Could the problem be that Micron, at least, already stopped manufacturing 2GB GDDR7 chips, and I wouldn’t be surprised if other suppliers also directed their limited capacity toward 3GB chip production.
I tried looking up the current price level, but information is scarce; an AI bot guessed 2–3x prices per chip (2GB vs 3GB), but the sources are pretty low-quality. We also don’t know what NVIDIA actually pays for them—presumably, as a major customer of memory manufacturers, they can squeeze out a better deal.
What’s clear is that if we get an RTX 5080 24GB, the price tag won’t be anywhere near the RTX 5080 16GB MSRP… My guess is that the starting price would be at least $1,500, and in the current market environment, it could be even higher. Meaning back home in Finland, at least around a couple of grand. Which is an utterly ridiculous price, but what can you do. A market economy works how it works. This is what we have to roll with until the absurd supply-demand imbalance corrects itself, at the latest when supply increases.
At least so far, they haven’t managed to make any progress on ramping up production for that new card to replace the discontinued one. Granted, the priorities are naturally on the AI side, and my guess is that we’re indeed heading toward some sort of paper launch with no real intention of actually delivering cards to retail.
NVIDIA has reportedly missed its own schedule for providing GeForce RTX 5080 24GB design kits to board partners. According to BenchLife, the company was expected to distribute the necessary documentation in October, but manufacturers had still not received it by the end of the month.