MediaTek will reportedly build a flagship SoC with an integrated Nvidia GPU for 2024

You might recall that the Samsung Exynos 2200 chipset was the first produced by Samsung to include a Radeon graphics processing unit (GPU) made by AMD. The partnership brought ray tracing to mobile devices allowing for a more realistic portrayal of light in video games played on mobile devices. Last month, Samsung and AMD said that they are extending their multi-year agreement which means that we should see AMD GPUs in more Exynos chips.

In a similar vein, fabless chip designer MediaTek (fabless meaning that it relies on a foundry like TSMC to build its chips) is reportedly talking to Nvidia to supply a GPU that would be used on a flagship MediaTek chip next year. According to DigiTimes Asia, MediaTek and Nvidia will also work together to create Windows on Arm (WOA) components for notebooks. These could be used in mid-range to high-end notebook computers. MediaTek currently concentrates on supplying SoCs for entry-level Chromebook devices and has a 20% share of that market.

The MediaTek chipsets developed with Nvidia GPUs will enhance AI capabilities of smartphones and will also improve gaming functionalities. The first MediaTek chips integrated with an Nvidia GPU are expected to start shipping sometime next year.

Sources inside TSMC tell Digitimes Asia that thanks to orders received from Nvidia for its A100/H100 and A800/H800 chips, the world's largest foundry was able to sell out its 7nm/5nm production capacity for the third quarter. A seasonal increase in demand for gaming devices should help Nvidia see business improve during the second half of the year.

Back in 2020, Nvidia offered $40 billion to buy chip architecture firm Arm Holdings from SoftBank. While Nvidia is the largest GPU designer in the world, the acquisition would have given Nvidia the chance to add Arm's Mali GPU chips to its lineup. The FTC put the kibosh on the deal not only because it would lead to the combination of two top mobile GPU suppliers but also because the regulatory agency was afraid that Nvidia wouldn't allow other chip designers like Qualcomm to license Arms's cores.