In 1999, Newsweek ran a piece on how the internet was going to kill shopping malls. The Industry Standard predicted that by 2005, most shopping would be done online. Webvan was going to deliver every grocery in the country in thirty minutes. Pets.com was going to be the future of pet supplies. Toys would come from eToys. Books would come from Amazon. The mall was a relic. The big box was a fossil. The corner store was over.
Twenty-five years later, the share of United States retail spending that happens online is around sixteen percent. The other eighty-four percent still happens in physical stores. Webvan went bankrupt. Pets.com went bankrupt. eToys went bankrupt. The mall changed but didn't die. The grocery store didn't die. Even the bookstore didn't fully die.
The technology was real. The prediction was wrong. The companies that bet on the prediction lost everything. The companies that bet on the technology, more carefully, won the next twenty years. That's the story most worth understanding right now, because the same shape is playing out with artificial intelligence.
What the 1999 Crowd Got Right
The internet did change retail. It changed the long tail of products you couldn't find at the local store. It changed how people compare prices. It changed how people decide what to buy before they walk in. It changed the back office of every retailer that survived. It built three of the largest companies in the world. The 1999 crowd was right about the direction.
What they got wrong was the speed and the shape. They thought the internet would replace retail. The internet absorbed retail. Most physical stores added a website. Most websites eventually added physical stores. The two distribution models melted into one another and the line between them stopped mattering. By 2024, you could order online and pick up in store, or buy in store and have it shipped home, or return online purchases at the register. The interesting question stopped being which channel won. The interesting question was which company could run both at once.
The companies that bet on a clean replacement, where the website ate the store, mostly died. The ones that bet on the technology as a layer on top of the existing system mostly lived.
The Survivors Looked Boring
Look at the companies that came out of the dotcom era as winners. Amazon spent the late nineties losing money and getting yelled at for it. eBay built a marketplace where the sellers were ordinary people. Booking.com, then Priceline, focused on the unglamorous work of inventory and pricing for hotels. Costco ignored the internet for a decade and ran the same warehouse playbook. Walmart built a website and an app and did not throw out the supply chain that made it Walmart in the first place.
The losers were companies whose entire pitch was that the internet would replace something. Webvan was going to replace the grocery store. Pets.com was going to replace PetSmart. Boo.com was going to replace boutique fashion. The pattern was the same. Lots of capital. Lots of marketing. A bet that the existing system would collapse on a schedule that would let the new entrant capture the whole market.
The existing system did not collapse. It absorbed and adapted. By the time the dotcom losers ran out of cash, the incumbents had figured out the website thing well enough that the new entrants no longer had a moat.
What the AI Crowd Is Saying Now
Pull up any tech publication today and you will find the same shape of prediction that ran in 1999. Lawyers are over. Software engineers are over. Customer service is over. Radiology is over. Drug discovery is solved. Schools will be replaced by chatbots. Most knowledge work will be automated by 2027. The valuations on the companies making those claims are larger than the entire United States retail market in 1999.
Some of those predictions will turn out to be directionally right. Artificial intelligence is real. It will change every industry it touches. The technology is not vapor.
But almost none of the predictions about timing or shape are going to be right. That's the part that the dotcom era teaches you. The market for taxis did not die when ride sharing arrived. It changed shape. The market for hotels did not die when home sharing arrived. It got bigger and more segmented. The market for newspapers did die, but it took twenty years and the survivors were unrecognizable. The market for grocery stores adopted online ordering and curbside pickup and is still mostly people walking through stores.
The pattern is consistent. The technology gets adopted. The structure of the industry changes. The replacement scenario almost never plays out the way the optimistic deck said it would.
The Capital Cycle Looks the Same
In March 2000, the technology sector of the Standard and Poor's 500 traded at fifty times earnings. Cisco was the most valuable company in the world on the bet that it would sell the routers for the internet. The bet was right. Cisco did sell the routers. The stock still hasn't fully recovered, in inflation adjusted terms, twenty-five years later. The technology won. The investors who paid 2000 prices for it did not.
The same dynamic is at work now. The leading companies in artificial intelligence are real businesses with real revenue. They are not Pets.com. The fact that the underlying technology works does not, by itself, mean the prices being paid for it today are reasonable. There are two questions. Will the technology change the world. Are these specific companies, at these specific prices, going to deliver returns to the people buying them now. The first answer can be yes while the second answer is no.
That was the dotcom story too. The internet changed the world. Most of the public companies riding the internet narrative in 1999 lost ninety percent or more from peak to trough. The technology was right. The expression of it through the public market was a disaster.
What Actually Happened to Retail
Here's the part that gets left out of the dotcom story. The companies that did the most damage to traditional retail weren't the ones that made the loudest predictions in 1999. Amazon spent its first ten years selling books, then media, then everything, while everybody else flamed out. Shopify didn't exist until 2006. Stripe didn't exist until 2010. The infrastructure of modern e-commerce got built mostly after the bubble popped, by people who watched the bubble pop and learned different lessons.
So if you wanted to know in 1999 which companies were going to dominate online retail in 2024, the right answer involved companies that didn't exist yet, infrastructure layers that hadn't been invented yet, and incumbents that had not yet figured out how to use the internet but eventually would. The right answer was not on the cover of any 1999 magazine.
That's where I'd put my chips on artificial intelligence too. The companies that will benefit most from the technology in 2040 are probably not the companies whose names you read every day in 2026. Some of them will be. Most of them won't be. A lot of the action will end up belonging to companies that haven't been started yet, infrastructure tools that aren't built yet, and old companies that figured out how to integrate the technology without rebuilding from scratch.
What to Actually Do With This
I'm not in the prediction business. I'm a data product manager who watched the last cycle play out and is watching this one.
Three things I keep in mind.
First, separate the technology from the company. The internet was real. Most internet stocks were not good investments at 2000 prices. Artificial intelligence is real. Most artificial intelligence stocks at 2026 prices may not be good investments either. Same logic, same trap.
Second, be skeptical of any prediction that says an existing system collapses on a clean schedule. The schedule almost never holds, and the system almost never collapses. It adapts. The most useful question isn't "what gets replaced" but "what gets reshaped, and which incumbents can ride that change." Those are different questions and the second one points to better answers.
Third, the boring answer is usually closer to the truth. Costco did not get worse because of the internet. Walmart did not get worse because of the internet. Trader Joe's did not get worse because of the internet. They added a website where it made sense and stayed focused on the part of the business that mattered, which was the store. The artificial intelligence equivalent will probably be the firms that quietly adopt the new tools, get a little more productive, and don't make it the headline.
The dotcom bust was the moment when "internet plus existing business" beat "internet replaces existing business." I'm betting "artificial intelligence plus existing job" will beat "artificial intelligence replaces existing job" for almost everything that matters. The technology is real. The replacement scenario is the part that's wrong, and it's wrong in roughly the same way it was wrong before.
If retail wasn't dead, neither is most of what people are now writing the obituary for.