What The Gold Rush Forgets To Buy
A Christian Science Monitor piece last Friday put the AI-in-schools market at seven hundred and thirty million dollars this year, projected at eighteen and a half billion by 2036, growing forty percent a year for a decade. What districts are buying is the tool. What tends not to make it onto the same purchase order is the layer underneath it.
Last Friday the Christian Science Monitor put a number on the market that grew up around the phrase AI in schools. Seven hundred and thirty million dollars in 2026. A projection of eighteen and a half billion by 2036, growing forty percent a year for a decade.1 The chart in the piece climbs the way the ones for the early ed-tech gold rushes climbed. Learning platforms. Devices. Analytics. Each one a wave the district budget rides for a season, buys, and then quietly asks, five years in, what it was.
The wave has a name this time. Generative AI. And the vendors have new names, though the marketing letters are familiar. MagicSchool, SchoolAI, Buddy.1 MagicSchool alone reports eight million educator sign-ups and a partnership footprint that reaches roughly one in five students in the United States.2 The purchase orders are moving faster than the guidance.
What The Order Line Says
A purchase order for an AI tool describes a product. Chat with your lesson plan. Give feedback on a student paragraph. Draft a rubric. Suggest a question a fifth grader might ask. All useful. All buyable. What the purchase order does not describe, because it is not a product line, is the record of what the child actually did when the tool was on her screen.
That absence is not accidental. A tool that produces output can be priced and demoed. A layer that captures a student's revising, her erasing, her deciding to keep the second version, is quieter. It does not demo well. It does not have a slide. Yet the layer is the thing a teacher who reads a paragraph on Monday needs on Tuesday, the thing a parent asking how her child is doing needs on Friday, and the thing the child, eventually, needs when she tries to remember how she got somewhere.
The Two Numbers Under The Number
The same Monitor piece pulled two numbers most gold-rush charts do not. Seventy percent of teachers say they worry AI is weakening students' critical thinking and research skills. Over half of students say using AI in class makes them feel less connected to their teachers.1 Both of these are the sound of a layer eroding. Not because the tools are bad. Because the tools are what got bought. The layer, the record of a student's thinking that used to sit in a teacher's head after a week of watching her work, is not what got bought.
A district signing a contract this month is answering a question quickly, under budget pressure, with a product. The question was what to do about AI. The answer was a subscription. The question underneath the question is harder. What did we mean by learning before, and what part of it are we still willing to see.
The Discipline Of A Slower Purchase Order
There is a version of this month that does not look like a rush. It begins by naming the layer the district wants to protect and then asks the vendor to describe how the tool touches it. A writing tool that produces a draft is a different purchase from a writing tool that captures the student's revising while she works. A tutoring tool that answers a question is a different purchase from one that leaves a legible trail of where the student paused. The difference is not in the price. The difference is in what the tool leaves behind for a human to read.
At Koan we spend our days thinking about the shape of that trail. The pauses. The rewritten sentence. The moment the student picked one word over another and can, later, say why. We think about it because when we lose the trail we lose the case for a teacher who is more than a monitor. Making the layer visible is one of the ways a teacher stays present in a classroom that has already bought the tool.
What The Gold Rush Does Not Ship
Every gold rush ends the same way. The wagons empty out. What was worth buying and what was decoration becomes clear a few years later, in hindsight. The districts that come out well tend to be the ones that spent a little less on the shiny thing and a little more on what the shiny thing did not include. In 2026 the shiny thing is a chat window. What it does not include, unless the district asks for it and pays for it, is the record of the child on the other side.
If your school this month is buying a tool that produces output, is it also buying a way to see the student who used it, or has that line been left off the order?
References
Education's AI 'gold rush' comes with risks for students
The Christian Science Monitor · August 7, 2026
'Nobody Wanted to Give A Former Principal Money': How An Educator Built An Edtech AI Startup With $63M From VCs
Crunchbase News · 2026
Sources cited in order of appearance. Click any inline number to jump.