At the beginning of the year, ASUS announced its intention to increase the prices of some PC components – especially for motherboards and graphics cards. At that time, ASUS officially responded to the increased import duties and logistics costs in the USA. There should be no changes in this country.
According to a DigiTimes report, MSI’s CEO, Joseph Hsu, announced similar steps when the last quarterly figures were announced, or blamed increased costs in logistics for the increased costs of PC components – at least in certain respects. For its own mainboards, graphics cards and gaming notebooks, MSI expects continued high demand over the full year 2021. In 2020, sales increased by 30 to 50% depending on the product category. For 2021, MSI also expects growth rates in the double-digit percentage range.
But there is a ray of hope: Hsu, or MSI, expects the costs for logistics to start falling again after March, as the “global logistics system has gradually returned to normal”. So nothing changes in the high demand, but the transport costs should have reached a normal level again, which should affect the supply and thus the price. As long as demand exceeds supply, as is currently the case, lower logistics costs are unlikely to have any impact on the price. Only a few days ago we reported on the completely out of control graphics card market.
ASUS also sees reasons in NVIDIA’s chip yield
ASUS has also published its latest quarterly figures and there was also an interesting comment with regard to the current tense market situation – especially with graphics cards.
“Our guess is that the gap might have been caused by lower yields upstream. As for when NVIDIA can increase that yield is something hard for us to predict.”
The higher demand on the overall market due to the pandemic and also the extreme demand for PC components from autumn onwards were initially difficult to predict. It is almost impossible to react to this at short notice. What the ASUS statement is aimed at, however, is that NVIDIA apparently does not achieve the planned yield at Samsung and therefore there are additional bottlenecks in GPU production.
With the use of Samsung’s manufacturing in 8 nm (and with a switch away from TSMC), NVIDIA apparently wanted to avoid bottlenecks at TSMC (and possibly higher costs). Only NVIDIA can answer whether this ultimately succeeded.
In the manufacture of GPUs, errors occur again and again, so that not all chips on the wafer meet the criteria for use on GeForce and Quadro cards. In some cases, NVIDIA wants to use these on the special CMP HX crypto mining cards. The GPUs used here do not meet the said criteria and are therefore scrap that could not previously be used otherwise. How high this proportion is in the manufactured GA102, GA104 and GA106 GPUs from NVIDIA will probably remain a secret.