The purchase of ARM Holdings by NVIDIA has undoubtedly been the transaction of the year in terms of the hardware market, but with it have come a series of unknowns of how this purchase can affect the rest of the market, especially in the Post-PC device market where the ARM instruction and register set is unanimously the standard as it is used by all software.
What problems can manufacturers encounter with the purchase of ARM by NVIDIA?
It is precisely the dependency on the software that makes all the companies that sell SoCs for Post-PC devices that creates a dependency problem on NVIDIA that makes them doomed to get along with NVIDIA and its policies whether they want to or not.
The consequences of the purchase are such that if, for example, Qualcomm decided to make a new CPU that was not compatible with the ARM set of instructions and registers, then that new processor would not be able to execute the programs on the market because it would not understand the code. binary of the program and therefore said processor would have no reason to exist.
The fact that 100% of the programs depend on the CPU being compatible with the ARM binary makes it completely impossible at this time for there to be a migration to another micro architecture.
How does this affect companies that design their own CPUs?
It should be clarified that the CPUs under the ARM register set designed by companies such as Qualcomm, Apple, etc. the only thing they share with fully ARM-designed CPUs is the set of registers and instructions.
Since each of these companies has its own developed portfolio and in the specific case of Qualcomm and Apple if we remove the parts related to the use of the set of records and architecture instructions, the rest is their property and ARM does not paint anything. except when collecting the royalty for the use of the set of records and instructions.
The ones that would be on the tightrope are those that are not only dependent on the ARM set of registers and instructions, but also do not have the ability to design their own processors and depend on the ARM portfolio. These are at the low end of the market and it is possible that NVIDIA will pay more to do without them permanently and sell the designs directly under its brand, which would leave many assemblers dependent on the ARM portfolio out of the market, since NVIDIA could completely replace them and sell them outright.
Indeed, one of the problems that SoftBank has faced is the price of royalties for each chip sold. In some specific cases they are too low for the business model to be profitable and normally these very low royalties come from the low end of the market, which is dominated by purely ARM processors in design to the GPU, so NVIDIA may not It is in the interest of continuing an agreement with these companies for reasons above all economic reasons.
Will NVIDIA-ARM compete against the Microsoft-Intel duopoly?
We don’t know, but NVIDIA could easily create an SoC with which to build an ARM-based PC. The problem is that they need to have the software part and we cannot forget that we are talking about an acquisition at the hardware level and therefore an additional partner would be needed in that aspect.
NVIDIA’s biggest home market is games and they run under x86 and not under ARM. No one can run the latest PC game on a non-x86 CPU so those NVIDIA chips wouldn’t run your favorite game since they speak a different language.
Thus, it would be necessary to convince the hundreds of software developers in the world to port their games to the NVIDIA platform and they do not want to become a Microsoft, or to be more exact, a kind of Apple with its own Operating System.
Is it possible for Microsoft to approach NVIDIA-ARM to create an optimized version of Windows and transition like Apple’s? It would be possible and do not doubt that it happens in the same way that we have seen Qualcomm SoCs in the odd model of the Surface for example, but the NVIDIA-ARM operation does not go in those directions, but rather it is a self-defense movement for part of NVIDIA when it comes to the hardware market.
What is it that has really led NVIDIA to buy from ARM?
One of the things that we are going to see in the following years are multi-chip modules (MCM) in which the CPU and GPU are going to be connected through a proprietary intercom technology from each of the manufacturers. Considering Intel‘s development of the Intel Xe range and that AMD has the Radeon Technology Group, both Intel and AMD could completely oust NVIDIA from some parts of the home and professional PC market by offering integrated CPU and Proprietary GPUs (and not only in the domestic market but in others such as the Data Center and Supercomputer market where Intel and AMD seem to have won the last big contracts).
How does it affect us at the Post-PC device level?
Most likely it will affect with a slight increase in the cost of Post-PC devices due to royalties. Actually, it is too early to know what the consequences will be but it must be clear that it is a titanic acquisition and NVIDIA is not generous in terms of the prices of its products, which will lead to the cost of using ARM technology is somewhat higher and therefore it would affect the margins in the low range making it less profitable for certain assemblers.
But what surely many of you are wondering is if we are going to see any GPU with GeForce architecture within our Post-PC devices. It is possible but not in the short term, since NVIDIA GPUs are not designed to operate under such low consumption so that for a time we may find SoCs that carry a GeForce GPU that is really a last generation Mali until NVIDIA develop its own GPUs for Post-PC devices using joint ARM and NVIDIA technology.
