The Gap Is Real. The Playbook Isn't.

Noah Epstein, writing on X, uses Census Bureau data, Pew Research, and a striking dot visualisation to argue that AI adoption is far earlier than your timeline suggests — and that this gap represents an "embarrassing" business opportunity. The adoption data is solid. The business advice that follows it is not.

The core observation deserves more attention than it's getting. Only 18.2% of American businesses use AI for any function. Only 34% of US adults have ever opened ChatGPT. The 0.3% of the global population paying for any AI tool is a real number, and it's a useful corrective to the warped perspective you develop from watching people debate agentic frameworks all day. If you think everybody is already using this stuff, you need to get out more. Fair point, well made.

But then comes the pivot, and it's the same pivot every one of these threads makes: here are seven businesses you can start from this gap. AI automation agency. Chatbot installer. Workflow auditor. Prices helpfully included. $2–5K setup, $500/month maintenance. Pick a vertical. Ship a broken V1 on Thursday. Bias to action.

He spends one paragraph — one — acknowledging that "everything comes down to whether you can sell." Then he moves on, because dwelling on it would ruin the thread. But the selling problem is the problem. It's not a minor implementation detail. Finding a dental office manager who will pay $5,000 for a chatbot, surviving the six-month procurement inertia of a law firm, handling the conversation when the automation confidently tells a client something false in week three — that's not the last mile. That's the entire marathon.

The broadband comparison is the tell. "People who helped businesses get online in 2004 made fortunes." Sure. But getting a business online meant installing a commodity technology with a binary outcome: you have a website or you don't. AI implementation is probabilistic, fragile, and requires ongoing maintenance against a backdrop of capabilities that shift every quarter. The website guy could say "done." The AI automation guy is never done.

Every price point in the piece is presented as fact. Not one is attached to an actual client, a named business, or a verifiable engagement. For an article built on "the data says," the business model section runs entirely on vibes.

The opportunity to help businesses adopt AI is real. It is also a consulting business — which means it requires domain expertise, sales ability, operational reliability, and the patience to build trust one client at a time. Describing it as something you can learn in weeks and ship on Thursday is how you get a generation of people burning through their savings building chatbots nobody asked for.

The gap between what AI can do and what businesses have adopted is enormous. The gap between identifying that opportunity and actually capturing it is the part this thread skips. That second gap is where all the money is, and all the difficulty, and all the reasons most people who try this won't make it.

Naming the opportunity is easy. That's a tweet. Building the business is the hard part. That's a career.