Subscriptions and the Quiet End of Owning Things

At the 2024 DealBook summit, Jeff Bezos recalled visiting a brewery in Luxembourg. The brewery dated from a time before national electricity grids, when businesses that needed power had to generate it themselves. Its old generator was still there, preserved as a museum exhibit [1].

To Bezos, it offered a glimpse of what would eventually happen to private data centres.

"This is what computation is like today; everyone has their own data center. That's not going to last. It makes no sense. You're going to buy compute off the grid."

He was talking about companies running their own servers. Some people later stretched the comment into a prediction that home computers would disappear, which was not what he had said [1].

Still, the broader shift he described is happening. Increasingly, we do not own the things we use. We pay for access to them, often one month at a time.

Tesla is one of the clearest examples.

For years, the company sold Full Self-Driving as a one-time upgrade costing $8,000. Elon Musk regularly described it as an appreciating asset. The idea was that a car capable of driving itself would eventually become more valuable than one that could not [2].

That argument ended on 14 January 2026, when Musk announced that the purchase option was going away.

"Tesla will stop selling FSD after Feb 14. FSD will only be available as a monthly subscription thereafter."

The replacement costs $99 a month. At that rate, a driver will have paid the old $8,000 price after about six years and eight months. The difference is that, at the end of those six years and eight months, the driver still owns nothing [2].

Whether the subscription is a good deal depends on how long someone keeps the car and how much the software improves. But the important change is not simply the price. It is the nature of the transaction. A feature that could once be bought is now something Tesla can continue charging for indefinitely.

BMW Tried Charging for Heated Seats

Car companies had already tested this model in a more literal way.

In 2022, BMW began asking some drivers to pay roughly $18 a month to activate their heated seats. The heating elements were already inside the car. So were the wiring and controls. The customer had paid for the vehicle, including the physical equipment, but still needed an ongoing subscription for permission to use it [3].

The programme did not reach the United States, but it appeared in BMW's online stores in markets including Britain, Germany, New Zealand, South Korea and South Africa. Customers could pay $180 for one year, $300 for three years or $415 for permanent access [3].

Then BMW did something unusual: it backed down.

In September 2023, the company withdrew subscriptions for heated seats and heated steering wheels. Pieter Nota, then BMW's board member responsible for sales and marketing, offered a blunt explanation [3] [4].

"People feel that they paid double, which was actually not true, but perception is reality, I always say. So that was the reason we stopped that."

He also admitted that customers had shown little interest in the offer [4].

"What we don't do anymore, and that is a very well-known example, is offer seat heating by this way. It's either in or out. We offer it by the factory and you either have it or you don't have it."

That sounded like a clear retreat, but only up to a point.

BMW stopped charging monthly fees for hardware such as heating elements. It did not abandon subscriptions for software. The company has continued selling driving and parking assistance as recurring services. It also spent a period charging owners $80 a year to use Apple CarPlay, even though most other manufacturers included it at no additional cost [4] [3].

The heated seat returned to being part of the car. The business model beneath it remained largely intact.

Most People Do Not Know What They Are Paying For

Subscription businesses benefit from a simple fact: people forget.

A YouGov survey across 17 international markets asked subscribers how many of the services they paid for they had used during the previous six months. Only 38% said they had used every one [5].

That means 62% were paying for at least one subscription they had not opened in half a year.

Seventeen per cent had one unused subscription. Another 15% had two. Eleven per cent had three, while 19% were paying for four or more services they had not recently used [5].

The pattern appeared across countries, although some were better at keeping track than others. In Britain, 56% of subscribers said they used everything they paid for. In Hong Kong, only about one in three did [5].

Britain's relatively strong showing makes its national figures no less striking. Citizens Advice estimated that unused subscriptions cost people in the UK £688 million in a single year. Two years earlier, the figure had been £306 million. The organisation also found that 26% of adults had signed up for at least one subscription without meaning to [6].

