How non-tech incumbents built moats nobody noticed — and why it’s happening again right now
The most consequential technology bet of the internet era wasn’t made by Amazon. It wasn’t Google, or Netscape, or any of the companies whose names defined the era.
It was made in Bentonville, Arkansas, by a retailer who had already been quietly building data infrastructure since 1988.
When Walmart launched Retail Link in 1991 — a supplier portal giving vendors real-time access to their own sales data — nobody wrote breathless magazine pieces about it. No IPO. No Silicon Valley pedigree. Just a supply chain intelligence system that forced Procter & Gamble, Unilever, and every other supplier to connect their operations directly to Walmart’s inventory data.
By 1997, Retail Link had become the business. Walmart’s supply chain was so far ahead of Kmart and Sears that the gap was essentially permanent. Both eventually filed for bankruptcy. Walmart didn’t.
This is not a story about the internet. It’s a story about what happens when an established company understands its data asset before anyone else does — and builds infrastructure to exploit it.
Three Companies. One Pattern.
Walmart had the data asset: a network of suppliers, real-time point-of-sale data from thousands of stores, and a CEO who understood that information was as valuable as inventory. The technology (EDI, then internet-based portals) was a means to an end. The end was frictionless supplier coordination at a scale nobody else could match.
Capital One had the data asset: credit application data and transaction history at a time when most banks were still issuing cards based on rough demographic segments. Richard Fairbank and Nigel Morris had a thesis they called Information-Based Strategy — that you could use statistical testing and mass customisation to price credit risk more precisely than any competitor. They built the systems to test 10,000 offers simultaneously. By the time the internet arrived in 1994, they already had the infrastructure. The internet just made it faster.
Domino’s had something different: a burning platform. By 2008 the brand was failing. Surveys said customers preferred the taste of cardboard. CEO Patrick Doyle made a counterintuitive bet — not on better pizza first, but on better data. Digital ordering launched in 2007. By 2015, more than half of US orders were digital. By 2019, 65%. The stock went from $3 to over $400 in a decade. Doyle’s famous line: “We’re a technology company that happens to sell pizza.”
The New York Times is the cautionary contrast that makes the others legible. The Times nearly didn’t survive the internet era. It gave away content for free for a decade while its print revenues collapsed. The metered paywall launched in 2011 — later than most observers thought possible, earlier than most competitors managed. Within four years it had become the business. Today the Times has more than 10 million digital subscribers. Most of its print peers are gone.
What separates the Times from the papers that died isn’t the paywall technology — anyone could build that. It’s that the Times had an asset (trusted journalism at scale) that was worth paying for, and eventually found the nerve to charge for it.
What These Cases Have in Common
All four companies were pioneers. None of them were technology companies. All four built internally rather than buying or partnering. All four eventually turned the technology into the core of the business — not a feature, not an efficiency gain, but the thing the business runs on.
And all four faced the same internal resistance: “We’re not a technology company. Our customers don’t want this. Our competitors aren’t doing it.”
The companies that listened to that resistance are, in most cases, no longer here.
The Pattern That Repeats
There’s a concept I call the Digital Ratchet. Each technology wave permanently raises the baseline for what established companies must be capable of. The companies that absorbed the internet wave are a different organism than those that didn’t. You can see it in their cost structures, their customer relationships, their talent density, their ability to move.
The ratchet never goes back.
What’s striking about the four cases above is not that they moved fast. Walmart’s EDI precursor to Retail Link started in 1988. Capital One’s statistical testing methodology predates the commercial internet. These companies moved early — but more importantly, they moved with conviction about what their data asset actually was.
That’s the harder part. It’s easy to identify a technology. It’s hard to look at your operations and see the data moat you’re sitting on that nobody else has.
The AI Parallel
Right now, in 2026, the same pattern is playing out.
Non-tech incumbents are sitting on data assets that AI makes newly exploitable. Patient records. Maintenance logs. Transaction histories. Claims data. Supplier networks. Decades of sensor data from industrial equipment.
The companies that understand what they’re sitting on — and build infrastructure to exploit it before the window closes — are building the Walmart Retail Links of the AI era.
The ones waiting for the technology to mature, or for competitors to move first, are making the same bet that Kmart made in 1991.
The internet wave took about a decade to separate the incumbents who got it from those who didn’t. The AI wave is moving roughly five times faster.
The ratchet is turning.