Back in June 2013 Nvidia said it would start licensing the Kepler GPU core, but since then we haven’t seen any developments on this front. However, things may be about to change with Maxwell.

Maxwell is a lot more efficient than Kepler and it may be more attractive to potential clients. Digitimes Research expects Nvidia to “rely on the licensing business” and allow its GPU patents to penetrate the mobile GPU market.

Analysts pointed out that Maxwell offers a 160% performance per area gain over Kepler, along with higher flexibility in the location of the GPU on the die. This added flexibility means chip designers should have a much easier time integrating it in various SoC designs.

Nvidia officially announced the Tegra X1 (Erista) just hours ago and this is the company’s first SoC with Maxwell graphics, built on the 20nm node. The company claims its 256-core GPU can outpace the competition with 1-teraflop performance, while at the same time providing competitive performance per watt.

Licensing Maxwell would make sense, but we still do not know how Nvidia plans to go about this. Whether or not it will offer the latest and most powerful designs remains to be seen. It should also be noted that GPU IP and licensing costs are comparatively low, making up only a tiny fraction of the overall cost of the chip.

A year ago industry sources told Fudzilla that the cost of deploying a mainstream GPU on a SoC tends to be very low, as little as 1-2%, or about 10 cents per mid-range chip. Mobile GPU licensing will clearly not be a big cash cow for Nvidia in the short term, but in the long haul the company could benefit indirectly by offering proprietary technologies and services tied to its GPU technology.