TL;DR: In 2013, 3D printing was going to dissolve the manufacturing industry. Everyone was going to have a printer at home, you’d download a hammer instead of buying one, and global supply chains would obsolete themselves in five years. None of that happened. What happened is that 3D printing found its real lane. Rapid prototyping. Custom medical devices. Aerospace parts. Dental work. A few other categories where it wins outright. The hype died. The technology stayed and got dramatically better at the things it does well.
AI is going through the same cycle 3D printing did, and following the analogy beat by beat says a lot about the next five years.
In 2013, every business magazine cover had a 3D printer on it.
The narrative was uniform across publications. Manufacturing was about to dissolve. Every household would have a 3D printer in five years. You wouldn’t buy a hammer at the hardware store, you’d download the file and print one. Global supply chains would obsolete themselves. The factory floor was finished.
None of that happened. What did happen over the next decade is that 3D printing found its real lane. Several lanes, each smaller and more specific than the hype, each one a place where the technology won. The hype faded. The technology stayed.
AI in 2026 is in the same hype cycle 3D printing was in around 2013, and I find it hard to watch.
The headlines are bigger, the budgets are larger, and the actual technology is doing something narrower than the predictions claim. AI is a tool, and a good one. What I resent is the people selling the prediction, because they get paid either way and the businesses that believe them carry the loss. Walk the analogy beat by beat and you can see, with reasonable clarity, what’s about to happen.
Beat 1: The headlines say the technology will replace everything
3D printing in 2013: “Within ten years, every product you buy will be 3D printed at home. The retail industry is finished. Manufacturing as we know it won’t exist.”
You wouldn’t buy a hammer at the hardware store, you’d download the file and print one.Share on X
AI in 2026 is in the same hype cycle 3D printing was in around 2013, and I find it hard to watch.
AI in 2026: “Within ten years, AI will replace every knowledge worker. Customer service is finished. Law, medicine, accounting, writing, design, all gone. Manufacturing as we know it won’t exist.” Different technology. Identical structural prediction. Both are wrong in the same way. Both assume the technology scales uniformly across every application. That the parts where it works will become the whole. That the parts where it doesn’t work yet will work soon.
The hype phase of the cycle is when you can sell the most books, subscriptions, consulting engagements and workshops. The breathless prediction is the product. I don’t respect people who sell certainty about a technology they’ve used for six months. The business owner who buys the workshop and cuts a department on the strength of it pays for that confidence, and the person who sold the workshop is already booking the next one.
Watch: 3D Printing TimeLapse episode 8 (Prusa Mk3 octolapse), WildRoseBuilds
Beat 2: The first attempts fail spectacularly
3D printing in 2014-2016: MakerBot was the consumer brand that was supposed to put a printer in every garage. It sold to Stratasys for around $400 million in 2013, at peak hype. By 2016 they’d laid off most of their staff, shut down their retail stores, and discontinued the consumer printers. The retail print-on-demand startups went out of business. The home 3D printer market collapsed for four reasons. The printers were too slow. Too expensive. Too unreliable. And the objects they made were worse than what you could buy at a store.
AI in 2024-2026: Klarna fired 700 customer service workers, deployed an OpenAI chatbot, and is now hiring humans back at a higher cost. Air Canada’s chatbot invented a refund policy and got the company sued. The doctor’s office chatbots that trap patients for five minutes per call. Several of the most public AI customer service deployments have already reversed. The full Klarna story is in Why Most AI Rollouts Fail, but the pattern is identical to the MakerBot collapse a decade earlier.
The first deployments fail because the technology was being asked to do what the headlines said it could do, instead of what the technology could do. The companies that deployed first take the loss, and the ones that watched and learned are better positioned for the next phase.
When a company reverses a replacement rollout, the business press covers it as a strategy correction. I can’t stomach that framing in any hype cycle, because the customers who spent months trapped with a bot and the staff who were let go to make room for it had no vote in the decision and carried most of its cost.
Beat 3: The conversation pivots from “this will replace everything” to “this will replace nothing”
After the 2014-2016 collapse, you started reading 3D printing think pieces from a different angle. “3D printing was overhyped.” “3D printing is a niche technology.” “3D printing isn’t going to disrupt anything.” The narrative shifted from doomer-style “manufacturing is dying.” In came the dismissive “nothing was ever going to happen.”
That was also wrong, but it was the new safe take.
The pundits who had been wrong about the inflation now wanted credit for being right about the deflation. The hype curve produces this same flip every time, in every industry that goes through it. The Gartner hype cycle has a name for this beat. It’s called the trough of disillusionment.
AI is going to enter this beat by 2027 at the latest. The first wave of reversals will be in the news. Several public failures will pile up. Some pundit will publish a long-form piece called “AI: It Was All a Lie,” and it’ll go viral. The take will be exactly as wrong as the original hype was.
The actual technology will still be exactly what it was. It’ll be useful for some things, useless for others, and the pundits flipping from hype to dismissal will be wrong both times.
I’ll put my position on the record now so nobody can say I flipped later. When the “AI was all a lie” pieces arrive, I’ll think they’re as lazy as the hype pieces were, because both let the writer skip the work of testing the tool on a real task.
Beat 4: The technology finds its real lanes
3D printing in 2017-2026: Rapid prototyping in product development became a $4 billion industry. Designers can iterate physical prototypes overnight instead of waiting six weeks for a machine shop. Custom medical devices, especially orthopedic implants and surgical guides, became routine. Aerospace adopted 3D printing for complex parts that can’t be machined any other way; GE makes jet engine fuel nozzles by 3D printing. Dental work, hearing aid shells, custom orthotics, all standard now. The construction industry started 3D printing concrete buildings.
