In addition, Java doesn't allow developers to go deeply enough into a phone's software stack to deliver the kinds of experiences customers want, Smith said. So instead of trying to solve Java's fragmentation, AT&T will try to standardize on as few operating systems as possible, and preferably open platforms, he said. That will give third-party developers more opportunity to reach AT&T subscribers with mobile applications, he said.

Mobile data services could "out-value" the Web by adding location awareness and personalized search, Smith said. But as the economy sags and subscriber growth in the U.S. slows, carriers can't keep relying on pulling more revenue out of subscribers to pay for it, he said. AT&T will push for many more advertising-based services over the next year or two, he said. Other possibilities include mobile banking and shopping services that generate transaction fees, and business-to-consumer or business-to-business applications in areas such as health care or transportation. Businesses themselves would pay for those in order to realize greater efficiencies, he said.

Asked how soon AT&T would whittle down its platforms, Smith said he expects "meaningful movement" on the consolidation within two years. The effort might lead to one, two or three platforms, depending on many factors, including AT&T's relationships with device makers and providers of mobile OSes, he said.

Hardware and software platform providers increasingly are putting their own stamp on the devices that run on AT&T's network, but the carrier still sees a market for phones built around customer experiences it builds, Smith said. It wants to own the customer experience for a major part of its core customer base, though it's a market Smith acknowledged is shrinking. But there are plenty of places AT&T could play, he said.

"There are lots of segments within our 75 million subscribers that no one's addressing," Smith said.