A handful of American households pay for AI. Is the future free — or a subscription?

Business A handful of American households pay for AI. Is the future free — or a subscription? June 4, 20265:00 AM ET By  Stephan Bisaha A woman uses a laptop as she lies on the grass in a park in the Manhattan borough of New York City on April 24, 2026. CHARLY TRIBALLEAU/AFP hide caption toggle caption CHARLY TRIBALLEAU/AFP Stay up to date with our Up First newsletter, sent every weekday morning. Kirby Plessas doesn't have an AI subscription. She has two. As a self-described technophile, she uses chatbots outside of work to plan family parties, tweak cocktail recipes and once to diagnose a broken wine cooler's motherboard. All the help Plessas gets from AI justifies the $40 a month she pays for both OpenAI's ChatGPT and Google's Gemini. "I wouldn't doubt that within a year I'll probably have a Claude subscription as well," she said. Sponsor Message Technology AI giant Anthropic prepares to sell stock to the public; files preliminary IPO paperwork But for most people, she believes, free AI is good enough. And judging by their spending, Americans agree. Only about 3% of households were paying for AI in February, using the most recent numbers available from the Bank of America Institute, which researches consumer trends based on the bank's customer transactions. Yet while the number of personal subscribers remains small — plenty of workplaces and universities do pay for AI services — their ranks are growing fast. About 10% more households paid up in February compared to a year earlier, according to the institute. Business Ask AI or just Google it? Google makes a big change to a little search box "If you think back to Netflix and streaming services, at the beginning the growth was quite slow," said Sekoul Krastev, cofounder of the Decision Lab, a research firm with an emphasis on behavioral science. Krasteve said it isn't the norm to pay for AI — yet. "Once that status quo is created, subscriptions will definitely start to go up sharply just the way we saw with str...