It all boils down to AMD, a much smaller company than Intel and NVIDIA and yet makes things difficult for both of them in almost every corner of the market. But when we analyze this situation, there are many problems that Lisa Su has to solve in order to maintain this standard.
Two of your competitors will go through everything in 2020 and 2021, is it possible to compete with fewer resources if they consider it a threat?
AMD vs Intel vs NVIDIA, by Lisa Su, between trying and succeeding?
This week we have already announced that Intel will launch its Sapphire Rapids platform with DDR5 ahead of AMD, which server will make a huge difference compared to the EPYC market options.
And that is, even though they include a few cores, the bandwidth will still be very clear that they will return to the remainder. Milan will come without the DDR5 and apparently with the same number of cores, but with the usual jumps it can be a bit, though there are changes to its design thanks to Zen 3.
It looks like it's not enough to compete with Intel's platform, much smaller in 2022, when its competitor will show 7 nm, while AMD is now moving to DDR5 with Genoa and Zen 4.
In the desktop / laptop environment AMD will have to deal with another collision with Tiger Lake-U, Rocket Lake-S and later in 2021 Meteor Lake-S, all in just one year, and the reds will have only Zen 3 on the market.
In the mobile market they have lost the battle and until the new order, on the desktop they will be more competitive, but Landet Lake-S will be the biggest change since Skylake at Intel, so it raises a lot of doubts about what they will be able to offer. The problem is that Meteor Lake will be arriving soon 7 nm, directing competition with Zen 4 and there is AMD that can encounter problems.
The GPU market is not a "bullish" CPU market
If there is already some doubt that AMD can tolerate the amazing speed Intel will set, especially to get out of the way at 10 nm early and go faster than 7 nm, what we will see on the GPU is still more challenging than war and blues.
NVIDIA is profitable, there are many, but this problem will be aggregated by Intel with its Xe GPUs, which is questionable and the mature data is one for rice and one for sand. Assuming they can't compete against the high end or being a novice in smart architecture, Intel will certainly be fighting for the mid-range, AMD's main frontline so far.
At the high-end NVIDIA will introduce Ampere and it looks like Hooper is already ready at 5 nm with Samsung and its new GAA (Nanosheet) transistors. No doubt RDNA 2 will be a Test competition, but it doesn't seem so with Ampere based on the definitions we've seen so far.
The middle and lower range will be a battle that AMD can throw away if Intel offers price pricing for its product and controls the entire chip making process with the card. Also, We know that Intel will try because of the statements Swwan made at the time. They don't expect a fight in this first generation to work out too high, but they do work / value.
Higher profitability equals higher consumer costs
Ultimately, AMD is in the process of changing costs and expenses. It generates revenue and profits, but it is not enough to equip its two competitors, which play on very large margins.
Lisa Su explained that this will focus on AMD in the future prices will fluctuate start producing more profits and invest more resources in R&D. In the end, AMD will cease to be a cheaper option to become a rival for its competitors as long as its products determine the same performance, similar or higher than that of Intel (CPUs and GPUs) and NVIDIA (GPUs).
It is a viable option, to be sure, but an army of supporters often see AMD as a people's choice. Comparing prices on the same or similar item is not an option currently available on AMD, especially because its competitors offer a very robust platform, with better software and more customization.
Thus, this process can take many years to complete, with the possibility that in the next decade AMD will gain products around as much as those of its competitors in standard terms. In the meantime they will continue to try to offer more for less, the bigger issue that gives them sales, and to adapt to the needs of big companies is to steal ratings when their competitors sometimes do not even try to compete for a cheap march.
Will we see AMD as a premium brand? How will this new trend affect your prices? Would you join this fight ARM as a possible successor at low prices? Ten years is a long time in this field, but it is clear that everyone pressing the accelerator and AMD will have to comply quickly if they want to stay in the army, is not easy when we compare the numbers, it is David versus Golium the book where they will be hard to win without taking a few hits.