(FORTUNE Magazine) – Ever since AT&T divested the Bell System two decades ago, the company has tried desperately to reassemble the assets that would enable it to once again become Americans' sole provider of communications services. In the '90s, management mostly tried to achieve that goal through costly, risky acquisitions. More recently AT&T has tried to offer customers bundles of long-distance and local calls by reselling other companies' services.

These efforts have turned out badly. AT&T eventually sold the wireless and cable companies it had bought. And now its reselling gambit is foundering too: The company is pulling back its residential offers in several states rather than pay higher rates to use the Bells' networks.

Despite that blow, AT&T actually plans to do more repackaging. The company is now renting space on Sprint's wireless network, and will soon begin selling cellular service under the AT&T brand. The company already resells Covad Communications' DSL service in 26 states. Analysts say AT&T may even get back into the pay-television game by striking a deal with one of the satellite providers. Ma Bell, it seems, has gone from being the feudal lord of the telecom industry to a lowly tenant.

AT&T says it is getting exceptionally good wholesale rates from willing suppliers such as Sprint. And by partnering rather than acquiring, the phone company can introduce new services almost immediately.

Still, this new approach has its share of risks, underscored by a recent decision by the Bush administration and communications regulators that effectively tosses out the cheap wholesale rates that AT&T pays to lease the Baby Bells' networks. AT&T now uses the Bell networks to deliver local service to about four million customers; AT&T says the regulatory decision makes it harder to serve those customers profitably, and it plans to stop adding local customers in seven states. This is doubly hard on AT&T: Without access to the Bell networks as an "anchor" to the customer, AT&T may have an even tougher time selling other services, such as wireless.

In other words, AT&T is at its landlords' mercy, in more ways than one. If Sprint experiences network problems, AT&T customers feel the pain too. Even AT&T's own VOIP offering, which delivers local phone service over a broadband connection, "rides" on other providers' DSL or cable-modem service--that means the AT&T service might suffer if the cable is on the fritz.

And making money as a reseller of commodity services is hard--ask any grocer. Many analysts believe the resale gambit isn't really a growth strategy, but one that keeps AT&T hanging on. The partnerships "all have exactly the same underlying theme, which is to maintain AT&T's substantial incumbent position in the long-distance market," says Jeff Halpern, a telecom analyst with Sanford Bernstein.

Fine, but that means AT&T is basically in the same position it was back in the '90s, before management tried to bolster the long-distance business by buying cable and wireless companies. "We are having to move more quickly to the end state that we've desired for some time, which is more control of our destiny," CEO Dave Dorman said in a presentation to employees in June.

Many on Wall Street think the end game for AT&T is selling out to a Baby Bell phone company. (AT&T came close to hooking up with SBC back in 1997 and more recently has held merger talks with BellSouth.) But some strategists are pushing a different kind of combination: an AT&T-Sprint deal.

Such a merger would give AT&T its own wireless network and local-phone assets in 18 states. And though AT&T has the smaller market capitalization ($12 billion, vs. Sprint's $24 billion), AT&T has the dominant position selling data and voice systems to coveted corporate customers, giving it a ton of leverage.

Of course, antitrust officials might have a problem with a deal that puts two big long-distance operators together, but at the right price the combination could make sense.

And at least AT&T would once again be in control of its own destiny. --Stephanie N. Mehta