This won't come as a huge surprise, but Cnet reports Nvidia didn't do very well at all in Q2 this year, with revenues dropping sharply by 71 percent compared to the same time last year, making only $52.6 million. Nvidia even made more from the patent license it has with Intel (66 million) than it did from the mobile chip business.
"These decreases were largely due to lower unit shipments of Tegra 3 processors as customers ramped down production of Tegra 3-based smartphones and tablets, as expected. They also reflect lower revenue associated with the sale of embedded products and game consoles, offset by increased revenue from automotive products."
Basically, Nvidia was at the very end of the product life cycle of Tegra 3 in Q2 this year, as Tegra 4 devices such as SHIELD and some tablets using it like the HP Slatebook X2 won't arrive until Q3 this year. Nvidia also failed to sign another contract with Google for the new Nexus 7, and last year the Nexus 7 represented the bulk of Tegra 3 sales.
This also means that Nvidia probably won't make a lot of money in the coming quarters either, because it doesn't look like Tegra 4 will be in some very popular products later this year, and neither will the "smartphone chip" Tegra 4i.
This is also what's the main problem with Tegra 4. It seems to have been specifically built for tablets – which is code word for not being a very efficient chip that could also be put in smartphones. They raised the clock speed of Cortex A15 too much for the 28nm process, and they made another small die size of 80mm2, which means there wasn't a lot of space left for the GPU, and they had to raise the clock speeds of the GPU cores too much in order to be competitive.
The mobile Kepler GPU is looking impressive, at least on paper, but there's a high chance Nvidia will still make a small 80mm2 Tegra 5 chip next year, and worse yet, they'll most likely build it at 28nm, instead of 20nm like everyone else next year. So either it will end up being a "mid-range" chip in terms of performance like Tegra 3 was, if they also want to put it in smartphones, or it will be competitive in performance, but too power consuming to be put in smartphones, or even most tablets, and then the story of Tegra 4 will repeat itself. It also seems better to repeat the Tegra 3 story rather than the Tegra 4 one. At least the "mid-range" Tegra 3 got them some sales and revenue.
I for one am pretty excited about mobile Kepler, even though it seems delayed, since initial rumors were saying it will arrive in Tegra 4, but I'm not as excited about the whole Tegra 5 chip. The one we can probably be excited about without expecting an inevitable disappointed yet again, is the Denver/Maxwell-based Tegra that will be made on the 16nm FinFET process. However, we'll have to wait until 2015 for that one. Hopefully, Nvidia's mobile chip business can survive until then, because we need to keep the competition high in the mobile chip market.