The gap between what people think they spend and what they actually spend can be enormous.

In one study, 1,000 American consumers were given ten seconds to estimate their monthly subscription costs. The average answer was $86. When researchers walked them through each category individually, the true average was $219 [7].

Nearly one-third had underestimated their monthly spending by between $100 and $199. Another quarter were wrong by more than $200 [7].

A separate survey of 2,500 consumers found that every respondent underestimated their spending. Over three years, the share of people underestimating by more than $200 rose from 24% to 66%, while actual annual spending climbed by roughly $430 [8].

This is one reason the subscription model works so well. Spending goes up while awareness goes down.

Charges for broadband, mobile service, television and software begin to feel less like purchases and more like utilities. They sit in the background, leaving an account every month without asking for attention [8].

A charge people barely remember is not a charge they are likely to cancel.

Cancellation Is Often Difficult by Design

In early 2024, 27 consumer protection authorities from 26 countries examined 642 websites and apps selling subscriptions [9] [10].

They were looking for what the OECD calls dark commercial patterns: design choices that push users towards decisions they might not otherwise make.

Three out of four businesses used at least one such technique. About two-thirds used more than one. Only 156 of the 642 services avoided them entirely [9].

The most common problem involved automatic renewal.

Among services offering auto-renewing subscriptions, 81% did not allow customers to turn renewal off while signing up. A user could accept the recurring charge, but could not refuse it and still complete the purchase [9].

Seventy per cent failed to explain the cancellation process at sign-up. Sixty-seven per cent did not clearly state the deadline for cancelling before the next payment. Another 66.4% required payment details before allowing customers to begin a free trial [9].

A separate investigation by 26 privacy authorities examined more than 1,000 websites and found that roughly 97% used at least one deceptive design pattern when users tried to make a privacy-friendly choice [10].

These are not accidental inconveniences. Someone has to design them, approve them, build them and test them.

A company makes registration easy because it wants people to join. It makes cancellation slower because it wants them to stay. The difference between a one-click sign-up and a phone call to cancel is not a technical limitation. It is a business decision.

The regulators did not determine whether each practice was illegal in the country where it appeared. Most of them probably were not [10].

That is what makes the findings uncomfortable. The behaviour can be deliberate, manipulative and entirely lawful at the same time.

The Price Is Based on the Customer, Not the Product

A subscription price is rarely a fixed reflection of what the service costs to provide.

Consider Netflix. In early 2025, its Standard plan cost $22.89 a month in Switzerland and $2.87 in Pakistan. The same tier was priced at $5.75 in India, $3.66 in Nigeria and $3.37 in Egypt [11] [12] [13].

A Swiss customer was paying about eight times more than a Pakistani customer for essentially the same service.

The software was the same. The catalogue varied somewhat by region, but not enough to explain a difference of that size. The price mainly reflected what Netflix believed customers in each market were willing and able to pay.

Subscription companies can revisit that judgement whenever they choose.

Netflix launched its standalone streaming service in the United States at $7.99 a month in 2011. By 2026, its ad-free Standard plan cost $19.99. That was an increase of 150% over 15 years, spread across nine price rises. There had not been a single reduction [14] [15].

Customers in Britain, Ireland, France, Egypt, Colombia and Nigeria also experienced repeated increases [12].

There is nothing inherently improper about raising prices. Films and television programmes are expensive to produce and license. Streaming companies compete for content, staff and technology, and they are entitled to charge what the market will accept.

But the pattern makes the relationship clear. The company controls the price. It changes that price on its own timetable. Then it relies on the customer's habit, forgetfulness or reluctance to cancel.

The same pricing history contains a neat illustration of how the bargain has changed.

In 2026, the cheapest Netflix plan in the United States cost $8.99 a month and included advertising. That is almost exactly what the ad-free service cost in 2014 [15] [14].

The old price still buys Netflix. It simply no longer buys it without interruption.

