Back in 2013, NVIDIA's (NVDA 0.48%) management team made a bold move to delay its flagship Tegra 4 product in order to bring Tegra 4i to market. Tegra 4i was an ARM (ARMH +0.00%)-based smartphone processor designed for the high-volume/mainstream portion of the market, and commensurate with that goal, NVIDIA equipped it with an integrated LTE baseband. In theory, this chip should have done well, but in practice it turned out to be a commercial dud with few design wins.
Looking at Tegra 4i: It was a good processor
It's hard to forget how compelling NVIDIA's Tegra 4i looked from a technical perspective, particularly for a platform intended for low-cost handsets:
Source: NVIDIA.
As far as CPU performance goes, this chip handily offers more oomph than the low-end Snapdragon 400 and MediaTek parts that are currently found in the plethora of low-end smartphones today. Ditto for graphics and, given that NVIDIA has traditionally given its chips beefy imaging capabilities, the Tegra 4i was quite competitive there, too.
While there is no doubt that any phone powered by a Tegra 4i would offer superb performance in both "normal" applications as well as gaming performance, the Tegra 4i is at a disadvantage relative to its competition in a number of key ways that, when all is said and done, far outweighed the graphical and computational prowess of the platform.
Connectivity and cost
The low-end/mainstream smartphone market is exceptionally cutthroat. Average selling prices are low so if you're not designing to a very lean cost structure (both at the chip level and the platform level), your platform -- no matter how high performance -- simply won't sell. While only NVIDIA knows for sure, the lack of a connectivity solution coupled with potentially a high cost structure may have been why Tegra 4i failed to gain traction.
On the connectivity side of things, Qualcomm (QCOM 1.43%) -- the leader in mobile processors -- actually integrates 802.11ac Wi-Fi, Bluetooth, and GPS right into the chip, which saves the OEM from needing to buy a separate connectivity combo chip (which run for about $3 apiece at the low end).
Source: Qualcomm.
On top of that platform cost disadvantage, NVIDIA probably suffered at the hands of Qualcomm's and MediaTek's economies of scale. While all three companies build their chips at Taiwan Semiconductor (TSM +0.42%), Qualcomm is Taiwan Semiconductor's largest customer and probably gets more favorable wafer pricing than NVIDIA does (although NVIDIA is not a small customer by any means).
On top of those issues, keep in mind that since the Tegra 4i is vastly more powerful than the Snapdragon 400, it likely consumes more die area in order to get that functionality. Interestingly enough, the Tegra 4i was pretty lean at about 60 square millimeters (NVIDIA claimed that Tegra 4i is about half the die size of Qualcomm's Snapdragon 800, which Chipworks claims is a 118 square millimeter chip). However, it is probable that this is bigger than the various Cortex A7 + PowerVR or Cortex A7 + Mali SoCs out there, which may mean an unfavorable cost structure, although, again, NVIDIA is offering significantly better performance.
It's a commodity market
The high-volume smartphone apps processor market is, according to NVIDIA CEO Jen-Hsun Huang, a commodity market. This is largely true given the ubiquity of CPU and GPU IP. The only way to really excel in a commodity market isn't to necessarily have the highest-performance part (since this means that your cost structure is probably higher), but to offer the desired set of features at the lowest price.
This means that the players with immense scale (Qualcomm, MediaTek) or those with their own manufacturing facilities (Samsung and Intel (INTC +5.42%) are going to have an edge over new players trying to disrupt the incumbents. Given that NVIDIA's strength has been technological differentiation (which has led to very high-value, high-margin GPUs), the mass-market smartphone market really doesn't seem like a good fit in a market that is principally concerned with cost.
Foolish bottom line
NVIDIA's strength as a company is that it is not afraid of leveraging its strong financial position to take risks with large potential payoffs. Was it a mistake for NVIDIA to pull in Tegra 4i at the expense of Tegra 4? In hindsight, probably. However, NVIDIA has learned a valuable lesson about where it is strong (high-performance, high-value, differentiated products), and where they aren't strong (commodity products). These learnings will drive the next set of management decisions and the company will be better for it.
