The cards RTX 40 graphics with a 5 nm manufacturing process will arrive next year, and NVIDIA invests billions of dollars to secure production with TSMC.
2021 is almost over and CES 2022 is just around the corner, and NVIDIA should release a variety of graphics cards, such as the RTX 3090 Ti. And in the second half of next year, the RTX 40s will be launched, using TSMC’s 5nm manufacturing process.
NVIDIA’s Next-Gen GPUs Will Come With Hopper and Ada Lovelace Architectures, both based on TSMC’s 5nm process. Hopper is primarily geared towards data centers, providing artificial intelligence, cloud computing, metaverse, and other scenarios.
Ada Lovelace will be NVIDIA’s architecture for gaming graphics cards and will belong to the RTX 40 family. more powerful core would be the AD102 which would have a maximum of 18,432 CUDA Cores. The amount of CUDA Cores will be increased by 71% and frequencies of up to 2.2GHz could be reached. According to rumors, in FP32 it can hit 81TFlops, which is a 1.25x increase over the RTX 3090.
It seems that companies like Apple, MediaTek and AMD are priority customers of TSMC and of course they paid to be first in line. That is why NVIDIA has had to give a big advance to ensure production at 5nm.
According to rumors, Nvidia advanced 1.64 billion dollars during the third quarter, while now they prepare a new payment of 1,790 million dollars. In total, Nvidia is expected to spend $ 6.9 billion on TSMC. The company will employ TSMC silicon in both its GeForce RTX 40 gaming GPU lineup, as well as its server GPU lineup.