According to a recent report from Hardware Times, Nvidia has spent about $9 -$10 billion protecting TSMC’s 5nm chips for the company’s next RTX 4000 series. According to Hardware Times, Nvidia will spend at least $10 billion to acquire a share of 5nm chips from competitors such as Apple, AMD, and Qualcomm.
In addition to Nvidia spending billions on 5nm chips, Intel also said that Intel wants to produce “millions” of Arc GPUs to solve current problems. Intel could also make its products more expensive in response to TSMC’s incredible growth and expansion; Intel plans to spend $25 billion to $28 billion on chip production in 2022. Designing the best graphics cards will be a challenge by 2022, but reports suggest Nvidia will spend at least $10 billion on GeForce RTX 4000-series components.
So Nvidia spends a lot, if that were true it certainly wouldn’t come as a surprise as mentioned. Nvidia recently said that Nvidia’s card offering will improve in the second half of this year when the next RTX 4000 series arrives, so while availability may be better, suggested prices could be much higher than expected.
Discrete NVDA GPUs likely outperform AMD (more on that later); therefore, investors should not look for growth here. NVDA dominates the discrete GPU market with an 80% increase in market share, and AMD will wipe out what’s left. When it comes to integrated GPUs for PCs, AMD is roughly on par with NVIDIA, while Intel (INTC) dominates about 68% of the market. Until recently, AMD was second only to INTC as a supplier of x86 chips. To strengthen its position in both business areas, the company last year planned to acquire ARM Holdings (ARMHF) from parent SoftBank for $40 billion.
In the last fiscal year, Intel (INTC) spent more than $13.5 billion on research and development, NVDA spent nearly $2.83 billion, and AMD spent just over $1.9 billion on research and development. Compare last year’s spending for the same quarter when Jensen Huang and company spent $2.54 billion. I should add that NVIDIA is shrinking AMD’s share of the discrete GPU market, and I believe this trend will continue, in part due to a mismatch in R&D budgets.
NVIDIA will update the RTX 3070 Ti 16GB to use more memory with higher efficiency than the current RTX 3080 series. In theory, this will be part of the fierce competition that Nvidia will face on the 5nm process as rival GPU manufacturer AMD is rumored to use 5nm for its high-end RDNA 3 GPUs, which are also expected to arrive later in 2022, for the RTX 4000 solution.