Hardware/ nvidia · gpu · rtx-5080-super · memory-prices

Nvidia's RTX 5080 Super Is Reportedly Dead Over Memory Costs

A leak says Nvidia told Chinese partners its 24GB RTX 5080 Super is off, with pricier memory chips as the likely cause and no new launch date.

Nvidia has reportedly cancelled the RTX 5080 Super, a 24GB refresh of the RTX 5080 that was rumored for January 2027.

The claim comes from a post on Board Channels, a Chinese forum that is a common source of leaks. It says Nvidia told graphics card makers in China the launch is off and gave no "clear subsequent launch schedule" (a translation from Chinese). That is stronger than a rumor from a week earlier, which only said the card was delayed. The same post says the 16GB RTX 5080 stays the top card in China, where the RTX 5090 is not sold. Nothing here is confirmed by Nvidia, and it rests on a single translated post.

The likely culprit is memory. To reach 24GB of VRAM (the video memory on a graphics card), the Super would need 3GB modules of GDDR7, versus the 2GB modules on the current RTX 5080. Those larger chips have become much more expensive.

Mid-generation "Super" refreshes are a normal part of a graphics cycle, so skipping one says more about the memory market than about the card itself. The broader RAM shortage is predicted to worsen toward the end of this year and get bleaker in 2027. The RTX 5090 shows the squeeze: stock is thin, prices run three to four times the list price in many regions, and the card has approached $10,000 in the US.

Gamers are pointing at Nvidia. One Redditor argued that its AI demand pulled memory production from GDDR toward HBM, the high-bandwidth memory used in data center AI cards. That is a commenter's claim, not an established fact, but it fits Nvidia's obvious focus on data centers.

The open question is what happens to the plain RTX 5080. One earlier rumor said it would be dropped in favor of the Super. If the Super is gone, that rumor looks shakier, and buyers may be left with one fewer high-end option and no sign of cheaper memory.

TR

The Revision

Written by an AI system from the public sources credited above. How we write →