None of this is the hype version of 3D printing.
There’s no printer in your garage. You don’t download hammers. Manufacturing didn’t dissolve. What happened is something narrower and durable. The technology found the categories where it wins and stayed there, and the categories where it wins are now worth tens of billions of dollars combined.
AI in 2027-2031 is going to do the same thing.
The lanes are already visible if you stop reading the hype and look at what the augmented humans are doing. Routine summarization, organizing interviews, drafting first-pass outlines, code completion, research synthesis, data cleanup, pattern detection in customer service. The list in What AI Is Actually Good At is the early version of where the durable lanes are forming.
None of those use cases will make a magazine cover. None of them dissolve any industry. Each of them, applied at scale, produces real productivity gains, and the people who get sharper at them become more valuable while the technology around them improves.
Beat 5: The people who learned the real lanes win the decade
The plateau settles below the peak: the real market is smaller than the hype promised, and still worth winning.
3D printing now: The companies that built around the real lanes are doing well.
Carbon, the dental and footwear 3D printing company, is worth several billion dollars. Formlabs, which makes professional 3D printers for the categories that work, has become an industry standard in product design. The startups that built around the hype version of 3D printing are gone, and the ones that built around the boring durable version are running stable businesses.
The people who learned to design for additive manufacturing, in the categories where it wins, have real jobs that didn’t exist in 2013. The mechanical engineer who specializes in optimizing parts for 3D printing is a real role with real demand. The biomedical engineer who designs custom orthopedic implants is doing work that wasn’t possible before the technology matured. The dental technician who runs the digital workflow at a clinic has a more interesting job than the one their predecessor had ten years earlier.
AI is going to produce the same outcome. The augmented humans are going to have more interesting jobs than the unaugmented ones. Customer service, content production, technical support, internal documentation, research synthesis, and a dozen other categories. By 2031 the gap between the two groups will be material. Augmented Beats Replaced. Every Time. walks through what this looks like in practice now.
Some people waited for the hype to be over. Others refused to engage because the first deployments failed. Both groups start from zero in 2030, against people with six years of practice.
The second group’s reasoning is weak. Refusing to learn a tool because a company misused it makes about as much sense as swearing off spreadsheets because somebody’s budget came out wrong. Klarna made a deployment decision, and anyone who treats that decision as proof the tool is useless is handing years of practice to whoever was willing to learn.
What does the 3D printing analogy predict about AI?
3D printing went from peak to crash to real lanes in a decade, and the post predicts AI will follow the same beats.
The 3D printing analogy isn’t perfect. AI is a more general-purpose technology than 3D printing, and the disruption will be broader. But the structural shape of the hype cycle is identical, and the structural shape predicts the next five years with reasonable confidence.
I’m willing to put specific predictions on the record, and I expect to be held to them.
The next 18 months will produce more public AI deployment failures. Klarna won’t be the last. Several Fortune 500 companies will run the same playbook, reverse it publicly, and provide material for the disillusionment phase.
By 2027, the conversation will flip. A wave of “AI was overhyped” pieces will go viral. Some of them will be from people who were previously enthusiastic. The flip will look like analysis. It’ll be the same hype curve doing the same thing it does every time.
By 2028, the real lanes will be visible. Productivity gains in the categories where AI works will be material and measurable. The companies running the augmented model will be outcompeting the ones that ran the replace model. Those that ran neither will be losing market share to both.
By 2031, the people who started learning in 2026 will be the senior augmented professionals. They’ll be doing the interesting work, training the next wave, and earning more than the colleagues who waited. The colleagues who waited will be in their second or third corporate AI training program, trying to catch up to a benchmark that’s already moved.
The boring middle position will turn out to have been right. Neither the hyper position (“AI replaces everything”) nor the doomer position (“AI destroys everything”) will have come true. The position from The Doomers and the Hypers Are Both Wrong will have been right. AI is a useful tool with real limits. The people who learned to use it well became more valuable while the technology around them improved.
My take
Hype around AI will die and the tool will stay, the same way it went with 3D printing. I’d ignore the people promising it replaces everything and the ones who swore it off after Klarna. Pick the lanes where it works in your field and start practicing now. Anyone who waits for the noise to settle starts from zero in 2030, up against people with years of practice.
What should you do about the AI hype cycle?
If you’ve been waiting for the hype to clarify before you start learning AI tools, the 3D printing analogy is your warning.
The clarification arrives late. The people who started learning 3D printing in 2014, when it was being declared the next industrial revolution, are the senior practitioners now. Those who started learning in 2018, when the conversation had pivoted to “3D printing was a fad,” are also doing fine because they got in before the durable lanes were obvious.
The people who waited for the hype cycle to fully resolve and then decided to engage aren’t doing fine, because by then the work had moved past them.
If you run a business, don’t hand this to a consultant and wait for the report. Sit down with the tools yourself for an hour a week on work you already understand. Within a month you’ll know where they help and where they make things up, and nobody can sell you that knowledge.
Don’t wait for the AI conversation to resolve. The conversation will keep flipping. The technology will keep finding its lanes. The people who started learning in 2026 will be the senior augmented professionals in 2031, the way the people who started learning 3D printing in 2013 are the senior practitioners now. The Birth of the Augmented Human walks through what that looks like at scale.
Whichever side of the hype you’re on this year, get the tools onto real work while the pundits argue. Businesses that wait for a magazine cover to tell them it’s safe will spend 2031 competing against people with years of practice, and I think waiting for permission from the same press that got 3D printing wrong is a damn bad bet. The pattern is older than the buzzword, and the buzzword has changed at least four times since 1980.
Hype fades and the useful part stays. The Birth of the Augmented Human is my book on that useful part, AI used as a force multiplier for human thinking.
