A little while ago, NVIDIA's (NVDA +2.08%) Chris Evenden, speaking at the Wells Fargo Fourth Annual Tech Transformation Summit, pointed out that Tegra 4i -- NVIDIA's first integrated baseband and applications processor -- didn't see the kind of OEM uptake that the company had initially hoped for. As a result, Tegra 4i will not be getting a successor. However, this does leave open the question of the ultimate fate of Icera, which NVIDIA acquired to develop cellular baseband technology for its platforms.
NVIDIA doesn't really need it
At the aforementioned conference, Evenden, NVIDIA's senior director of investor relations, asserted that connectivity -- in this case, presumably cellular connectivity, as NVIDIA doesn't develop other connectivity technologies like Wi-Fi, Bluetooth, and NFC -- was still an important part of the company's automotive strategy. The implication here is that NVIDIA would continue to develop standalone cellular baseband processors.
That said, developing cellular basebands is expensive, particularly if the chip company wants to stay on the leading edge of that technology. For automotive applications (as well as high-end tablets and phones), NVIDIA doesn't actually need to provide its own cellular baseband and could instead pair its Tegra processors with third-party chips.
In fact, do you remember the Xiaomi Mi3 smartphone that NVIDIA won with its Tegra 4? That used a Spreadtrum cellular baseband. Further, Tegra has been successfully paired with third-party connectivity combo chips (think Wi-Fi, Bluetooth, NFC, and so on) from the likes of Broadcom (NASDAQ: BRCM) in phones, tablets, and the infamous Project Shield.
For the very high end of the market, owning all of the pieces of the platform is nice, but it's not strictly necessary, as OEMs will mix and match the best parts from various vendors.
NVIDIA should wind down or sell Icera
At this point, it seems inevitable that NVIDIA will wind down or sell Icera; cellular baseband simply isn't NVIDIA's core competency. Fortunately, since the markets the company is targeting with Tegra don't require integrated baseband or connectivity, it makes little sense to spend all of that R&D money to come up with a solution that's inferior to what could simply be bought off the shelf from the likes of Qualcomm (QCOM 2.85%).
The argument that some will make, then, is that by using off-the-shelf cellular products, NVIDIA doesn't capture as much of the platform value and has to pay Qualcomm's margins. This may be true, but the R&D expense for developing modems is quite high, so -- depending on the size of the market -- it may make much more sense to take the gross-margin hit to get to better overall operating margins.
Foolish bottom line
I contend that NVIDIA does not need Icera. The company has no problems using Wi-Fi, Bluetooth, and NFC products from third parties like Qualcomm, and even in its most publicized smartphone win to date, a third-party cellular chip was used.
The R&D required to stay current with cellular modems and RF is extremely high, and the incumbents have deeper pockets and a head start on NVIDIA and Icera. There is no shame for NVIDIA to admit that RF and cellular is not its core competency and to ultimately wind down or sell that part of its business. The research and development saved could either be shifted to better return-on-investment projects or simply returned to the shareholders.
