PC hardware is more and more integrated, there are fewer and fewer parts, and we are getting dangerously close to a chair game where the choice of components is increasingly limited. If to this we add that Intel has started making GPUs, NVIDIA is looking to make CPUs and AMDs to support an ecosystem only with its products, then we have the clear conclusion.
Intel, AMD or NVIDIA PCs only for the future?
The moves of Intel, NVIDIA and AMD in recent years have allowed the three companies to not only increase their technology portfolio, but also to be able to realize new classes of hardware where previously there was no involvement of the one of these three companies. If we start with Intel, we bought it a few years ago from FPGA maker Altera and the recent development of its Intel Xe GPUs, with which they intend to enter a market so far occupied by AMD and NVIDIA at different levels, either integrated high-performance GPU for high-performance computing.
The second in the line is NVIDIA, which is in the midst of an ARM purchase operation, but they have already been manufacturing their own CPUs for some time and the takeover of Mellanox has allowed them to increase their portfolio in the face of interconnecting components . both at the processor level and between processors. Finally, do not forget AMD which has one foot on the graphics card market and the other on that of processors. The recent takeover of Xilinx also allowed them to increase their portfolio.
All this allowed them not only to penetrate new markets, allowing them to create complete systems without the participation of the other two. This has been made worse by the use of proprietary interconnections which only allow the use of equipment from the same manufacturer. Especially in laptops where standard interconnects such as PCI Express are not used to communicate components with each other.
Proprietary interconnections are the key
One of the problems that different hardware design companies have studied the most is the problem of the energy consumed by the transport of data. Today, the biggest bottleneck in terms of design is no longer the compute capacity that can be reached, but the consumption of bandwidth power to perform those computations.
This has led different manufacturers to create new interconnection technologies to reduce the power consumption of data transmission. The problem? Each builder has created their own interconnection technology, which leads them to create a proprietary ecosystem in each case. How does this translate? Well, in the fact that we are heading towards a market in which Intel, AMD or NVIDIA will be able to create an ecosystem where apart from certain components that do not depend on these brands, we will end up with solutions from a single Mark.
So we may find ourselves in a future situation where when we buy an Intel processor, we may find that it can only be paired with an Intel GPU and vice versa. A phenomenon that can occur with other manufacturers, and is now starting to occur, an example being the leak of future PCB gaming laptops with Intel Xe-HPG from Intel which pairs its new gaming GPUs with its proprietary processors. .
