AT&T is again rethinking wireless service. At an investor conference on Tuesday, AT&T CEO Randall Stephenson spoke about changes in how the company provides devices and services to its customers.

“When you’re growing the business initially, you have to do aggressive device subsidies to get people on the network. But as you approach 90 percent penetration, you move into maintenance mode. That means more device upgrades. And the model has to change. You can’t afford to subsidize devices like that.”

To unpack and analyze this, a little history, wireless and otherwise, is in order.

AT&T has been at or near the forefront of changing directions in the industry. It successfully moved large numbers of its customers away from unlimited data plans, with more than 70% now paying for a fixed amount that they can use. Without the simplicity of unlimited bundling of voice, data and text, AT&T has still tried to simplify by grouping those formats in shared plans.

But then there were the devices to deal with.

The mobile device industry is out of control, which is what you would expect in a free market for an exploding technology. Manufacturers can do more and more, more quickly, asking more or less for it, depending on the configuration and profitability demands. The upshot is that smartphones are on an annual improvement cycle (the typically-used 18-month cycle is just bandied about to make it seem a little less crazy). And those smartphones are genuinely expensive, befitting miraculous pocket-sized computers, which they are.

This is where wireless providers like AT&T came in and how it became such a mess.

As the gatekeepers of wireless service, providers find themselves playing two supposedly synergistic roles. When you get to the gate, they sell you service and devices to use the service. One of those roles is relatively simple and straightforward. The other—as a reseller of hardware—has become the problem.

Back in the day, before changes in telecomm capped by the breakup of “Ma Bell” in 1982, it was this easy for AT&T and its customers. You leased a phone from AT&T and you paid whatever regulatory bodies allowed for your service. The attempt to inject the free market into this process more or less worked to radically reform that. You could get your service elsewhere and you could get your phone elsewhere and ultimately anywhere. Pricing for service and devices dropped accordingly and precipitously. AT&T and its emancipated children did not have to be in the business of selling phones, though particularly in the business sector, they still did.

Then came wireless and all bets were off.

Networks and devices came in pairs: if you want AT&T service, these are the cell phones that work on its network. AT&T did not want to be in the device-selling business, but as Stephenson pointed out, that was how you get customers on the network, where you sell your actual moneymaker.

The evolution to smartphones and data seemed on its face an opportunity. Those devices would be hungry for exactly the sort of meal that AT&T cooked up. AT&T would make the devices relatively easy to own. It was Business 101: Give away the razors, sell the blades. Manufacturers devised really nifty devices, applications for those devices proliferated like rabbits, and all should have been right with the world for AT&T, even if it had to subsidize those devices.

But few could foresee the frenetic hyperspeed at which devices would develop. A smartphone barely two years old could become a technological and experiential antique—or so it was made to seem to consumers. AT&T and others always had appropriate upgrade paths, still predicated on the seductively-priced device model.

When the tenability of that model came into question, the industry looked over its shoulder to another industry that has long had to deal with expensive devices: automakers. While the idea of owning a telephone would have seemed strange to consumers in the 1960s, so would the idea of leasing a car. But when money became tight for car buyers, that is exactly what the auto industry turned to. And when wireless providers decided to no longer deeply subsidize $600 smartphones, they came up with the same solution. For the moment AT&T is still offering smartphones at a somewhat reduced price with a two-year contract, but not as reduced as it once was. AT&T would love to leave that model behind and it may well disappear entirely. Instead, you can pay full price for the device or bring your own (with a small monthly service discount if you do), or you can pay on installments. After 20 months, the device is yours, but as with a car lease, you will owe the entire residual amount if you end paying installments early.

Here are items that Stephenson did not account for or disclose, at least publicly at the conference.

By pulling away from its role in the device distribution chain, AT&T will not curb the device development madness or the consumer desire for the latest and greatest, which is always a few months around the corner. Stephenson’s taking a stand is completely bottom-line rational, but is likely to prompt a new dynamic, in which synergy diminishes, replaced by some still-to-be-determined forces.

In essence he has said: This is nuts and we are not playing this game anymore. But if he thinks that the other players—consumers and manufacturers—are about to adapt to AT&T’s direction, it may not be that easy.

Manufacturers have been granted extraordinary freedom by the subsidy-model, freedom which certainly contributed to the accelerated upgrade cycle. They have developed expensive devices that they knew would be discounted and therefore more accessible to consumers. But they are in the business of innovation, and they can’t and won’t just stop. Either they slow down innovation, or they make devices more affordable, or they expect people to shell out big bucks every couple of years. This may or not be what Stephenson had in mind to do: shift the onus, get out from the tight space, and put the manufacturers between the rock and the hard place.

Consumers also don’t want to be left behind. The only thing moving faster than smartphone development are expectations of user experience. A good part of that is software-based, not necessarily requiring a newer or better device. But some of the most appealing and desired features and functions are device-bound. In keeping with Stephenson’s comments, the free market conclusion is that if customers want something, they should be willing to pay for it, if they are able. People might want to drive a Lexus or BMW, but some are just going to have to settle for a reliable Chevy. But that doesn’t mean customers are going to be happy, no matter how rational it is, when they’ve been driving the best for less up to now.

That isn’t the biggest question or unspoken prospect.

Consumers may not want or need as much service as AT&T has prepared to provide and plans to sell. It is evident, from research and from the rise of non-phone tablets, that this is now a Wi-Fi device world. The expense of data drove consumers there, and once they discovered that most of the capability of their smartphones could be accessed via free or cheap, and nearly ubiquitous, Wi-Fi, data and even phone service became the sometimes necessary sidekick turned to if and only if there was no Wi-Fi available. Which other than travelling, is increasingly rare.

All of us—manufacturers, providers and consumers—are rethinking the possibilities.