returns?

Financial services leads, and by a wide margin. Sixty-seven percent of financial-sector companies reported tangible AI improvements over the past 12 months. Communication services ranked second at 61%, followed by information technology at 51%.

Banks, payment processors and investment firms are using AI to automate administrative processes, improve customer service, strengthen risk management and analyze financial information. The dull work, in other words, is where the returns show up first.

Morgan Stanley also found a potential link between adoption and earnings. Companies in the top 10% for frequently reporting quantifiable AI benefits have shown stronger forward earnings trends than the broader S&P 500 on a sector-neutral basis. Correlation, not proof — but a pattern worth tracking.

Why are AI costs suddenly a talking point?

The proportion of S&P 500 companies discussing AI-related expenses rose from 8% to 22% over the past year, nearly tripling. Token usage, inference and API expenses now make up the largest reported cost category at 51%, exceeding general computing and infrastructure spending.

That last figure deserves attention. The cost of running AI has overtaken the cost of building it. The shift indicates companies now evaluate AI investments on financial returns rather than the technology's potential alone. Boards have stopped clapping and started asking for receipts.

Who made the list of 28 leaders?

Morgan Stanley identified 28 companies showing consistent evidence of quantifiable AI benefits in corporate transcripts plus at least 10% upside to the firm's price targets. The group spans:

• Financial services
• Health care
• Industrials
• Consumer businesses
• Energy
• Technology

Named companies include Bank of America, JPMorgan Chase, Citigroup, Visa, GE Aerospace, UnitedHealth Group, Procter & Gamble and AppLovin. Most carry Overweight ratings from Morgan Stanley; seven were rated Equal-Weight. The full list with tickers appears in the accompanying graphic. Analyst ratings, as the standard disclaimer goes, are not guarantees of future performance.

What does this mean for earnings season?

The analysis points to a changing phase of enterprise AI adoption. Companies now face expectations to show how AI lowers operating costs, raises productivity, improves customer experience or adds revenue — not merely that they have an AI strategy.

For investors, the distinction between companies experimenting with AI and those reporting measurable financial benefits could become increasingly important as the technology moves into everyday operations. The findings also flag a structural challenge: maintaining attractive returns as usage, and the operating expenses attached to it, keep climbing.

The next few earnings cycles should show whether the 26% reporting direct financial benefits grows toward the 44% claiming any benefit at all. That convergence — or the absence of it — will tell us whether enterprise AI is a profit story or a productivity story with good publicity.