In what would be a smart move, it looks like Nvidia isn't keen with putting all its eggs in one basket when it comes to its next generations of graphics cards. DigiTimes reported today that the chipmaker is dividing orders for its next-gen GPUs, based on the Ampere and Hopper architectures, between both TSMC and Samsung.

Citing unnamed sources familiar with the matter, DigiTimes said TSMC will continue being Nvidia's main source of chips for the upcoming GPUs, which shouldn't come as a surprise. TSMC is a long-standing Nvidia partner, and the Taiwanese foundry has sufficient production capacity to fill big orders. Nvidia has reportedly put in a significant 7nm order for its upcoming Ampere graphics cards, which could include the RTX 3080 line.

Furthermore, Nvidia has even reportedly reserved TSMC's 5nm production capacity in 2021 for Hopper, which is believed to be the microarchitecture succeeding Ampere. This would be wise on Nvidia's part, considering that TSMC is pegged to produce Apple's 5nm Arm-based processors and AMD's rumored Ryzen 5000-series CPUs.

So where does that leave Samsung? Anonymous sources have whispered to DigiTimes that Nvidia is tapping Samsung to produce lower-end Ampere graphics cards. By doing this, Nvidia could effectively lessen the load on TSMC and maximize the yields from both foundries.

It's currently speculated that low-end Ampere products could leverage Samsung's 7nm EUV or 8nm process nodes. Samsung seemingly transitioned to the newer and improved 5nm EUV manufacturing process in the second quarter of this year, DigiTimes said. Its sources also claimed that Nvidia and Samsung are in negotiations as to whether the latter can get a small piece of Nvidia's 5nm orders.