Ubisoft Disabled a Game Millions Had Bought

At least a subscription tells the customer that access is temporary.

The more difficult cases involve products people believed they owned.

In December 2023, Ubisoft removed its racing game The Crew from online stores. On 31 March 2024, it shut down the servers required to run it [16].

The game stopped working for everyone, including people who had bought physical or digital copies. Around 12 million players had used it during its lifetime [16].

When challenged, Ubisoft argued that customers had not purchased ownership of the game itself. They had purchased a limited right to access it. The distinction had been included in the terms of service, where almost nobody would have noticed it [16].

The shutdown helped trigger the largest consumer campaign the video game industry had faced.

A European Citizens' Initiative called Stop Destroying Videogames collected 1,294,188 verified signatures. Germany supplied the largest number, followed by France. The total was high enough to require a formal response from the European Commission, as well as a hearing in the European Parliament in April 2026 and a plenary debate the following month [17].

The Commission responded on 16 June 2026 and chose not to introduce a new legal requirement.

It concluded that forcing publishers to keep games playable after withdrawing them from sale would be disproportionate. Instead, it said it would bring industry representatives together before the end of 2026 to discuss a voluntary code of conduct for retiring games [17].

The British government had reached a similar position in 2025, saying it had no plans to change consumer law to address the disabling of video games [17].

Millions of customers had paid for a product. The company turned it off. Under the current rules, that did not necessarily mean anything unlawful had happened.

The Cancellation Rule That Never Took Effect

American regulators did make one serious attempt to address the imbalance.

The Federal Trade Commission introduced a rule requiring companies to make subscription cancellation at least as easy as sign-up. It would also have prevented businesses from forcing app subscribers to deal with chatbots or retention agents before allowing them to leave [18].

The rule was scheduled to take effect on 14 July 2025.

Six days before that date, the Eighth Circuit Court of Appeals struck it down. The court found that the FTC had failed to complete a required preliminary analysis of the rule's costs and benefits [18].

The judges made clear that their decision was about procedure, not an endorsement of the behaviour the rule was meant to stop [19].

"While we certainly do not endorse the use of unfair and deceptive practices in negative option marketing, the procedural deficiencies of the Commission's rulemaking process are fatal here."

The rule did not fail because the court decided cancellation traps were acceptable. It failed because the agency had not followed the correct process.

Among the groups that had sued to block it were the US Chamber of Commerce, a trade body representing companies including Charter, Comcast and Cox, along with media businesses such as Disney Entertainment and Warner Bros. Discovery [18].

In other words, companies that make money from subscriptions helped fund the challenge to a rule intended to make those subscriptions easier to cancel.

The government has continued pursuing individual cases. It has alleged, for example, that Adobe automatically selected a plan carrying an early termination charge equal to half of the remaining annual payments. According to the complaint, the fee was disclosed inside a popup hidden behind a link on the payment page [20].

Those allegations have not been proven.

Even so, the complaint closely resembles what international regulators found elsewhere: signing up is simple, leaving has a cost, and the important details appear somewhere few customers are likely to look.

Ownership Is Becoming an Exception

The line between owning and renting has become surprisingly difficult to find.

The law draws only part of it. Many of the practices identified by consumer authorities appear to be legal. Regulators have tried to strengthen protections, but one of their most significant efforts disappeared over a procedural mistake less than a week before it was due to begin.

What remains is public resistance.

Sometimes enough customers notice the same thing at the same time and complain loudly enough to force a company to change course. That is what happened when BMW tried to charge drivers for heated seats already installed in their cars.

But most subscriptions never produce such a clear moment of irritation.

They cost a few dollars or pounds a month. They renew quietly. They hide among dozens of other transactions. They are attached to cars, printers, software, televisions, games consoles and phones.

BMW's customers noticed because they were being asked to pay not to be cold.

Most subscriptions are less obvious. They take a small amount each month, in exchange for continued access to something that once would have been bought, kept and called yours.