Amazon's logistics gambit: From 2-hour deliveries to becoming commerce's invisible infrastructure

Amazon has cracked a problem that seemed impossible just a few years ago: the three-hour delivery window. Order a phone charger at three in the afternoon, and before you've had time to regret the purchase or start a load of laundry, a driver is at your door with the package in hand. The company showed off this capability this week at its Delivering the Future summit, held at its largest European logistics hub outside London, and made clear it has no intention of slowing down.

The speed isn't magic—it's mathematics. Behind the scenes, a predictive algorithm powered by artificial intelligence is essentially guessing what you'll want before you know you want it. The system analyzes your purchase history, what your neighbors are buying, the weather outside, and local consumption patterns to move products into position hours before you click to order. Amazon calls this anticipatory shipping taken to its logical extreme. The algorithm doesn't require massive warehouses in the middle of empty plains. Instead, it forces the company to position its fastest-moving inventory in a network of small urban micro-warehouses, tucked into the neighborhoods where people actually live. When you hit buy, the product is already three kilometers from your house. All that's left is a label with your name.

This year, Amazon plans to expand two- to three-hour delivery to 25 additional European cities, though it hasn't named which ones. In the United States, the company is investing $4 billion to extend ultra-fast delivery into small towns and rural communities, a deliberate effort to make instant gratification available everywhere, not just in New York, London, or Madrid. In the United Kingdom, Amazon has already begun testing 30-minute delivery windows for groceries and essentials—a service previously available only in America.

But the real ambition extends far beyond getting packages to customers faster. Amazon has built a logistics empire that costs obscene amounts of money to maintain: over a hundred cargo planes, thousands of facilities worldwide, and a delivery network in the United States that now moves more volume than UPS or FedEx. The problem is that this infrastructure was designed to handle the savage peaks of Black Friday and Christmas, which means for most of the year, Amazon has excess capacity sitting idle. On Wall Street, idle machinery is considered a mortal sin.

So the company has decided to rent out its infrastructure to anyone willing to pay. Amazon Supply Chain Services is the name of the new business, and it represents a fundamental shift in how the company thinks about itself. Amazon no longer cares as much whether you buy from its platform. What it wants is for you to use its machinery even if you're selling from the store across the street. An order placed on Shopify, TikTok Shop, or even Walmart could end up traveling in an Amazon truck, managed by Amazon's algorithms, without the customer ever knowing. The ultimate goal is to become the invisible operating system of global commerce.

This playbook is familiar because Amazon executed it two decades ago with cloud computing. In the early 2000s, the company built a massive server infrastructure to keep the web from crashing during the Christmas rush. When executives realized how much money that infrastructure cost to maintain while sitting idle the rest of the year, they decided to rent it to other companies. Amazon Web Services was born, and it became the financial engine that sustains the entire group today. The logic is relentless: the bill for building the infrastructure is already paid. Now Amazon just needs other companies to help amortize it.

AWS succeeded by turning the idle capacity of computers and servers into a profitable business. Now Amazon wants to repeat the trick, swapping data bytes for cardboard boxes. The plan is to become the operating system of delivery vans. Major brands like Procter & Gamble and 3M have already started using the service. For a company, the pitch is nearly irresistible: instead of juggling four different shipping companies and three separate warehouses, you hand the keys to one organization that has every incentive to make the network run flawlessly—not just to keep you happy, but to keep its own e-commerce empire generating money.

Amazon's real advantage isn't trucks or warehouses. It's data. While traditional logistics operators work with static forecasts, Amazon's software recalculates in real time: traffic patterns, demand in the next neighborhood, the probability of combining three orders into a single delivery route. The company has assembled an almost obsessive collection of solutions for every scenario. In city centers, it uses electric bikes. In the United States, it deploys Rivian vans. In the Mediterranean, it operates ships connecting Spain and Italy. There are cargo planes, mules in the Grand Canyon for inaccessible zones, and boats along the Maine coast for island deliveries. Drones get the headlines, but they're almost an afterthought in a philosophy oriented toward using the most efficient method for each meter of the journey. Even in pharmacy, Amazon uses small allied drugstores and AI to process paperwork, delivering prescription medications the same day.