Founders and builders who judge AI by their feed on X see a crowded race. Every visit brings another big funding round, for new research labs, for brokers selling data to model makers, for startups across consumer and enterprise software. Read that way, the money has been handed out and the winners picked.
Put the phone down and call people who run businesses outside San Francisco. Ask chief executives of private and public companies, the owner of the local lawn care firm, or friends and neighbors how they use AI, and many have no strategy and no plan. Some have been talking to consultants and still do not know what AI means for them. Inside big companies, the question is often met with “You mean ChatGPT?”
That answer points to a split that survey numbers also show, between people using a chatbot and companies putting AI to work. In McKinsey’s 2026 survey on the state of AI, eight in ten respondents said AI had improved their own productivity, but only 37% credited it with any effect on their company’s operating profit, about the same share as a year earlier. About 6% counted as high performers, crediting AI with at least 5% of operating profit.
AI agents show how far most companies have to go. At companies with more than $1 billion in annual revenue, 40% of respondents said their organizations were scaling agents, up from 27% a year earlier. At smaller companies the share held flat at 22%.
The industry selling those agents is still the Wild West. There is no standard way to price an AI agent and no standard way to deploy one. The trade in data for training models is chaotic, and evals, the tests used to grade models, have multiplied beyond counting.
That leaves endless room to build companies in AI or around it, to do research, or to work at the frontier, and the work runs a decade or more ahead. Much of the ground is still empty, so a founder can pick something no one else is doing. None of that work depends on eliminating the workforce. The companies in the headlines have their money, and the businesses still asking what AI means for them have barely started.





