We describe it as buying a machine. The invoice increasingly describes something else. A modern postal meter weighs the envelope, prints indicia, pulls USPS rate updates overnight, pushes mailing spend into a cloud dashboard, and flags address problems before anything leaves the tray. It is still called “the meter” in most offices, because that is what the box on the counter looks like. What the vendor is actually selling is a software subscription with a physical endpoint.
This is the small, unglamorous edge of a story playing out across every category of business equipment. The device is the loss leader. The platform is the business.
Pitney Bowes is the obvious case, though not quite in the way the shorthand suggests. The company has not walked away from hardware: its two continuing reportable segments are SendTech Solutions, which sells physical and digital shipping and mailing technology along with financing for the equipment, and Presort Services, which physically sorts large volumes of mail for postal discounts. What changed is where the durable money sits — recurring services, SaaS subscriptions, postage financing, and shipping APIs, all attached to a device that is the cheapest part of the relationship. The company also shows the limits of the pivot. It sold its Software Solutions business — location intelligence, address data, customer information management — to Syncsort in December 2019, and the recurring revenue that survived is the recurring revenue attached to the endpoint. The object on the counter is a customer acquisition cost. The data flowing through it is the asset.
What makes the postal-meter version interesting is how ordinary it looks. A firm is not thinking about digital transformation when it renews the meter lease. It is thinking about a piece of office furniture. And yet the renewal quietly extends a software contract, a data-sharing arrangement, and a dependency on a vendor whose real product lives in a datacenter. The cloud infrastructure spending in the second quarter of 2026 is the aggregated shadow of a million decisions like this.
The same shape shows up wherever a fleet of small devices gets replaced. Label printers in a hospital network are a fair example. The procurement memo says hardware refresh. What the organisation has actually joined is a platform that manages label templates, compliance rules, audit logs, and firmware updates from the vendor’s cloud. The printers still print. But the value the vendor extracts, and the leverage the vendor holds, sits in the software layer. If the buyer ever wants to switch, the printers are the easy part. The workflows built around the platform are the hard part.
The industry term for this is “servitization,” though nobody outside a strategy deck uses that word. The practical version is simpler: sell the machine cheap, or lease it, and monetize the ongoing relationship. John Deere did it with tractors. Rolls-Royce did it with jet engines, charging airlines per engine flying hour for availability rather than selling the engine outright. HP did it with printer ink so aggressively that the printers themselves became nearly free. Peloton built a subscription business on top of exercise bikes and discovered how quickly the model strains when hardware demand collapses. The pattern works when the software genuinely improves the product. It curdles when the software becomes a rent-extraction mechanism attached to something the customer already owns.
The Swedish freight technology company Einride, which raised $113 million in the first half of 2026 ahead of its public-market debut, is explicit about this. Its trucks matter less than its Saga software platform, which coordinates electric fleets, charging, and autonomous operations across Europe, North America, and the Middle East. The truck is the endpoint. The optimization layer is the company. Investors are not buying a vehicle manufacturer. They are buying a logistics operating system that happens to ship with wheels.
The same reframe explains why Google Cloud grew 82 percent year over year to $24.8 billion in the second quarter of 2026 while AWS, still the market leader at 28 percent share, saw its share slip about two points. The hyperscalers are not selling servers. They are selling the substrate that every servitized business runs on. When a postal meter, a tractor, a hospital label printer, and an EV fleet all phone home to the same three clouds, the economics of the endpoint become almost incidental to the economics of the platform beneath it.
There is a quieter part of this shift that deserves attention. When the meter becomes a platform, the buyer’s relationship with the vendor changes shape. Most corporate RFP processes were built over decades to compare boxes: price, throughput, service interval, warranty. They were not built to compare software ecosystems, data policies, integration overhead, or exit costs. That is why a procurement process can be run correctly, by competent people, and still produce a decision that looks cheap on the invoice and expensive on the balance sheet three years later.
The shift also changes who inside a company owns the decision. A postal meter used to be an office manager’s call. A platform decision touches IT, legal, security, and finance. The endpoint is a rounding error. The layer above it is where the margin lives.
Financial services noticed early. Canada’s big banks reported third-quarter results in late August 2026, with RBC’s profit up 11 percent to $6 billion and TD’s net income up 38 percent to $4.62 billion. The branch, once the physical centre of the business, is increasingly a wedge for a much larger set of platform services. TD’s plan to open 100 new U.S. branches by the end of 2028 is not really a real estate strategy. It is a customer acquisition strategy for a business whose economics run on software and happen to include buildings.
Not every servitization story ends well. Meta’s reported plan to convert a large share of its workforce into AI agents, described in August 2026, ran into internal reliability warnings and staff resistance, and the company pulled back — cancelling layoffs and unwinding part of the reorganisation. The lesson was not that platforms are bad. The lesson was that a platform without adequate human oversight is a liability dressed up as a strategy. The postal-meter version of this failure is quieter but familiar: vendors who over-promise what the software layer can do, then leave the customer holding an expensive subscription attached to a box that used to just work.
There is also a risk that gets underweighted in the enthusiasm. Every platform generates data, and data generates its own governance problems. The postal meter that logs every piece of outgoing mail is, in a real sense, keeping records that regulators, litigants, and adversaries may eventually want. When the device becomes a platform, the platform becomes a record. The record becomes discoverable. Few procurement teams price this in.
The measurement layer has its own weak spots. Analytics dashboards attached to platform devices can be gamed, misread, or corrupted by the same kind of noise that has plagued digital marketing metrics for a decade. A dashboard is not a source of truth. It is a source of numbers that a vendor has chosen to show you.
What all of these buyers have in common is that they are being asked to make platform decisions using vocabulary built for hardware. The meter. The printer. The truck. The branch. The language flattens the actual choice. When a buyer signs for a device, they are signing for the software, the data flows, the update cadence, the integration surface, the switching cost, and the vendor’s roadmap. The box is the smallest part of what changes hands.
The mismatch between what companies think they are buying and what they are actually buying is where most of the friction, and most of the value transfer, now lives. The vendors have already made the shift. The buyers, mostly, have not.
The postal meter on the desk still hums when it stamps an envelope. It looks like the same machine firms have used for twenty years. It isn’t. And the sooner the people signing the invoices understand what that machine actually is, the better their next contract will read.