According to The Information, Nvidia may be considering using memory with reduced bandwidth compared with the top-tier products the market has to offer. Has to offer… so to speak, because it would in fact be the shortage of these products on the market that is pushing Nvidia in this direction. This would be bad news for leaders in the memory sector, which can obviously sell top-of-the-line products with higher margins and virtually no competition.
That’s the market, one might say. If the goods are not available, and become too expensive because supplies are limited, some will look for alternatives. According to The Information, those alternatives would involve using lower-end memory.
What consequences for the markets?
It should be stressed that, at least for now, this is only a possible option under consideration. Such a decision would clearly also have significant repercussions for technology, performance and efficiency. The signal, however, is clear: the current shortage—or perhaps it would be more accurate to say demand that is too high relative to supply—will lead many companies to make different choices whenever possible.
Returning to the consumer segment, Apple has already imposed significant price increases across its entire product line. Other PC and notebook assemblers are slowing down upgrades, while still others are struggling to offer the market devices in the lower price range, particularly those costing a few hundred euros.
The most significant consequences should be felt by Samsung and SK Hynix, whose valuations also take into account their enormous backlog—that is, currently unfilled orders—which could partly disappear if Nvidia were to make such a radical decision.
How much of this is true?
For now, we are dealing with considerations. The fact that Nvidia is thinking about it is an important signal, but before making such a decision, we imagine that significant assessments of performance and efficiency will have to be carried out.
The systems that could potentially be affected are the Rubin Ultra machines, the next iteration of Nvidia’s powerful artificial intelligence systems.
This brings us back to common sense: no matter how much money is poured into a sector, supply cannot immediately adjust.
