Dean Bubley: Wireless infrastructure & handset software - counter-synergies?

Dean Bubley at Disruptive Wireless ponders why there aren't more companies/partnershior that provide both wireless network infrastructure and applications servers, and embedded software for mobile handsets. He states:

Generally, the only reason that an infrastructure company needs its own handset client software is if it has a proprietary product. If the technology isn't standardised or at least "open" to external developers, this means that the vendors needs to offer an "end-to-end solution". Great in principle, but it inevitably means it will only work on a restricted range of devices. Which is fine if the service is so compelling that customers will "take what they're given" in terms of phones, but not if they're average consumers deciding they need a pink RAZR or a black Chocolate before they think about what applications it can support.

Bubley cites some "successful examples of one company producing both sides of the equation, servers/infrastructure and also devices or device software" such as RIM , Motorola with its iDEN network & devices for Nextel. He then states "Macromedia/Adobe might get there with Flash. Opera's Java-based Mini client is getting some traction too. Maybe Real, and bits of Microsoft."

He also points out that "with the exception of iDEN (which looks like it's now on the wane anyway) all of these are "high level" applications, which can be ported relatively easily across devices and sit on top of smartphone OS's or featurephone app stacks or inside Java. They don't need too much messy integration with chipsets and protocol stacks."

One area where he thinks infrastructure vendors need their own handset clients or integration expertiseis with UMA. Bubley writes that

even though it's now been adopted by 3GPP as a standard, it's notable that its prime advocate, Kineto Wireless, has had to develop a large handset client arm to help foster development of UMA chipsets & phones. I can't believe it's actually making much money out of this - it's actually a "means to an end" to help it sell its networking hardware. No infrastructure company really wants to get their hands dirty with porting to 101 varieties of handset silicon, and the onerous & expensive development and testing work that entails, for a measly 50c or $1 per handset and some low-margin consulting fees. They'd much rather sell $100k or $1m lumps of tin & software.

Another example is IPWireless. Nobody seems particularly interested in making UMTS-TDD phones at the moment, so it has to try & stimulate demand itself, with its own reference design and a prototype developed with Atmel and UTStarcom.