The Only Number in the Q3 2026 AI Earnings Story
Companies promised AI gains, and CNBC looked for the payoff in third-quarter earnings. I read the article for numbers and found one, from Santander.
CNBC's roundup covers four European companies: Santander, Novartis, Standard Chartered and Carlsberg. Santander is the only one of the four that gives a figure: its AI initiatives generated 194 million euros of business value in the first nine months of 2026. In the same article, Santander's CFO says AI is embedded in all the bank's processes, "so it's really difficult to just say how much comes from AI." Novartis called the third quarter the first quarter in which AI tangibly improved Novartis's profit and loss (P&L). Novartis did not quantify AI's share. Carlsberg said it does not see AI benefits at scale.
For comparison, the article cites Stanford HAI data, which shows that corporate AI investment reached $581.7 billion in 2025.
The missing numbers do not show whether AI raises productivity. Economists saw the same gap with computing. In 1987 Robert Solow wrote that you could see the computer age everywhere but in the productivity statistics.
Can companies measure AI's ROI?
In a budget meeting, I would ask how we would know whether the AI investment earned a return. These sources mostly couldn't answer that question. "Business value" is a number that a company defines for itself. A measurement needs a baseline, which is an estimate of the quarter without AI. If an AI program has no baseline, its figure is a story.
Managers like to say that you can't manage what you can't measure. By that rule, a company that can't separate AI's share of its results also can't manage its AI investment. If the plan is to spend first and find the number later, is the plan a strategy at all? Broadcom's Hock Tan called AI investors gamblers last month.
